June 3, 2025 • 35 mins
I am always looking for ways to improve. Send me a text and let me know your thoughts! - Kevin
Chris O'Neill, CEO of GrowthLoop and former leader at Google and Evernote, reveals how AI is creating an entirely new marketing paradigm through what he calls "compound marketing."
This fascinating conversation explores how marketing is shifting from slow, manual processes to a world where AI agents function as team members, handling the routine while humans focus on creativity. O'Neill explains the concept of compounding in marketing – where small, consistent improvements create exponential growth. Just as compound interest generates wealth through small percentages over time, compound marketing leverages AI to increase the velocity of marketing experiments and improvements.
The most revolutionary aspect of this approach is how it removes the pain points of cross-channel marketing. GrowthLoop's technology abstracts away the tedious work of activating campaigns across multiple platforms and analyzing siloed data. The result? Marketers can move from idea to impact dramatically faster, with AI agents working as "glue" teammates that stitch everything together behind the scenes.
O'Neill offers a compelling vision for how marketing roles will evolve – not by being replaced, but by being elevated. He suggests individual contributors will need managerial skills to effectively guide these AI teammates: setting clear goals, providing feedback, and knowing when to delegate versus remain involved. The transformation extends to business models as well, with O'Neill predicting a shift from software-as-a-service to results-as-a-service, where companies pay for outcomes rather than seats.
Interested in learning how AI agents could transform your marketing team's capabilities and results?
Listen now to understand the fundamental shift happening in marketing operations and how to position yourself at the forefront of this revolution.
🎧 Tech Marketing Rewired is hosted by Kevin Kerner, founder of Mighty & True.
New episodes drop regularly with unfiltered conversations from the frontlines of B2B and tech marketing.
Catch the video version of the podcast on our YouTube channel.
👉 Subscribe, share, and connect with us at www.mightyandtrue.com
📩 Got a guest idea or question? Email us at kevin@mightyandtrue.com
🔗 Follow Kevin on LinkedIn for more insights and behind-the-scenes takes.
📰 Subscribe to Kevin's Substack at: Subscribe to Kevin's Substack at: https://kevinkerner.substack.com/
📰 Want more tech marketing insights? Subscribe to the Drag & Drop newsletter for fresh thinking, real talk, and tools that make your job easier.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Kevin Kerner (00:00):
Hey everyone, this is Kevin Kerner with Tech
Marketing Rewired.
Today's episode is a great one.
I sat down with Chris O'Neill,ceo of GrowthLoop, someone who's
led at places like Google,evernote and is now helping
reshape how marketers actuallyget work done.
We got into Chris's idea ofcompound marketing and what it
really means to have AI agentsas teammates, and why Chris
(00:20):
thinks we're headed into aresults-as-a-service world, not
just software.
We, chris, thinks we're headedinto a results as a service
world, not just software Talkedabout speed, smarter systems and
, honestly, how to take most ofthe pain out of modern marketing
while leaving time for all thecreativity.
And if you've ever been stuckin cross-channel chaos, you'll
definitely want to hear howGrowth Loop is solving for that.
Let's get into it.
(00:42):
This is Tech Marketing Rewired.
All right, hello everyone, andwelcome back to Tech Marketing
Rewired.
I'm your host, kevin Kerner, andtoday I'm very excited to have
Chris O'Neill, who's the CEO ofGrowth Loop.
For those of you that don'tknow Chris, he has a really
incredible career.
Up to this point, he's beenmanaging director at Google
Canada and Google X.
(01:03):
He served as the CEO andchairman of Evernote and Glean.
I loved Evernote, by the way.
I love it.
Thank you.
It was one of my firstnote-taking tools and is
currently a board member at Gap.
So, chris, I am super excitedto have you here on the podcast.
Yeah, it's great to be with you.
Yeah, yeah, it's going toreally be amazing to talk about
(01:23):
this new category of eugenicmarketing and what you're
building at Growth Loop.
One fun fact that I'm not sureeveryone knows is that Mookie
Betts is on your board, which isamazing to me.
Chris O'Neill (01:33):
He is he is?
He's an amazing, obviouslyathlete and human, very curious
and really wants to learn moreabout business, and we're
fortunate to have.
Kevin Kerner (01:48):
Yeah, that's great , I think, having that different
perspective.
Now I'm a Guardians fan and theDodgers just killed the
Guardians in this life series,so tell him to take it easy on
us this year, which I'm sure hewon't.
Chris O'Neill (01:57):
He will not.
Kevin Kerner (01:58):
They are very competitive.
Okay, so there are a lot ofthings I want to dig into, but
let's start with just thisamazing transformation that's
going on right now and theimpact for leaders.
You've seen all this waves ofinnovation across some really
impressive companies, and youhave an amazing co-founder in
David, and then Tamim, the CTO,who's doing some incredible work
.
I'm curious, from yourperspective, what feels
(02:19):
fundamentally different aboutbuilding a company right now,
from a CEO perspective, aboutbuilding a company right now
from a CEO perspective.
Chris O'Neill (02:24):
Yeah well, I've been around long enough to have
been through lots of differentpretty big disruptions or
changes from the shift to mobile, obviously, the advent of the
cloud and this one certainly haselements of disruption like
that in terms of how it meansthe interaction between
consumers or customers and howthey consume information.
That's all true.
(02:45):
The big shift here is really onthe supply side, in other words
, it's how we create, in thiscase, marketing, but really how
we go about the work of work.
Ai is fundamentally changingthat.
So it's really really, I think,the biggest thing that's
happened in my lifetime.
And then the pace with whichit's happening is unprecedented.
(03:07):
I mean, it literally feels likeevery week there's a new model,
a new protocol and it'sunlocking a whole host of use
cases we could never evenimagine.
So it's just so incredible andI know there's a lot of hype and
I do think that it iswell-deserved and in many ways
we'll see, the impact isprobably greater than we can
(03:28):
imagine right now.
Kevin Kerner (03:29):
Yeah, it really is incredible.
Yeah, mcp, microsoft's new NLweb announcement.
It's just, it's just reallyhappening so fast.
The models are getting so muchfaster, quicker.
Well, now clicking into thegrowth loop thing what was the
catalyst that that led you tobecome interested in this more
operational role inside ofgrowth loop?
Chris O'Neill (03:50):
Yeah, I've been inside companies, I've been an
investor, I've dealt with everytool over the years, and the sad
reality is marketing is justtoo slow, too manual.
It doesn't really allow toscale up to the opportunity
potential to deliver delightfulexperiences consistently to
consumers.
That's always bothered me.
(04:10):
It's bothered me for over twodecades I mean going back to my
time at Google.
We were obviously solving apart of that, but the large and
sad reality is most experiencesthat customers have are very
impersonal, they're verysuboptimal, and that doesn't
need to be that way.
So this is about, really, howcan you use technology to serve
(04:30):
up fundamentally differentloyalty experiences, lifecycle
interactions with customers?
And that's really what it'sabout is how do you apply
agentic AI in this case, on topof what's called first party
data, right To basically drivegrowth faster and more relevant
for consumers?
And we'll talk about thisconcept of compound marketing
(04:51):
because, really, that themissing gap is not only that
you're driving gains, but you'redoing so faster, right?
So this concept of compoundingis one I'm personally obsessed
with and I'm happy to dive intoit a little bit more with you.
Kevin Kerner (05:03):
Yeah, I would love to.
Yeah, it's one of the reasons Ihave not as much hair anymore
is because how hard marketingand marketing has been for the
last 30 years yeah.
So I'm, I'm, I'm very excitedabout where you're headed,
because it won't grow back, butmaybe it'll make things easier.
Um, let's hope, yeah, yeah.
So, um, you've been using youmentioned this term, compound
(05:23):
marketing, and I've seen thedemo and I can't seem to sort of
shake how impactful this couldbe.
Let's click into that a bit andexplain from your perspective
what this new category ofcompound marketing is and how
it's different than today'smarketing.
Chris O'Neill (05:39):
Let me zoom out for just a second.
I just have a fundamentalbelief that all great companies
create categories.
Right.
It's not just about saying, hey, we're like this person or this
company, but we're better.
It has to be about difference.
It's to say like we'refundamentally approaching the
world in a different way.
Think of all great companies,companies like Qualtrics or
Google.
Evernote, in its day right, setout to create a new category.
(06:02):
It's just fundamentallydifferent.
We're solving a differentproblem in a different way.
So that's the framing abroadand that really has to do with
how you drive growth and how youtalk about benefits.
When we looked across the marketand it's super crowded, as we
know tens of thousands ofdifferent tools and companies
chasing this in this broader,you know, marketing technology
(06:22):
landscape.
So you have to cut, cut throughthe noise.
So it starts with that.
The other thing I'll add isI've been obsessed with compound
interest in investing since avery young age.
