Customer Success Playbook Podcast S3 E54 - Katie Smith - How to Use AI in Marketing Without Losing Your Brand Voice

Send us a text

In this final part of the Katie Smith trilogy on the Customer Success Playbook, we enter the brave new world of AI and marketing. Host Kevin Metzger explores the promises and pitfalls of generative tools with Katie, who brings her thoughtful and grounded take on how businesses can embrace automation while fiercely protecting their authenticity. If you’ve ever wondered how to use AI effectively without sounding like a robot, this one’s for you.

Detailed Analysis: AI isn't going anywhere—and that's exactly why it's time to get strategic. Katie Smith walks listeners through the essentials of adopting AI in a way that enhances rather than dilutes your marketing. The episode kicks off with her advice on building internal AI policies: what your team will use AI for, what it won’t, and how to protect sensitive data along the way. Her mantra? Be proactive, not reactive.

Katie also shares her go-to applications of AI in the creative process:

• Use AI as a co-creator to spark content ideas and draft early versions
• Train AI with your brand’s voice and tone to maintain consistency
• Stay vigilant about hallucinations and homogenized content

She emphasizes the importance of human review at every stage, especially when publishing customer-facing materials. AI is a brilliant assistant, but not a final authority.

The discussion evolves into deeper insights on lead generation and real-time responsiveness. Kevin adds his own tricks for applying brand tone through prompt engineering and post-processing, offering a compelling use case that blends Claude, GPTs, and content repurposing magic.

Finally, the two zoom out to a broader question: How do you optimize your brand for AI-driven search and recommendations? It’s an emerging discipline with massive implications, and Katie teases what’s to come from leaders in B2B and digital strategy.

Whether you're testing the AI waters or already building internal GPTs, Katie’s thoughtful approach provides the guardrails needed to preserve quality and trust in a world of automation.

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 (00:05):
Customer success.
Welcome back everyone to theCustomer Success Playbook
podcast.
I'm Kevin Metzker.
Again, Robin's unable to join usthis week, but today we are
wrapping up our series withKatie Smith, the queen of the
Customer journey.
Katie has taken us through theimportance of sustaining

(00:29):
customer engagement and thepower of cross-functional
teamwork between marketing andcustomer success, and really
across the business.
And now we turn to our finalepisode.
AI and marketing and how to keepit real while embracing
technology.
Katie, AI tools are everywhere.
They promise better targeting,faster response times and deeper

(00:50):
analytics.
But how can marketers harnessthis power without losing the
human touch and brandauthenticity You value so much.

Katie Smith (01:00):
This is such an important question.
We're kind of on our own rightnow with how to deal with ai,
with our policies, with thingslike that.
So I think there's, there's twothings that you need to do right
away, which is create an AIpolicy within your company so
you can work with your lawyers,look at your brain guidelines,

(01:20):
look at what that says, and thenkind of figure out.
How you will use it.
This will change as thingschange, but it's just really
good to know how your companyplans to use ai.
Written down things that youwill do, things that you won't
do, especially if you're dealingwith data, especially customer
data.

(01:40):
So that can look like, um, ifwe.
Use AI to analyze data.
We will take out any identifyinginformation like name, address,
website, things like that, anyidentifier so that you know, you
can just replace that with like.
A, a naming convention or a codethat you have the code to.

(02:02):
Then it's also thinking aboutlike what kind of quality do you
want?
So I like to think about AI as aco-creator when we're talking
about marketing, things likethat.
It's really good at being kindof the middle, right?
So you have the idea, you can.
Freeform your idea in there, oryou can ask it for ideas, it can

(02:24):
organize it, but then you needto be there, there to, um, to
read it through, to make surethat everything is correct.
Um, you know, we talk a lotabout, uh, chat, GPT,
hallucinations, things likethat, coming up with, with, um,
information that's incorrect.
So, so you need to be there to.

(02:47):
To, to outline it.
And you need to put that intoyour policy of like how things
get checked and approved thathave been used.
Ai.
So that's really important thatyou kind of have a plan going
into it because nobody issetting that plan or regulation
for you.
And then it's, okay, how are wegonna use this efficiently?
So there's a couple ways Ialways suggest getting the paid

(03:10):
version.
So if you're using, um.
If you're using one of Chad GPTor Gemini, um, use the paid
versions and read through theirdocuments to see how they might
use your information to traintheir, um, these LLMs.
Because we wanna make sure Idon't want my information
training this model, and I don'twant my client's information

(03:33):
training this model.
They may not want that, so Ialways pay for it.
And I look for Gemini right nowis kind of the one that I think
has the best privacy policy.
Nothing is, I think we can sayguaranteed, but I wanna be doing
my due diligence and activelyunderstanding how they're using
my data.
Um, just like if you're going touse AI in any of your other

(03:56):
programs, read through theirprivacy policy and learn how
they're going to use your data,if they're going to use your
data, and if there's a tier thatyou can protect your data from
being used, that's superimportant.
So those are kind of the, likethe rules, I think.
For protecting yourself,protecting any clients,

(04:17):
protecting your, your customers,um, or your company's, um,
intellectual property.
So really important to lookthrough that.
So that's kinda the boringstuff.
Now, the fun stuff is how do youuse it?
So I like to use it as, uh, aco, like I said, a co-creator.

