Customer Success Playbook Podcast S3 E63 - AI Revolution in Customer Success: From Chorus to Custom GPTs

Send us a text

AI Friday delivers cutting-edge insights as John Huber reveals how artificial intelligence is transforming customer success operations. The conversation explores practical AI applications that are already delivering results, from conversation intelligence tools like Chorus for team coaching and customer context gathering, to emerging platforms like Sturdy that analyze unstructured data across email, Slack, and support tickets to identify churn risks and expansion opportunities. John challenges the overhyped notion that AI will replace CSMs, emphasizing instead how it amplifies human capabilities and enables more strategic engagement. The discussion culminates with an intriguing experiment: using custom GPTs for renewal pricing strategy that combines deal structure recommendations with benefit articulation. This customer success playbook episode demonstrates how forward-thinking CS leaders are leveraging AI to scale their impact while maintaining the human connections that drive customer loyalty.

Detailed Analysis

The episode showcases a mature understanding of AI implementation in customer success, moving beyond theoretical possibilities to practical applications with measurable business impact. John's progression from early adoption of Chorus to exploration of comprehensive platforms like Sturdy illustrates the rapid evolution of AI tools specifically designed for CS operations.

The discussion of unstructured data analysis represents a significant leap forward in customer intelligence capabilities. Traditional CS platforms focus primarily on structured data points, but John's experience with AI-powered analysis of emails, support tickets, and communication channels opens new possibilities for early risk detection and opportunity identification. This capability addresses a long-standing challenge in customer success: the inability to systematically analyze the vast amount of unstructured communication that contains critical insights about customer health and growth potential.

John's perspective on AI replacing CSMs demonstrates thoughtful leadership in an era of technological disruption. His emphasis on AI as an enabler rather than a replacement aligns with successful digital transformation strategies across industries. The human element remains crucial for building trust, navigating complex customer relationships, and making nuanced decisions that require emotional intelligence.

The custom GPT experiment for renewal pricing represents the frontier of AI applications in CS operations. This use case demonstrates how AI can be trained on specific business contexts to provide both analytical recommendations and strategic guidance, potentially transforming how CS teams approach contract negotiations and renewal conversations.

For CS leaders, this episode provides a roadmap for AI adoption that balances innovation with practical implementation, emphasizing tools that enhance rather than replace human capabilities.

Now you can interact with us directly by leaving a voice message at https://www.speakpipe.com/CustomerSuccessPlaybook

Please Like, Comment, Share and Subscribe.

You can also find the CS Playbook Podcast:
YouTube - @CustomerSuccessPlaybookPodcast
Twitter - @CS_Playbook

You can find Kevin at:
Metzgerbusiness.com - Kevin's person web site
Kevin Metzger on Linked In.

You can find Roman at:
Roman Trebon on Linked In.

Episode Transcript

Kevin Metzger (00:11):
Happy Friday everyone.
I'm Kevin Metzger without myco-host Roman, our guest, John
Humor.
We've already talked aboutJohn's number one tip for
building a CS function.
And debated who should ownexpansion.
Today we're zoning in on ai,specifically how John has

(00:32):
started using AI as a customersuccess leader.
John, we're super excited.
AI Friday, we always lovetalking about ai.
It's, it's, it's where we're atthese days.
Everybody's starting to look athow to use it best, best use
cases.
And so I think I wanna startthere.
Like, what's, what's your bestuse case for AI and customer

(00:54):
success right now?

John Huber (00:56):
Well, happy Friday.
And, uh, excited to, uh, to jumpinto this topic.
I, I think there's a couple,I'll go back probably a few
years and, and we implemented atool called Chorus ai, which,
you know, is a, is a AI notetaker.
I.
And I quickly learned howvaluable of a tool it was, and
really from a coachingperspective at that point, I

(01:18):
had, you know, roughly 35 peopleon my team across, uh, CSMs and
account executives.
And I, I couldn't physicallyjoin.
I.
All of the customer calls andinternal calls.
So to be able to go and not onlylisten to the recordings, but
also, you know, have it as atool to be able to coach, uh,
the team, uh, you know, on aperiodic basis.