I remember parents I told thisstory recently and I grew up in
a very small town in the snowbelt of Canada and I was given
this book called the WealthyBarber and it talks about this
(06:43):
metaphorical wealthy barbernamed Sam who basically lives a
very simple life, but followed aformula that really involved
compound interest and doingsmall things over time to build
some substantial wealth, notjust in a quick way, but
overnight, slowly and surely.
And that's the power ofcompounding the eighth wonder of
the world overnight, slowly andsurely, and that's the power of
(07:05):
compounding, you know, theeighth wonder of the world.
So it got us to thinking how doyou apply that concept that's
very familiar in investing tomarketing?
So the difference between, youknow, a 1% gain every month
versus a 2% gain every weekisn't just a little bit, it's
actually 13, 14, 15x, right.
So we're here to basically helpmarketers apply AI to their
(07:26):
data to move quickly, morequickly, through the entire
cycle.
Right From figuring out, I havean idea that's going to move
churn or reduce churn or perhapsboost lifetime value.
How do you take it from thatidea to actual impact?
We're about compressing thatimpact.
So, idea to impact cyclethrough marketing who do we talk
(07:48):
to?
What do we talk to them about?
When do we talk to them?
And then how do we measure itall?
And then lather, rinse andrepeat by basically figuring out
what worked against ourobjective function of lifetime
value or reducing churn orwhatever it happens to be and
just continuously doing that,and that's the name of our
company, growth Loop, but itreally is leveraging the power
of compounding which, again Imentioned, is considered the
(08:11):
eighth wonder of the world.
So we're thinking and borrowingphrases and concepts from other
parts of the world and applyingthem to marketing.
Kevin Kerner (08:20):
Yeah, fascinating.
Yeah, we talk a lot about speedto outcome and so when I heard
the compound term, I was like Ilinked directly to that.
It's like how do you get tothat outcome faster?
It's like moving that outcomefrom here down to here and
there's a bunch of little stepsto get to that outcome and so if
you can compress that, you justget there faster.
What were the act like?
(08:49):
Like seems like solving this,solving this compound marketing
thing must have had severalactivators.
Like what did you work on tomake sure that that you could
begin to realize this vision ofcompound marketing?
There's a bunch of differentparts of it.
Like what are they?
Chris O'Neill (08:58):
Yeah, yeah, it really all starts with data.
It really all starts with dataand one of the things that is
also happening in the backgroundis not by accident or
coincidence is the improvementof the quality of data and the
semantic layer of data.
So the rise of data clouds isreally a big part of what's
going on.
So for too long data has beensiloed off so it's very
(09:19):
difficult to kind of stitch ittogether and then access it.
So the democratization of datathrough the rise of things like
Snowflake and Databricks andGoogle BigQuery and AWS, et
cetera, that's a really big partof it.
So it really starts there.
It has to be the quality ofyour data and your data strategy
and the guardrails you putaround.
It is where it has to start.
(09:40):
Ai is not very useful, machinelearning not very useful if
you're doing so with poor data.
So it starts there and then itreally we started to experiment
with not just removing frictionfrom very manual processes,
we're sort of automating theend-to-end marketing cycle.
So that has to do withself-serving for marketers,
right?
Historically it means I have anidea I got to go talk to a data
(10:02):
team to write some SQL and pulla query and then you bounce
back and forth to see is thistoo big, too small?
Do we have the right audienceas a starting point to at least
have the conversation, make anoffer that would help boost
loyalty, or whatever it happensto be?
Now, first step was to removeand automate the entire
end-to-end process, so thatwe've done that and that's
(10:22):
really well-received, and that'swhere we got our start.
Now it's about infusing AI atevery one of those steps and
having a supervising agent tobasically make sure that these
swarms of agents are actuallyworking together in unison,
consistent with the guardrails,consistent with governance
practices, so it doesn't returngarbage.
So, essentially, we have thismindset of an end to end process
(10:45):
.
We're trying to automate.
Now AI allows us to do eachstep of the way better in some
meaningful way, and then, in thefullness of time, we see, with
humans in the loop, an end toend process.
So that's really what this isabout, and we're really
delighted with the early results.
I want to be equally clear,though like we have a long way
(11:05):
to go.
These models are incredible andthey're advancing at an
unprecedented rate.
The slope of the improvementsis astoundingly great and
there's way more work to do tobasically boost the quality of
each one of the steps, whetherit's the audience agent we have,
or our journeys agent, or, uhwhat.
There's literally a dozen plusof them that we're experimenting
(11:27):
with right now.
Uh, and we're, we're nevergonna, we're not gonna rest
until the quality continues toto exceed anything even
imaginable.
Um, with humans doing the sametask.
Kevin Kerner (11:37):
Yeah it's the real unlock for me is the fact that
you know the agents talking toworking together and not being
singular agents that are working.
They are working on their pointfunctions, but the fact that
they're able to talk together,work together, is really a lot
for me.
Getting siloed marketing databetween channels and analyzing
(11:58):
it is one of the biggest pains.
It's the biggest pain in thebutt ever.
You're in Google, you're insocial, you're in some other
channels.
You get those, those data,typically separately and then
you have to analyze it yourself.
Are you?
If you're, you know, maybeyou're smart enough to put it
into some sort of llm and haveit combine it, but you're doing
that pretty much manually andyou're saying what you figured
(12:18):
out as a way to, um, have thoseuh channels analyzed in a
consolidated fashion using theseagents.
Chris O'Neill (12:27):
Yeah, that's right, and it's actually
upstream of that.
It's actually activating acrossall the channels, so once you
have ideas and you have specificoffers and specific things you
want to do across your channelsright now.
Not only do you have to stitchtogether all that stuff upstream
, to go and activate it requiresseparate discrete efforts and
(12:48):
manual efforts, separatediscrete efforts and manual
efforts.
We have abstracted that awaywhere you have a single place to
activate and then a singleplace to pull back and read back
into the data cloud the results.
These are the little thingsthat get in the way, right.
So it's about automating theworkflow and this is the work of
the work and why I think thisis so powerful.
(13:08):
So this application layer, ai,application layer things are so
powerful because the real worldis messy, right, and the insight
and the task-specificalgorithms you know, the
step-by-step algorithms thatstitch together, are going to
create and are creating realvalue right now.
So it's not just like any oneof the steps require manual
(13:30):
stitching together, it's kind ofall of them.
So that's where these agentscome into play and really I
think of them as like the glue,the glue guy or gal on a team.
They're really, they're reallystitching, doing the work to
stitch and pull things together.
So it's like having a thousanddifferent interns going off and
doing things you know again with, with supervision and bringing
it back in the loop at the righttime.
(13:51):
But really, that's what it'sabout is abstracting away modern
day TPS reports.
For those office space fans,right, it's like there's so much
mundane, soul sucking work thatreally can go away.
That's where agents will shineand that will allow humans to
(14:12):
unleash their creativity, theirpotential, by basically focusing
on the things that that humanscan uniquely do.
So, yes, jobs will befundamentally transformed.
Some jobs will go away, butmany others will be recreated in
a much more hopefullysatisfying and way that's
simpatico with the humancondition and what humans are
great at.
So that's how I see thisunfolding and, of course, no one
fully knows, but I do see theopportunity for humans to
(14:35):
interact with literallythousands of these agents in a
very simpatico way, uh, in asymbiotic way, um, to really
deliver real results that youcan quantify yeah, I totally
agree.
Kevin Kerner (14:46):
It's the 80 is 80, 20 rule.
Like% of the work is justmundane work and, as someone
that's been doing this for, youknow, 30 years, there's just a
lot of work that seems like itshould be able to be done by the
agents.
It'd be interesting to see ifthere's a new version of office
space out there that will be forthe AI.
Chris O'Neill (15:05):
Some other DPS report.
Well the agents are hard atwork at it already, kevin,
that's right.
Kevin Kerner (15:11):
Did you see VO3 last week?
Yes, I did.
Chris O'Neill (15:14):
How mind-blowingly cool is that.
I mean holy cow 8Z.
Kevin Kerner (15:22):
Yeah, I was playing it with my kids and they
were going nuts and I have onekid that's in more art school
and she did not like it at all.
She was like no, this is notgood.
Although I agree with you, Ithink I do think that the like I
do think that I mean I've beenthrough these technology
transformations before andthere's more work to be done.
As a marketer, there's it, andthe work that will you'll be
able to do, that you'll need todo, will be a different category
(15:44):
of work.
It'll give you more time to bethinking about the things that
need to be done and sort ofrunning at a whole different
level with these assistants.
Chris O'Neill (15:52):
Yeah, I think it's going to unlock more
creativity.
I might be in the minority here, but I actually think if you
have more tools in the hands ofcreators, designers, I think
it's going to unlock creativity.
So the mundane goes away andthen it unlocks more time and
energy.
It's going to be massive,massive change around the change
for people to absorb, let's beclear.
(16:13):
But I'm of the conviction andthe belief that it's going to
unlock creativity and we're just, you know, on the verge of
seeing that.
But it will be bumpy for sure.
Kevin Kerner (16:23):
Yes, I agree.
I have one quick question I'mjust curious about from the CDP
side of things Is your, do yousee an opportunity for the
agents to create data new fieldsinside of the CDP or begin
their own?
Are you doing that now?
Chris O'Neill (16:40):
We are.
We're experimenting.
The way we're experimentingwith that is like synthetic data
, right?
So what we'll do is we'llcreate a synthetic like a
synthetic profile human and it'sremarkable how well and robust
you can make a synthetic profileand then you simulate the
interactions with that.
So why does that matter to amarketing Like?
(17:02):
Is it the same as a human?
No, it's not, but it's really.
It actually allows you to drawconclusions with a simulated
behavior, which is generallywe're finding remarkably
powerful and accurate.
You don't have to go spend abunch of money on marketing to
test a hypothesis using A-Btesting in this case, right, you
can do a lot of that in alaboratory.
(17:23):
We're leaning into doing that.
It's one of the most excitingthings we're doing, frankly, is
just like this simulation ofhuman behavior.
Uh, without having to gothrough all the steps and spend
lots of time, energy and moneyto figure out what the response
was, you can simulate itbeforehand, which is super
exciting that is amazing,absolutely amazing.
Kevin Kerner (17:42):
I talked to a company yesterday that was a
research.
They had a long history inconsumer research and they were
doing they were talking aboutdoing that same thing, like they
were using their research datahistory you know, decades of
research data and putting itinto an LLM and then creating
these personas.
It was data that they had.
I thought it was superinteresting.
So the fact that you could talkto your persona or at least
(18:04):
learn from your persona usingsynthetic data is pretty wild.
Chris O'Neill (18:09):
Well, let me tell you a fun story about how we
created our category.
Kevin, we did a version of this.
It wasn't quite as extensive ascreating thousands and
thousands and thousands ofdifferent consumer personas, but
what we did, we said who areour target personas?
We're talking to CMOs, we'retalking to CTOs or CTDOs,
depending on the thing and webasically said, ok, great, we
(18:29):
fed it a bunch of informationusing the same technology
underlying Google's notebook LLMand we basically simulated
conversations back and forth tosay, hey, this is our
positioning, this is what we doand does this resonate with?
Like?
We simulated a podcast, so whatwe're doing and it helped us
refine it.
I'll tell you.
(18:49):
We actually narrowed ourcategory names down to two
things.
We said it was agenticmarketing or compound marketing.
We made the decision to docompound marketing for many of
the reasons I've alreadydiscussed, so I won't repeat
them, and also because itresonated with people, these
simulated personas that we wentinto LinkedIn Navigator and said
here are some actual personas,here is an actual content that
(19:11):
we're thinking of talking, andwe simulated these conversations
.
So it's not just a hand wavything.
We're using it to help informhow we position our company by
using these powerful tools and,we think, pretty innovative ways
.
So this is stuff we're actuallydoing today.
Kevin Kerner (19:27):
Amazing.
That's a great use case.
So I want to dig into a bit ofthe skills, the new stack and
the new skill sets that areneeded now.
I love the idea.
I hadn't thought of it beforeuntil I watched one of your
keynotes about agents as teammembers.
Like actually seeing the agentsas a worker on behalf of you
doing things.
It's kind of a really cool wayto sort of get your mindset
around it.
What kind of skillset shift doyou think the next generation of
(19:51):
marketers will need to adopt toharness these news agent call
them agentic teammates.
Chris O'Neill (19:58):
Yeah, many, many of the skills are actually the
same.
To be clear, I'll start thereand then I'll talk about what's
different.
The same is like look, I see aworld where individual
contributors are are going toneed the managerial skills that
great managers have today.
What does that look like?
Well, you need to be able toset really good goals, right.
You need to give reallyeffective feedback and you need
to figure out how and when todelegate.
(20:19):
Those are classic managerialskills.
In this case, it's just going tobe using an agent, right?
So agents are remarkably goodat following instructions, so
you've got to set a really cleargoal.
Are remarkably good atfollowing instructions, so you
got to set a really clear goal.
They're really good atresponding to sub prompts or
(20:39):
subsequent feedback.
And then the last part is likefiguring out when the human
should uniquely be in the loopor or when the agent can be set
free.
Like that.
That is a classic managementthat even individual
contributors are going to need,because I believe everyone in
the company is going to interactwith agents.
So that's where I'd say there'sa lot more in common than not
and this concept of as a teammember.
I'm a big believer.
I love this term glue guy orgal right.
(21:01):
These are the people that justelevate everyone's performance,
and that's what I believe agentsare right.
They're basically going toallow you to focus on the things
where you uniquely add value,and it's going to rise the
overall performance and elevatethe performance of everyone.
What does that mean?
Well, it means a differentmindset.
These are not here to take yourjob.
It's here to elevate andaugment.
So I'd say that the other thingI would say is you've got to
(21:23):
think about velocity right.
The real advantage here is theability to compound and go
faster, as we described.
So really not trying to findperfection off the bat.
It's about gosh.
You can launch so many moreexperiments, so it doesn't mean
you just do it for its own sake,but you basically have the
ability to not let perfection bethe enemy of done or good
enough, because with theseiterative loops, with agents
(21:47):
doing most of the work for you,you can get through it quicker.
So you have to have thismindset that velocity matters
more, and velocity is a functionof both speed and direction.
It's speed times direction.
If you think about physics,it's not just doing stuff for
its own sake.
It's about being intentionalabout the direction you're
taking and then really usingthese agents to think more
faster.
And then I guess, moregenerally, I'm a big fan of
(22:09):
slope over intercept in terms ofpeople's capabilities.
What I mean by that is theability to learn and understand
and apply insights to the nexttask you're going to go after or
the next set of challenges youtake on.
And I have a high school-ageddaughter and I just just started
(22:31):
college age son and what I'mreally encouraging them to do
and it's remarkable to see howthey use these tools.
It's amazing to me.
Yes, he's really become.
Become a ninja at using thesetools.
Be so good that people can'tignore you like this is like a
superpower.
The people, the youngergenerations, that are using this
stuff natively.
I mean, I watch how he studies,I watch how he basically
(22:58):
creates these flashcardsinstantly, compresses hours of
lectures into minutes of supersummary notes.
It's amazing and we're justscratching the surface.
So I would say lean into thesetools, become a master at using
them, because we have only juststarted to see what an operating
system that's rooted in thesellm models uh, with deep
personalization and deep context, will have.
It's just, it's mind-blowing tome at this stage and we're just
(23:19):
early in it.
Kevin Kerner (23:20):
Yeah, I have four college-aged kids right now and
um it's, I think they're movingfaster into this experimentation
, to this stuff, than thecolleges are certainly.
I mean, they are getting a lotof pushback from the colleges
and I am, of course, I'm on the.
I'm on the outer and bleedingedge of this stuff and I'm
trying to talk to the professorsand deans to say, look, you got
to get on board here this stuff.
(23:40):
You should be, you should betaking growth loop into
universities and showing thesome of the professors some of
the stuff that's going on here,because you would.
There's a uh, there can be aproblem with people not being
told not to use it at this point, which is really important.
Chris O'Neill (23:56):
We are doing a version of that.
Actually, a former colleague ofmine, Jim Lissinsky, is a, is
an amazing professor at Kelloggand uh, he, he and I, uh, he's,
he's, he's fantastic.
He is fantastic, yeah, so sowe're we're working with Jim on
a couple of different fronts.
Kevin Kerner (24:11):
That's great.
Chris O'Neill (24:12):
Yeah, like, like students need to to, to feel
comfortable with these thingsLike it really is.
I think it's going to be, youknow, again, as I said, it's
going to be like a superpowerfor them.
Kevin Kerner (24:21):
So it just their world, is going to be so much
different than I think what we,what we can even imagine right
now and I, I'm a very optimisticperson on all this yeah, I
think though, that there'll beone of the things you, one of
the things you've I'm sorry,chris, um, one of the things
that, um, I think you guys havesolved, for that is really
interesting from a skill setperspective is you've tried to
(24:41):
make the ui very easy andintuitive, like I like.
When I saw the demo, I was like, okay, the problem with getting
and I have a team that isreally good at using these tools
, but they're, if they use atool and they can't, it doesn't
work.
When they first use it, they'lljust drop it.
I'm curious, just from a userexperience perspective and
(25:03):
intentionality behind the toolbuilding the tool.