(01:41):
But where I also found it reallyvaluable was I.
I could jump into a recordingahead of meeting with a customer
and really understand where arewe at in the customer journey
with this, uh, with thiscustomer?
What's the context of theconversation?
Are they a happy customer?
Are they a not so happycustomer?
And be able to get that beforeyou walk into a meeting.

(02:02):
For an onsite with a customerwas hugely valuable.
I think that's one example.
You know, another example, uh,more recently as I launched
Customer Success Architectsearlier this year, I spent a lot
of time really, I.
Trying to reeducate and reorientmyself with the technology,
right?
Your traditional customersuccess platforms, a lot of

(02:25):
those organizations are, arelooking at different ways to
incorporate AI into theirplatform from, you know, health
scoring from, you know,engagement opportunities with
customers.
But what I've found is.
There's a whole host of newcompanies that are starting to
emerge, that are built on AI andfocused on that, and I think one

(02:48):
that has really caught my eyes,a company called Sturdy, and
what they do is their purposebuilt for customer success.
They take all of theunstructured data that you
communicate with your customers,such as email.
Um, slack channels supporttickets, CRM data, and they take

(03:10):
and aggregate that data and runit through their large language
model to identify risk and risk.
Early on, and I'll be honest,Kevin, this is a tool I wish I
had when I was, you know, anoperator last year getting ready
for customer health and churnmeetings.
So it's pretty cool to see howquickly the technology evolves.

Kevin Metzger (03:31):
That's pretty cool.
I actually haven't heard ofSturdy yet, so that's the first
time I'm gonna have to checkthat one out.
That's definitely one of theareas where I'm starting to hear
more and more as a primary usecase is looking at real churn
reasons and uh, I.
Sat in the local, in Atlanta,there's Atlanta Customer
Success, and we had our monthlymeeting, I think it was last

(03:53):
week.
Yeah, it was last week actually.
Former guest was on or wasspeaking, uh, at that meeting,
but talking about the exact sameI.
Use case, they developed theirown process, but mm-hmm.
For doing it.
They weren't using a sturdy,but, um, went back through all
of the, they had gong in placeand they went through, back,

(04:15):
through all of the calls, gongcalls, and started looking for,
basically did a reclassificationof churn risks and reasons for
churn.
And we're able to really kind ofdo a, a much more detailed
analysis of why customers werechurning and it really actually
was able to see some significantchanges from what they thought

(04:39):
it was versus the updatedanalysis.
When you're really able to lookat that unstructured data and
capture the actual reasons whenyou go back looking through it.

John Huber (04:50):
Absolutely.
What's.
Really interesting about thosetypes of platforms, you know,
not only to be able to identifyrisk early on and, and those,
you know, hearing kind ofconversationally from your
customer, but also starting toidentify opportunities, right?
On the other side of it,boosting net retention and

(05:11):
seeing, you know, opportunitiesfor cross-sell and expansion,
you know, within your customerbase as well.
So it, it's fascinating howquickly this technology is
moving.

Kevin Metzger (05:21):
And it's really interesting because every use
case you kind of get and refine.
Companies are trying to figureout, okay, where do I apply?
Where do I spend money?
What the general guidance froman AI perspective is, start
small, find a use case thatprovides some kind of return
that makes sense to do.
Mm-hmm.
Try that.

(05:43):
And then make a decision aboutwhat you do next.
Coming up with each of theseindividual types of use cases,
whether, whether if you're notalready using it to do
transcription, recording and,and all that, you're definitely,
as a company, you're definitelybehind, right?
I mean, those are, that, that'skind of a fundamental thing, but

(06:03):
once you start capturing thatdata.
And really having all of thatdata now you can really, it, it,
the tools are there and theimplementation process is
available.
It takes some time, it takessome effort.
You gotta figure out what youwant to achieve with it.
But you can reclassify all yourturn cases and actually figure

(06:24):
out where, what, what, whereyour indicators were, what your
early indicators were.
You can really figure out what.
Hey, this customer, thesecustomers are talking about this
need they have, and our CSMshaven't identified it, but we
realize it's a need becauselook, you can have the AI go
back and look across all theconversations, and now you've

(06:47):
got product updates.
You've got expansionopportunities.
You, I mean, all of theinformation that has been tapped
is, or, or as you.
If you capture it, you can thennow tap it and you can tap it
with intelligence that's able tolook across all of the, all of

(07:08):
the data in a way that you justnever could before.
It's amazing.