Did you take that into accountas you were, as you were
building the thing, trying tomake it as easy as possible,
take the technical part awayfrom the marketers so they could
focus on the goal?
Chris O'Neill (25:15):
Yeah, it's very intentional, to be clear, and
I'll comment more generally interms of the importance of
design, because I do think itbecomes more important.
But, yeah, absolutely, this wasdesigned for marketers.
We say it's designed formarketers, loved by data teams.
The data teams love it becauseagain, it removes a lot of the
low value added work of goingback and forth with SQL.
(25:36):
So the first iteration of thiswas very intentional to design a
beautiful UI that marketerscould just intuitively
understand from the beginning,and that is how actually we've
won businesses.
Like literally, when we'redoing demos and proof of concept
with people, you can almost seethe marketers.
It's so fun.
They're like wait, is that?
Like what?
Where's am I being punked right?
(25:57):
now, I just do that, did I, did,I just did I just literally
create an audience in like threeseconds with ai?
Like they're like there's thisbig smile comes across their
face so that's awesome so,absolutely it is.
And and I just on the on thecomment of design, I actually
think chats of not a great.
It's a suboptimal interface.
It's the best we've got, sowe're using it, but it's it
(26:18):
requires too much cognitive loadon the user itself.
You have to prompt and it'slike you know.
So I see a world with theseagents sort of these thousands
of agents, like these thousandsof interns you have will fade
into the background and you know, I think agentic AI is the
spark.
But it will be design thatreally brings the true power in
the form that we haven't evenimagined yet.
So I think that that is anelevating again back to the
(26:42):
point.
Like design becomes moreimportant, creativity becomes
more important to really figureout how we harness the
technology and allow it to fadeinto the background so we can
really focus on the businessoutcomes and delivering
creativity in service ofcustomers.
That's really how I see ithappening.
But, yes, that is adifferentiator for us that
(27:03):
people really do enjoy using ourplatform.
Kevin Kerner (27:06):
Yeah, that's an incredible vision.
I mean, if you can pull thatoff, it's pretty amazing.
A couple other thoughts here aswe're wrapping things up.
I've heard you mention the termresults as a service, and I
love the idea that there's achain.
There's not only you'rebuilding a category here, which
is hard to do, but you're kindof building a different go to
(27:26):
market.
Talk to me a little about youridea of results as a service
versus software as a service andhow that plays out for you guys
.
Chris O'Neill (27:34):
Yeah, a couple of things.
So software as a service isreally rooted in the concept of
a seat.
Right, it's linked to a human.
There's this X, many humansusing the software product,
right, and so what agents allowyou to do is to couple that
right, agents are workingcontinuously, they work
autonomously with supervision,so it breaks down that
connectivity, okay, well, whydoes that matter?
(27:56):
Well, it allows amazing thingsto be possible.
It allows for almost unlimitedscalability, right, there's like
all these different thingswhich humans just can't get to.
That you know, working 24,seven and having like the
ability to span, like thecapabilities, is really limited
by the compute, which is nothuman limited.
(28:18):
Okay, and then that also allowsyou to think differently about
the pricing model, the businessmodel more generally.
It's not we're charging you perseat.
You can start to do things liketasks or you can, even better,
do the actual outcomes.
You're selling work.
You can basically say look, wecan prove, quantifiably, prove
that there's a different outcomehere.
(28:38):
Right, maybe we moved churn ormaybe we improved lifetime value
.
So it's going to create newways, fundamental new ways for
business value to be exchangedand charged for.
You're seeing it with SierraBrett Taylor's company, which is
doing customer serviceinteractions right, they're not
charging per seat.
They're basically investingheavily upfront and they're
(28:59):
saying look, we'll charge for aresolved customer service case.
Right, that's their pricingmodel.
So I think that that's one ofthe more exciting things.
With agents, right, we havethese agent swarms and the
protocols that allow them.
We have the 80-20 you'retalking about before we're going
to, basically, humans are goingto work on the 20% that
actually is unique to them and80% of the other work will go to
(29:19):
agents.
And then, above it all is,basically we're going to start
to see oh my gosh, wouldn't itbe amazing if we could actually
align value here, where there'sactually a business outcome
based pricing, as opposed tojust artificial proxy called
seats, which has no bearinggoing forward?
And, if I may.
I think this is what's going on.
This is why Salesforce is goingout having to buy growth.
Like, if I may, I think this iswhat's going on.
(29:40):
This is why Salesforce is goingon having to buy growth.
The systems of record arereally in trouble.
You have the rise of these dataclouds.
You have agentic AI on topSystems of record have seen
their high watermark.
I really believe that.
So it's really going to be afascinating time in business to
see how this all unfolds.
Kevin Kerner (29:55):
As a service provider, an agency.
We're seeing the same thing.
It's like there should be somepayoff for can't be the old
(30:19):
model and I think the a lot ofthe unfortunately a lot of the
older, um larger holdingcompanies are going to have a
really hard time making thatshift, that cost model shift to
something that's more resultsoriented.
It's really interesting timesin this category I do think
you're right.
Chris O'Neill (30:37):
You saw it with programmatic, right,
programmatic was a pretty,pretty tough change.
This is even bigger than that,of course.
It really is.
I have lots of stories beyondthe scope of today's
conversation just about how youthink about all the different
agency jobs that they do and howthey're being changed and how
that value gets exchanged withtheir actual clients.
(30:58):
But this is not justadvertising.
This is an audit, this is anaccounting.
It's again back to our earlierconversation.
It's every corner of companies,you know, from the supply chain
all the way through to customerfacing, pre and post sales.
So it's really a big deal.
And now that's going to be thebiggest and longest poll, I
(31:19):
believe, is change management.
How do you get people toembrace these things and change
their behavior?
As always, right it becomes.
I had an operations professorin business school who used to
end every class we talk aboutgnarly supply chain problems and
like really the operations of acompany, and he'd always I'll
never forget this he said youknow, in the end it's an OB
problem, right, Organizationalbehavior or a human problem,
(31:41):
right, Like it's an opportunityor a problem that really traces
back to how do you get humans tochange their mindset and their
behaviors, and that's reallybeing laid bare.
Kevin Kerner (31:51):
So true, it's a revolution.
Chris O'Neill (31:54):
I'm excited about it, but it's a real human
problem.
Kevin Kerner (31:59):
Well, chris, I have one thing I do at the end
of these podcasts where I do anAI roulette question, and so
this has been great.
I could keep going with allkinds of stuff.
So I load a question intoperplexity and it says the
prompt is I am doing a podcastwith Chris O'Neill.
Here's his LinkedIn profile andlatest posts, along with his
company website.
At the end of the podcast, Iwant to ask him an AI roulette
question, where you will createan unexpected and spicy question
(32:22):
that I can ask.
Chris, give me a good one.
So I just hit go here andhere's a spicy, unexpected AI
roulette question.
This is unedited.
Chris, you've led teams atGoogle, evernote and now
GrowthLoop, all at the cuttingedge of data and AI.
Imagine it's 2027.
Growthloop's AI has become sogood at predicting customer
(32:42):
behavior that it startsrecommending not just marketing
actions but actual businessstrategies, sometimes
contradicting what yourexecutive team thinks.
If your AI chief strategyofficer and your human board are
deadlocked, who gets the finalsay?
And would you ever let the AIoverrule your own judgment if
the data was compelling enough?
Chris O'Neill (33:04):
I love this question, but I do believe that,
for the foreseeable future,humans will remain in the loop
and ultimately be making finaldecisions.
There's nuance and judgmentthat is hard won over the years,
so this one's a relatively easyquestion for me to answer Is
(33:25):
the humans have the final say?
Ai and data in general.
It should inform decisions andstrategy, but not make it right.
I think that that is a uniqueskill that humans and effective
leaders have.
They can hold lots of context,and it's about having
(33:46):
consistency across a set ofinterlocking questions which
form your strategy.
So you know, I think that AIcan help and challenge and help
and maybe pave the way to gooddecisions, but no, I don't see
this being something that the AIoverlords way to good decisions
.
But no, I I don't see thisbeing being something, uh, that
the ai overlords just um, justto make decisions, so to speak,
and I do think that's agovernment's question too for
(34:07):
companies.
They need to figure out wherehumans oh yeah gonna remain in
the loop and, where agents canbe, can be working autonomously
on their own yeah, I totallyagree.
Kevin Kerner (34:15):
Um, I've done this since I've started the podcast
and every time the AI questionis so much better than my
questions Every time.
Chris O'Neill (34:25):
It's embarrassing .
I'm not so sure, kevin, you'veasked some really great
questions, but yeah, it is agood use case, right?
It's really fun.
It thinks much moreholistically than maybe we do
sometimes.
Kevin Kerner (34:37):
Well, chris, I've had such a great time talking to
you.
I could keep going, but I knowyou're a busy guy.
I've watched the I will put aplug in if you haven't seen the
Scott Brinker MarTech Daypresentation.
It's a demo so you really get agood feel for, like the actual
product and the UI which is.
I think that's why I asked theUI question.
It really struck me how YouTubestudio slash Google.
(34:59):
It is.
It's so easy to go through.
But if people want to getstarted with this or want to get
a little information, moreinformation, or just want to
connect with you, what's thebest way to to get started?
Chris O'Neill (35:10):
You can.
You can check out our websiteor you can find me in the
socials.
I try to maintain a goodpresence on LinkedIn is where I
spend most of my timeinteracting with folks, so if
people want to check out there,I'd be happy to interact with
folks there.
And, Kevin, thank you for thisconversation and for what you're
doing for the community.
It's an important thing you'redoing at a really obviously a
(35:33):
really interesting time in theworld of business and marketing,
so thank you.
Kevin Kerner (35:36):
No doubt.
Well, thank you so much, chris,and best of luck to you, and
hopefully we'll talk again soon.
Chris O'Neill (35:41):
And same to you.
Hey everyone, this is Kevin Kerner with Tech
Marketing Rewired.
Today's episode is a great one.
I sat down with Chris O'Neill,ceo of GrowthLoop, someone who's
led at places like Google,evernote and is now helping
reshape how marketers actuallyget work done.
We got into Chris's idea ofcompound marketing and what it
really means to have AI agentsas teammates, and why Chris
(00:20):
thinks we're headed into aresults-as-a-service world, not
just software.
We, chris, thinks we're headedinto a results as a service
world, not just software Talkedabout speed, smarter systems and
, honestly, how to take most ofthe pain out of modern marketing
while leaving time for all thecreativity.
And if you've ever been stuckin cross-channel chaos, you'll
definitely want to hear howGrowth Loop is solving for that.
Let's get into it.
(00:42):
This is Tech Marketing Rewired.
All right, hello everyone, andwelcome back to Tech Marketing
Rewired.
I'm your host, kevin Kerner, andtoday I'm very excited to have
Chris O'Neill, who's the CEO ofGrowth Loop.
For those of you that don'tknow Chris, he has a really
incredible career.
Up to this point, he's beenmanaging director at Google
Canada and Google X.
(01:03):
He served as the CEO andchairman of Evernote and Glean.
I loved Evernote, by the way.
I love it.
Thank you.
It was one of my firstnote-taking tools and is
currently a board member at Gap.
So, chris, I am super excitedto have you here on the podcast.
Yeah, it's great to be with you.
Yeah, yeah, it's going toreally be amazing to talk about
(01:23):
this new category of eugenicmarketing and what you're
building at Growth Loop.
One fun fact that I'm not sureeveryone knows is that Mookie
Betts is on your board, which isamazing to me.
Chris O'Neill (01:33):
He is he is?
He's an amazing, obviouslyathlete and human, very curious
and really wants to learn moreabout business, and we're
fortunate to have.
Kevin Kerner (01:48):
Yeah, that's great , I think, having that different
perspective.
Now I'm a Guardians fan and theDodgers just killed the
Guardians in this life series,so tell him to take it easy on
us this year, which I'm sure hewon't.
Chris O'Neill (01:57):
He will not.
Kevin Kerner (01:58):
They are very competitive.
Okay, so there are a lot ofthings I want to dig into, but
let's start with just thisamazing transformation that's
going on right now and theimpact for leaders.
You've seen all this waves ofinnovation across some really
impressive companies, and youhave an amazing co-founder in
David, and then Tamim, the CTO,who's doing some incredible work
.
I'm curious, from yourperspective, what feels
(02:19):
fundamentally different aboutbuilding a company right now,
from a CEO perspective, aboutbuilding a company right now
from a CEO perspective.
Chris O'Neill (02:24):
Yeah well, I've been around long enough to have
been through lots of differentpretty big disruptions or
changes from the shift to mobile, obviously, the advent of the
cloud and this one certainly haselements of disruption like
that in terms of how it meansthe interaction between
consumers or customers and howthey consume information.
That's all true.
(02:45):
The big shift here is really onthe supply side, in other words
, it's how we create, in thiscase, marketing, but really how
we go about the work of work.
Ai is fundamentally changingthat.
So it's really really, I think,the biggest thing that's
happened in my lifetime.
And then the pace with whichit's happening is unprecedented.
(03:07):
I mean, it literally feels likeevery week there's a new model,
a new protocol and it'sunlocking a whole host of use
cases we could never evenimagine.
So it's just so incredible andI know there's a lot of hype and
I do think that it iswell-deserved and in many ways
we'll see, the impact isprobably greater than we can
(03:28):
imagine right now.
Kevin Kerner (03:29):
Yeah, it really is incredible.
Yeah, mcp, microsoft's new NLweb announcement.
It's just, it's just reallyhappening so fast.
The models are getting so muchfaster, quicker.
Well, now clicking into thegrowth loop thing what was the
catalyst that that led you tobecome interested in this more
operational role inside ofgrowth loop?
Chris O'Neill (03:50):
Yeah, I've been inside companies, I've been an
investor, I've dealt with everytool over the years, and the sad
reality is marketing is justtoo slow, too manual.
It doesn't really allow toscale up to the opportunity
potential to deliver delightfulexperiences consistently to
consumers.
That's always bothered me.
(04:10):
It's bothered me for over twodecades I mean going back to my
time at Google.
We were obviously solving apart of that, but the large and
sad reality is most experiencesthat customers have are very
impersonal, they're verysuboptimal, and that doesn't
need to be that way.
So this is about, really, howcan you use technology to serve
(04:30):
up fundamentally differentloyalty experiences, lifecycle
interactions with customers?
And that's really what it'sabout is how do you apply
agentic AI in this case, on topof what's called first party
data, right To basically drivegrowth faster and more relevant
for consumers?
And we'll talk about thisconcept of compound marketing
(04:51):
because, really, that themissing gap is not only that
you're driving gains, but you'redoing so faster, right?
So this concept of compoundingis one I'm personally obsessed
with and I'm happy to dive intoit a little bit more with you.
Kevin Kerner (05:03):
Yeah, I would love to.
Yeah, it's one of the reasons Ihave not as much hair anymore
is because how hard marketingand marketing has been for the
last 30 years yeah.
So I'm, I'm, I'm very excitedabout where you're headed,
because it won't grow back, butmaybe it'll make things easier.
Um, let's hope, yeah, yeah.
So, um, you've been using youmentioned this term, compound
(05:23):
marketing, and I've seen thedemo and I can't seem to sort of
shake how impactful this couldbe.
Let's click into that a bit andexplain from your perspective
what this new category ofcompound marketing is and how
it's different than today'smarketing.
Chris O'Neill (05:39):
Let me zoom out for just a second.
I just have a fundamentalbelief that all great companies
create categories.
Right.
It's not just about saying, hey, we're like this person or this
company, but we're better.
It has to be about difference.
It's to say like we'refundamentally approaching the
world in a different way.
Think of all great companies,companies like Qualtrics or
Google.
Evernote, in its day right, setout to create a new category.
(06:02):
It's just fundamentallydifferent.
We're solving a differentproblem in a different way.
So that's the framing abroadand that really has to do with
how you drive growth and how youtalk about benefits.
When we looked across the marketand it's super crowded, as we
know tens of thousands ofdifferent tools and companies
chasing this in this broader,you know, marketing technology
(06:22):
landscape.
So you have to cut, cut throughthe noise.
So it starts with that.
The other thing I'll add isI've been obsessed with compound
interest in investing since avery young age.
I remember parents I told thisstory recently and I grew up in
a very small town in the snowbelt of Canada and I was given
this book called the WealthyBarber and it talks about this
(06:43):
metaphorical wealthy barbernamed Sam who basically lives a
very simple life, but followed aformula that really involved
compound interest and doingsmall things over time to build
some substantial wealth, notjust in a quick way, but
overnight, slowly and surely.
And that's the power ofcompounding the eighth wonder of
the world overnight, slowly andsurely, and that's the power of
(07:05):
compounding, you know, theeighth wonder of the world.
So it got us to thinking how doyou apply that concept that's
very familiar in investing tomarketing?
So the difference between, youknow, a 1% gain every month
versus a 2% gain every weekisn't just a little bit, it's
actually 13, 14, 15x, right.
So we're here to basically helpmarketers apply AI to their
(07:26):
data to move quickly, morequickly, through the entire
cycle.
Right From figuring out, I havean idea that's going to move
churn or reduce churn or perhapsboost lifetime value.
How do you take it from thatidea to actual impact?
We're about compressing thatimpact.
So, idea to impact cyclethrough marketing who do we talk
(07:48):
to?