John Huber (07:13):
It really is.
And it's, I think it's, youknow, it's gonna break down all
of those silos of data, youknow, that we've been talking
about for years and, you know,be able to tie all that data
together, especially theunstructured data.
So it's exciting

Kevin Metzger (07:27):
as you walk through this and, and see this,
and as we identify theseobviously.
Have you had any pushback fromCSMs or, or seen anything where
you've got people who areconcerned about what it's doing?

John Huber (07:40):
Not really.
When we did implement chorus,there was a little hesitation,
but I think I.
You know, when CSMs quicklyrealized how much more efficient
they could be, you know, wherethey could automate the task or
capture recordings, you know,versus trying to take notes and
have a conversation with thecustomer at the same time.
You know, everybody was quicklyon board and I think those that.

(08:04):
Aren't willing to adopt orembrace AI and AI platforms.
You know, they're gonna quicklyget left behind, unfortunately.
So, but I think, I think mostare quickly starting to realize
the benefits

Kevin Metzger (08:16):
I had put in a little, kind of lightning round
this week.
I'm not sure if you maybeanswered one of the questions
already, but we'll go ahead and,and, and give it a shot anyway.
Uh, so what's your favorite AItool right now?
I

John Huber (08:27):
would say sturdy from an enterprise perspective.
And then I'm a chat GPT guy, soI use that, uh, daily,
personally and for work.
So do you use the subscriptionor do you use the free

Kevin Metzger (08:39):
model?

John Huber (08:39):
Uh, I use the free model right now, so I've been
kind of pushing that and, uh,but I'm, I'm close to, to
pulling the trigger on thesubscription.
Subscription.

Kevin Metzger (08:49):
Yeah.
Cool.
All right.
Most overhyped AI customersuccess.

John Huber (08:55):
I think this is easy.
I think, uh, this, this myththat, you know, AI is going to
replace, you know, the CSM is,is silly.
There's still that human tohuman connection that I think is
required.
I think AI is gonna allow us todo more, but, uh, I don't think
you're ever gonna replace thehuman element.

Kevin Metzger (09:15):
And what's the, uh, next AI experiment on your
roadmap?
So

John Huber (09:20):
this is an interesting one.
I've heard this on a, um.
On a webinar about two monthsago, and it really p piqued my
curiosity.
So, um, the, the premise issomebody built a custom GPT to
support all of their renewalpricing, so to align with kind
of their pricing strategy andtheir pricing policy.

(09:42):
And so, A CSM could go in andsay, customer A, B, C.
Is interested in a three yearand a five year renewal term,
um, help me derive pricing forit.
And, you know, you can enter inuser data and product data and
just as you know, our deal deskor a pricing, you know, um,

(10:03):
calculator would, you know,provide those answers back.
It would provide it back, butalso provide it back with, you
know, benefits on why.
Um.
You know, that go along withthat renewal.
So I just, I found itfascinating that people are
starting to build these andtrain these models for, you
know, things like renewalpricing.

(10:24):
I think that would be a reallyinteresting, uh, you know,
perspective to, to explore

Kevin Metzger (10:29):
that.
That actually sounds like apretty cool, um.
Use case.
I, I, I like the idea, andespecially with the idea that it
can kind of help return, notjust the model, but hey, here's
the, here's the benefits, here'sthe upsell.
Package, basically givedelivering broader than just the
pricing, but the, the fullpackage benefits, the how you

(10:52):
deliver it, all, that probablycan really get, uh, wrapped in
That's a, that's a good usecase.
Jake.
Thank you for your time andreally appreciate all the, your
insights on how you're using ai.
That wraps up our three partseries with, uh, John John
Huber.
Connect with him on LinkedIn at.
www.linkedin.com/in/ John M.

(11:16):
Huber, H-U-B-E-R.
If today's episode sparks Ideas,hit subscribe.
Leave a five star review andshare the Customer Success
Playbook podcast with acolleague.
We'll be back next week withfresh tactics to elevate your CS
game.
Until then, keep on playing.