What do we talk to them about?
When do we talk to them?
And then how do we measure itall?
And then lather, rinse andrepeat by basically figuring out
what worked against ourobjective function of lifetime
value or reducing churn orwhatever it happens to be and
just continuously doing that,and that's the name of our
company, growth Loop, but itreally is leveraging the power
of compounding which, again Imentioned, is considered the
(08:11):
eighth wonder of the world.
So we're thinking and borrowingphrases and concepts from other
parts of the world and applyingthem to marketing.
Kevin Kerner (08:20):
Yeah, fascinating.
Yeah, we talk a lot about speedto outcome and so when I heard
the compound term, I was like Ilinked directly to that.
It's like how do you get tothat outcome faster?
It's like moving that outcomefrom here down to here and
there's a bunch of little stepsto get to that outcome and so if
you can compress that, you justget there faster.
What were the act like?
(08:49):
Like seems like solving this,solving this compound marketing
thing must have had severalactivators.
Like what did you work on tomake sure that that you could
begin to realize this vision ofcompound marketing?
There's a bunch of differentparts of it.
Like what are they?
Chris O'Neill (08:58):
Yeah, yeah, it really all starts with data.
It really all starts with dataand one of the things that is
also happening in the backgroundis not by accident or
coincidence is the improvementof the quality of data and the
semantic layer of data.
So the rise of data clouds isreally a big part of what's
going on.
So for too long data has beensiloed off so it's very
(09:19):
difficult to kind of stitch ittogether and then access it.
So the democratization of datathrough the rise of things like
Snowflake and Databricks andGoogle BigQuery and AWS, et
cetera, that's a really big partof it.
So it really starts there.
It has to be the quality ofyour data and your data strategy
and the guardrails you putaround.
It is where it has to start.
(09:40):
Ai is not very useful, machinelearning not very useful if
you're doing so with poor data.
So it starts there and then itreally we started to experiment
with not just removing frictionfrom very manual processes,
we're sort of automating theend-to-end marketing cycle.
So that has to do withself-serving for marketers,
right?
Historically it means I have anidea I got to go talk to a data
(10:02):
team to write some SQL and pulla query and then you bounce
back and forth to see is thistoo big, too small?
Do we have the right audienceas a starting point to at least
have the conversation, make anoffer that would help boost
loyalty, or whatever it happensto be?
Now, first step was to removeand automate the entire
end-to-end process, so thatwe've done that and that's
(10:22):
really well-received, and that'swhere we got our start.
Now it's about infusing AI atevery one of those steps and
having a supervising agent tobasically make sure that these
swarms of agents are actuallyworking together in unison,
consistent with the guardrails,consistent with governance
practices, so it doesn't returngarbage.
So, essentially, we have thismindset of an end to end process
(10:45):
.
We're trying to automate.
Now AI allows us to do eachstep of the way better in some
meaningful way, and then, in thefullness of time, we see, with
humans in the loop, an end toend process.
So that's really what this isabout, and we're really
delighted with the early results.
I want to be equally clear,though like we have a long way
(11:05):
to go.
These models are incredible andthey're advancing at an
unprecedented rate.
The slope of the improvementsis astoundingly great and
there's way more work to do tobasically boost the quality of
each one of the steps, whetherit's the audience agent we have,
or our journeys agent, or, uhwhat.
There's literally a dozen plusof them that we're experimenting
(11:27):
with right now.
Uh, and we're, we're nevergonna, we're not gonna rest
until the quality continues toto exceed anything even
imaginable.
Um, with humans doing the sametask.
Kevin Kerner (11:37):
Yeah it's the real unlock for me is the fact that
you know the agents talking toworking together and not being
singular agents that are working.
They are working on their pointfunctions, but the fact that
they're able to talk together,work together, is really a lot
for me.
Getting siloed marketing databetween channels and analyzing
(11:58):
it is one of the biggest pains.
It's the biggest pain in thebutt ever.
You're in Google, you're insocial, you're in some other
channels.
You get those, those data,typically separately and then
you have to analyze it yourself.
Are you?
If you're, you know, maybeyou're smart enough to put it
into some sort of llm and haveit combine it, but you're doing
that pretty much manually andyou're saying what you figured
(12:18):
out as a way to, um, have thoseuh channels analyzed in a
consolidated fashion using theseagents.
Chris O'Neill (12:27):
Yeah, that's right, and it's actually
upstream of that.
It's actually activating acrossall the channels, so once you
have ideas and you have specificoffers and specific things you
want to do across your channelsright now.
Not only do you have to stitchtogether all that stuff upstream
, to go and activate it requiresseparate discrete efforts and
(12:48):
manual efforts, separatediscrete efforts and manual
efforts.
We have abstracted that awaywhere you have a single place to
activate and then a singleplace to pull back and read back
into the data cloud the results.
These are the little thingsthat get in the way, right.
So it's about automating theworkflow and this is the work of
the work and why I think thisis so powerful.
(13:08):
So this application layer, ai,application layer things are so
powerful because the real worldis messy, right, and the insight
and the task-specificalgorithms you know, the
step-by-step algorithms thatstitch together, are going to
create and are creating realvalue right now.
So it's not just like any oneof the steps require manual
(13:30):
stitching together, it's kind ofall of them.
So that's where these agentscome into play and really I
think of them as like the glue,the glue guy or gal on a team.
They're really, they're reallystitching, doing the work to
stitch and pull things together.
So it's like having a thousanddifferent interns going off and
doing things you know again with, with supervision and bringing
it back in the loop at the righttime.
(13:51):
But really, that's what it'sabout is abstracting away modern
day TPS reports.
For those office space fans,right, it's like there's so much
mundane, soul sucking work thatreally can go away.
That's where agents will shineand that will allow humans to
(14:12):
unleash their creativity, theirpotential, by basically focusing
on the things that that humanscan uniquely do.
So, yes, jobs will befundamentally transformed.
Some jobs will go away, butmany others will be recreated in
a much more hopefullysatisfying and way that's
simpatico with the humancondition and what humans are
great at.
So that's how I see thisunfolding and, of course, no one
fully knows, but I do see theopportunity for humans to
(14:35):
interact with literallythousands of these agents in a
very simpatico way, uh, in asymbiotic way, um, to really
deliver real results that youcan quantify yeah, I totally
agree.
Kevin Kerner (14:46):
It's the 80 is 80, 20 rule.
Like% of the work is justmundane work and, as someone
that's been doing this for, youknow, 30 years, there's just a
lot of work that seems like itshould be able to be done by the
agents.
It'd be interesting to see ifthere's a new version of office
space out there that will be forthe AI.
Chris O'Neill (15:05):
Some other DPS report.
Well the agents are hard atwork at it already, kevin,
that's right.
Kevin Kerner (15:11):
Did you see VO3 last week?
Yes, I did.
Chris O'Neill (15:14):
How mind-blowingly cool is that.
I mean holy cow 8Z.
Kevin Kerner (15:22):
Yeah, I was playing it with my kids and they
were going nuts and I have onekid that's in more art school
and she did not like it at all.
She was like no, this is notgood.
Although I agree with you, Ithink I do think that the like I
do think that I mean I've beenthrough these technology
transformations before andthere's more work to be done.
As a marketer, there's it, andthe work that will you'll be
able to do, that you'll need todo, will be a different category
(15:44):
of work.
It'll give you more time to bethinking about the things that
need to be done and sort ofrunning at a whole different
level with these assistants.
Chris O'Neill (15:52):
Yeah, I think it's going to unlock more
creativity.
I might be in the minority here, but I actually think if you
have more tools in the hands ofcreators, designers, I think
it's going to unlock creativity.
So the mundane goes away andthen it unlocks more time and
energy.
It's going to be massive,massive change around the change
for people to absorb, let's beclear.
(16:13):
But I'm of the conviction andthe belief that it's going to
unlock creativity and we're just, you know, on the verge of
seeing that.
But it will be bumpy for sure.
Kevin Kerner (16:23):
Yes, I agree.
I have one quick question I'mjust curious about from the CDP
side of things Is your, do yousee an opportunity for the
agents to create data new fieldsinside of the CDP or begin
their own?
Are you doing that now?
Chris O'Neill (16:40):
We are.
We're experimenting.
The way we're experimentingwith that is like synthetic data
, right?
So what we'll do is we'llcreate a synthetic like a
synthetic profile human and it'sremarkable how well and robust
you can make a synthetic profileand then you simulate the
interactions with that.
So why does that matter to amarketing Like?
(17:02):
Is it the same as a human?
No, it's not, but it's really.
It actually allows you to drawconclusions with a simulated
behavior, which is generallywe're finding remarkably
powerful and accurate.
You don't have to go spend abunch of money on marketing to
test a hypothesis using A-Btesting in this case, right, you
can do a lot of that in alaboratory.
(17:23):
We're leaning into doing that.
It's one of the most excitingthings we're doing, frankly, is
just like this simulation ofhuman behavior.
Uh, without having to gothrough all the steps and spend
lots of time, energy and moneyto figure out what the response
was, you can simulate itbeforehand, which is super
exciting that is amazing,absolutely amazing.
Kevin Kerner (17:42):
I talked to a company yesterday that was a
research.
They had a long history inconsumer research and they were
doing they were talking aboutdoing that same thing, like they
were using their research datahistory you know, decades of
research data and putting itinto an LLM and then creating
these personas.
It was data that they had.
I thought it was superinteresting.
So the fact that you could talkto your persona or at least
(18:04):
learn from your persona usingsynthetic data is pretty wild.
Chris O'Neill (18:09):
Well, let me tell you a fun story about how we
created our category.
Kevin, we did a version of this.
It wasn't quite as extensive ascreating thousands and
thousands and thousands ofdifferent consumer personas, but
what we did, we said who areour target personas?
We're talking to CMOs, we'retalking to CTOs or CTDOs,
depending on the thing and webasically said, ok, great, we
(18:29):
fed it a bunch of informationusing the same technology
underlying Google's notebook LLMand we basically simulated
conversations back and forth tosay, hey, this is our
positioning, this is what we doand does this resonate with?
Like?
We simulated a podcast, so whatwe're doing and it helped us
refine it.
I'll tell you.
(18:49):
We actually narrowed ourcategory names down to two
things.
We said it was agenticmarketing or compound marketing.
We made the decision to docompound marketing for many of
the reasons I've alreadydiscussed, so I won't repeat
them, and also because itresonated with people, these
simulated personas that we wentinto LinkedIn Navigator and said
here are some actual personas,here is an actual content that
(19:11):
we're thinking of talking, andwe simulated these conversations
.
So it's not just a hand wavything.
We're using it to help informhow we position our company by
using these powerful tools and,we think, pretty innovative ways
.
So this is stuff we're actuallydoing today.
Kevin Kerner (19:27):
Amazing.
That's a great use case.
So I want to dig into a bit ofthe skills, the new stack and
the new skill sets that areneeded now.
I love the idea.
I hadn't thought of it beforeuntil I watched one of your
keynotes about agents as teammembers.
Like actually seeing the agentsas a worker on behalf of you
doing things.
It's kind of a really cool wayto sort of get your mindset
around it.
What kind of skillset shift doyou think the next generation of
(19:51):
marketers will need to adopt toharness these news agent call
them agentic teammates.
Chris O'Neill (19:58):
Yeah, many, many of the skills are actually the
same.
To be clear, I'll start thereand then I'll talk about what's
different.
The same is like look, I see aworld where individual
contributors are are going toneed the managerial skills that
great managers have today.
What does that look like?
Well, you need to be able toset really good goals, right.
You need to give reallyeffective feedback and you need
to figure out how and when todelegate.
(20:19):
Those are classic managerialskills.
In this case, it's just going tobe using an agent, right?
So agents are remarkably goodat following instructions, so
you've got to set a really cleargoal.
Are remarkably good atfollowing instructions, so you
got to set a really clear goal.
They're really good atresponding to sub prompts or
(20:39):
subsequent feedback.
And then the last part is likefiguring out when the human
should uniquely be in the loopor or when the agent can be set
free.
Like that.
That is a classic managementthat even individual
contributors are going to need,because I believe everyone in
the company is going to interactwith agents.
So that's where I'd say there'sa lot more in common than not
and this concept of as a teammember.
I'm a big believer.
I love this term glue guy orgal right.
(21:01):
These are the people that justelevate everyone's performance,
and that's what I believe agentsare right.
They're basically going toallow you to focus on the things
where you uniquely add value,and it's going to rise the
overall performance and elevatethe performance of everyone.
What does that mean?
Well, it means a differentmindset.
These are not here to take yourjob.
It's here to elevate andaugment.
So I'd say that the other thingI would say is you've got to
(21:23):
think about velocity right.
The real advantage here is theability to compound and go
faster, as we described.
So really not trying to findperfection off the bat.
It's about gosh.
You can launch so many moreexperiments, so it doesn't mean
you just do it for its own sake,but you basically have the
ability to not let perfection bethe enemy of done or good
enough, because with theseiterative loops, with agents
(21:47):
doing most of the work for you,you can get through it quicker.
So you have to have thismindset that velocity matters
more, and velocity is a functionof both speed and direction.
It's speed times direction.
If you think about physics,it's not just doing stuff for
its own sake.
It's about being intentionalabout the direction you're
taking and then really usingthese agents to think more
faster.
And then I guess, moregenerally, I'm a big fan of
(22:09):
slope over intercept in terms ofpeople's capabilities.
What I mean by that is theability to learn and understand
and apply insights to the nexttask you're going to go after or
the next set of challenges youtake on.
And I have a high school-ageddaughter and I just just started
(22:31):
college age son and what I'mreally encouraging them to do
and it's remarkable to see howthey use these tools.
It's amazing to me.
Yes, he's really become.
Become a ninja at using thesetools.
Be so good that people can'tignore you like this is like a
superpower.
The people, the youngergenerations, that are using this
stuff natively.
I mean, I watch how he studies,I watch how he basically
(22:58):
creates these flashcardsinstantly, compresses hours of
lectures into minutes of supersummary notes.
It's amazing and we're justscratching the surface.
So I would say lean into thesetools, become a master at using
them, because we have only juststarted to see what an operating
system that's rooted in thesellm models uh, with deep
personalization and deep context, will have.
It's just, it's mind-blowing tome at this stage and we're just
(23:19):
early in it.
Kevin Kerner (23:20):
Yeah, I have four college-aged kids right now and
um it's, I think they're movingfaster into this experimentation
, to this stuff, than thecolleges are certainly.
I mean, they are getting a lotof pushback from the colleges
and I am, of course, I'm on the.
I'm on the outer and bleedingedge of this stuff and I'm
trying to talk to the professorsand deans to say, look, you got
to get on board here this stuff.
(23:40):
You should be, you should betaking growth loop into
universities and showing thesome of the professors some of
the stuff that's going on here,because you would.
There's a uh, there can be aproblem with people not being
told not to use it at this point, which is really important.
Chris O'Neill (23:56):
We are doing a version of that.
Actually, a former colleague ofmine, Jim Lissinsky, is a, is
an amazing professor at Kelloggand uh, he, he and I, uh, he's,
he's, he's fantastic.
He is fantastic, yeah, so sowe're we're working with Jim on
a couple of different fronts.
Kevin Kerner (24:11):
That's great.
Chris O'Neill (24:12):
Yeah, like, like students need to to, to feel
comfortable with these thingsLike it really is.
I think it's going to be, youknow, again, as I said, it's
going to be like a superpowerfor them.
Kevin Kerner (24:21):
So it just their world, is going to be so much
different than I think what we,what we can even imagine right
now and I, I'm a very optimisticperson on all this yeah, I
think though, that there'll beone of the things you, one of
the things you've I'm sorry,chris, um, one of the things
that, um, I think you guys havesolved, for that is really
interesting from a skill setperspective is you've tried to
(24:41):
make the ui very easy andintuitive, like I like.
When I saw the demo, I was like, okay, the problem with getting
and I have a team that isreally good at using these tools
, but they're, if they use atool and they can't, it doesn't
work.
When they first use it, they'lljust drop it.
I'm curious, just from a userexperience perspective and
(25:03):
intentionality behind the toolbuilding the tool.
Did you take that into accountas you were, as you were
building the thing, trying tomake it as easy as possible,
take the technical part awayfrom the marketers so they could
focus on the goal?
Chris O'Neill (25:15):
Yeah, it's very intentional, to be clear, and
I'll comment more generally interms of the importance of
design, because I do think itbecomes more important.
But, yeah, absolutely, this wasdesigned for marketers.
We say it's designed formarketers, loved by data teams.
The data teams love it becauseagain, it removes a lot of the
low value added work of goingback and forth with SQL.
(25:36):
So the first iteration of thiswas very intentional to design a
beautiful UI that marketerscould just intuitively
understand from the beginning,and that is how actually we've
won businesses.
Like literally, when we'redoing demos and proof of concept
with people, you can almost seethe marketers.
It's so fun.
They're like wait, is that?
Like what?
Where's am I being punked right?
(25:57):
now, I just do that, did I, did,I just did I just literally
create an audience in like threeseconds with ai?
Like they're like there's thisbig smile comes across their
face so that's awesome so,absolutely it is.
And and I just on the on thecomment of design, I actually
think chats of not a great.
It's a suboptimal interface.
It's the best we've got, sowe're using it, but it's it
(26:18):
requires too much cognitive loadon the user itself.
You have to prompt and it'slike you know.
So I see a world with theseagents sort of these thousands
of agents, like these thousandsof interns you have will fade
into the background and you know, I think agentic AI is the
spark.
But it will be design thatreally brings the true power in
the form that we haven't evenimagined yet.
So I think that that is anelevating again back to the
(26:42):
point.
Like design becomes moreimportant, creativity becomes
more important to really figureout how we harness the
technology and allow it to fadeinto the background so we can
really focus on the businessoutcomes and delivering
creativity in service ofcustomers.
That's really how I see ithappening.
But, yes, that is adifferentiator for us that
(27:03):
people really do enjoy using ourplatform.
Kevin Kerner (27:06):
Yeah, that's an incredible vision.
I mean, if you can pull thatoff, it's pretty amazing.
A couple other thoughts here aswe're wrapping things up.
I've heard you mention the termresults as a service, and I
love the idea that there's achain.
There's not only you'rebuilding a category here, which
is hard to do, but you're kindof building a different go to
(27:26):
market.
Talk to me a little about youridea of results as a service
versus software as a service andhow that plays out for you guys
.
Chris O'Neill (27:34):
Yeah, a couple of things.
So software as a service isreally rooted in the concept of
a seat.
Right, it's linked to a human.
There's this X, many humansusing the software product,
right, and so what agents allowyou to do is to couple that
right, agents are workingcontinuously, they work
autonomously with supervision,so it breaks down that
connectivity, okay, well, whydoes that matter?
(27:56):
Well, it allows amazing thingsto be possible.
It allows for almost unlimitedscalability, right, there's like
all these different thingswhich humans just can't get to.
That you know, working 24,seven and having like the
ability to span, like thecapabilities, is really limited
by the compute, which is nothuman limited.
(28:18):
Okay, and then that also allowsyou to think differently about
the pricing model, the businessmodel more generally.
It's not we're charging you perseat.
You can start to do things liketasks or you can, even better,
do the actual outcomes.
You're selling work.
You can basically say look, wecan prove, quantifiably, prove
that there's a different outcomehere.
(28:38):
Right, maybe we moved churn ormaybe we improved lifetime value
.
So it's going to create newways, fundamental new ways for
business value to be exchangedand charged for.
You're seeing it with SierraBrett Taylor's company, which is
doing customer serviceinteractions right, they're not
charging per seat.
They're basically investingheavily upfront and they're
(28:59):
saying look, we'll charge for aresolved customer service case.
Right, that's their pricingmodel.
So I think that that's one ofthe more exciting things.
With agents, right, we havethese agent swarms and the
protocols that allow them.
We have the 80-20 you'retalking about before we're going
to, basically, humans are goingto work on the 20% that
actually is unique to them and80% of the other work will go to
(29:19):
agents.
And then, above it all is,basically we're going to start
to see oh my gosh, wouldn't itbe amazing if we could actually
align value here, where there'sactually a business outcome
based pricing, as opposed tojust artificial proxy called
seats, which has no bearinggoing forward?
And, if I may.
I think this is what's going on.
This is why Salesforce is goingout having to buy growth.
Like, if I may, I think this iswhat's going on.
(29:40):
This is why Salesforce is goingon having to buy growth.
The systems of record arereally in trouble.
You have the rise of these dataclouds.
You have agentic AI on topSystems of record have seen
their high watermark.
I really believe that.
So it's really going to be afascinating time in business to
see how this all unfolds.
Kevin Kerner (29:55):
As a service provider, an agency.
We're seeing the same thing.
It's like there should be somepayoff for can't be the old
(30:19):
model and I think the a lot ofthe unfortunately a lot of the
older, um larger holdingcompanies are going to have a
really hard time making thatshift, that cost model shift to
something that's more resultsoriented.
It's really interesting timesin this category I do think
you're right.
Chris O'Neill (30:37):
You saw it with programmatic, right,
programmatic was a pretty,pretty tough change.
This is even bigger than that,of course.
It really is.
I have lots of stories beyondthe scope of today's
conversation just about how youthink about all the different
agency jobs that they do and howthey're being changed and how
that value gets exchanged withtheir actual clients.
(30:58):
But this is not justadvertising.
This is an audit, this is anaccounting.
It's again back to our earlierconversation.
It's every corner of companies,you know, from the supply chain
all the way through to customerfacing, pre and post sales.
So it's really a big deal.
And now that's going to be thebiggest and longest poll, I
(31:19):
believe, is change management.
How do you get people toembrace these things and change
their behavior?
As always, right it becomes.
I had an operations professorin business school who used to
end every class we talk aboutgnarly supply chain problems and
like really the operations of acompany, and he'd always I'll
never forget this he said youknow, in the end it's an OB
problem, right, Organizationalbehavior or a human problem,
(31:41):
right, Like it's an opportunityor a problem that really traces
back to how do you get humans tochange their mindset and their
behaviors, and that's reallybeing laid bare.
Kevin Kerner (31:51):
So true, it's a revolution.
Chris O'Neill (31:54):
I'm excited about it, but it's a real human
problem.
Kevin Kerner (31:59):
Well, chris, I have one thing I do at the end
of these podcasts where I do anAI roulette question, and so
this has been great.
I could keep going with allkinds of stuff.
So I load a question intoperplexity and it says the
prompt is I am doing a podcastwith Chris O'Neill.
Here's his LinkedIn profile andlatest posts, along with his
company website.
At the end of the podcast, Iwant to ask him an AI roulette
question, where you will createan unexpected and spicy question
(32:22):
that I can ask.
Chris, give me a good one.
So I just hit go here andhere's a spicy, unexpected AI
roulette question.
This is unedited.
Chris, you've led teams atGoogle, evernote and now
GrowthLoop, all at the cuttingedge of data and AI.
Imagine it's 2027.
Growthloop's AI has become sogood at predicting customer
(32:42):
behavior that it startsrecommending not just marketing
actions but actual businessstrategies, sometimes
contradicting what yourexecutive team thinks.
If your AI chief strategyofficer and your human board are
deadlocked, who gets the finalsay?
And would you ever let the AIoverrule your own judgment if
the data was compelling enough?
Chris O'Neill (33:04):
I love this question, but I do believe that,
for the foreseeable future,humans will remain in the loop
and ultimately be making finaldecisions.
There's nuance and judgmentthat is hard won over the years,
so this one's a relatively easyquestion for me to answer Is
(33:25):
the humans have the final say?
Ai and data in general.
It should inform decisions andstrategy, but not make it right.
I think that that is a uniqueskill that humans and effective
leaders have.
They can hold lots of context,and it's about having
(33:46):
consistency across a set ofinterlocking questions which
form your strategy.
So you know, I think that AIcan help and challenge and help
and maybe pave the way to gooddecisions, but no, I don't see
this being something that the AIoverlords way to good decisions
.
But no, I I don't see thisbeing being something, uh, that
the ai overlords just um, justto make decisions, so to speak,
and I do think that's agovernment's question too for
(34:07):
companies.
They need to figure out wherehumans oh yeah gonna remain in
the loop and, where agents canbe, can be working autonomously
on their own yeah, I totallyagree.
Kevin Kerner (34:15):
Um, I've done this since I've started the podcast
and every time the AI questionis so much better than my
questions Every time.
Chris O'Neill (34:25):
It's embarrassing .
I'm not so sure, kevin, you'veasked some really great
questions, but yeah, it is agood use case, right?
It's really fun.
It thinks much moreholistically than maybe we do
sometimes.
Kevin Kerner (34:37):
Well, chris, I've had such a great time talking to
you.
I could keep going, but I knowyou're a busy guy.
I've watched the I will put aplug in if you haven't seen the
Scott Brinker MarTech Daypresentation.
It's a demo so you really get agood feel for, like the actual
product and the UI which is.
I think that's why I asked theUI question.
It really struck me how YouTubestudio slash Google.
(34:59):
It is.
It's so easy to go through.
But if people want to getstarted with this or want to get
a little information, moreinformation, or just want to
connect with you, what's thebest way to to get started?
Chris O'Neill (35:10):
You can.
You can check out our websiteor you can find me in the
socials.
I try to maintain a goodpresence on LinkedIn is where I
spend most of my timeinteracting with folks, so if
people want to check out there,I'd be happy to interact with
folks there.
And, Kevin, thank you for thisconversation and for what you're
doing for the community.
It's an important thing you'redoing at a really obviously a
(35:33):
really interesting time in theworld of business and marketing,
so thank you.
Kevin Kerner (35:36):
No doubt.
Well, thank you so much, chris,and best of luck to you, and
hopefully we'll talk again soon.
Chris O'Neill (35:41):
And same to you.
Popular Podcasts
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.