(04:38):
It's really good at generatingideas, and it's really good at
generating first drafts, soyou're gonna have to take some
time, but take some time totrain it.
So you, if you've purchased the,the higher ones that say they're
not going to use yourinformation.
You know, load all of yourcompany stuff into it.
Um, your brand, your website.

(04:59):
You have to teach it on yourcompany.
Teach the.
The tone, um, the, you know, allof those pieces and then you can
start asking it to kind of comeup with different ideas.
So content is a really greatco-creator to use.
You know, you can use, um, chatas a co-creator for that.
And you can come up withdifferent, you know, custom GPTs

(05:22):
for your marketing.
You have a social media one.
You train it, you know, you, youmake sure you get the right
prompts.
You can look up prompts.
Um, you can just Google goodprompts for marketing.
Uh, Pam Diner has a couple ofreally great AI books out right
now if you want marketingprompts.
Uh, she's wonderful and she'sgot a whole book that's just

(05:43):
marketing AI prompts that arereally helpful.
Uh, if you're not sure if you'renew to using prompts.
Just Google it.
Whatever you wanna do.
I wanna write a LinkedIn post.
What are some prompts I can ask?
I want to write, uh, severalblog posts.
What are some good prompts?
Um, look it up and then you can,you'll kind of get the rhythm of
how to write a good prompt.

(06:04):
Um, and then it's trial anderror.
So you try it out, you read it,you ask it to change, and you'll
get better.
But also it will get better at,at learning your program.
So.
Use it to come up with ideas,use it to write your messy first
draft, and then you come in andyou edit it, make sure things

(06:26):
are correct.
Um, and then it just, it takesthat content creation time.
You know, it cuts it in half andthen you can start asking it for
more ideas.
The, the only thing you wanna becareful of is if you're, you
don't wanna just ask it to writeits own blog post.
And if you're gonna do longercontent, sometimes it's good to
ask it, you know, like if you'regonna.

(06:47):
Write a blog post, you'll getbetter results if you ask it to
write it in snippet.
So here's the first bullet thatI wanna have my blog post.
Ask it that one bullet and thenyou can tell it how short or how
in three paragraphs, and thenyou can break it down.
If you ask it to do kind of likelong, long content, it will,

(07:10):
it'll just, it just gets kind ofvague and it sounds like
everybody else.
Um, and that's where you get thekind of content that just.
Doesn't, people are, it's justnot good.
Um, but if you work with it, useit as a co-creator, then it'll
cut your time in half.
But you'll still keep thepersonality of your brand.
You'll still be creating newthings.

(07:30):
'cause those ideas are gonna becoming from your brain.
And, um, and you'll, you'll beable to, to.
Kind of jump over the traps thatpeople are falling into with ai.

Kevin Metzger (07:42):
I think that was a, a lot of really good content.
As somebody who's spent a lot oftime in the AI space and focus,
you obviously have spent timelearning about it and like it's
really a lot of great tips thatyou kinda ran through there.
I have a couple questions and,and actually a suggestion that
I'm interested to see whetheryou've used yet or not, but

(08:04):
something.
You might wanna try, if youhaven't, one of the things that
you can do is create a GPT basedoff of the tone and voice that
you want to speak in, and thenyou can apply it post.
So like if you're writing a blogpost, you can actually at that
GPT and say, please apply thisto the post after you've.

(08:27):
Created it to make sure thatit's written in the correct tone
and voice.
So it's a really good way of,especially if you have content
that you've written over time.
The little trick I use is I feedthat content into Claude, ask
Claude to do an analysis of it,and gimme all the, the gimme an
analysis of what the tone ofthis writing is like, and then

(08:48):
take that analysis and make thatinto my prompt.
For how I want my, so it's kindof, it's a little bit meta, but
it kind of, it works reallywell.
Oh, that's a

Katie Smith (08:58):
great suggestion.
Yeah.

Kevin Metzger (08:59):
So yeah, little, little something that if you
haven't done that yet, you mightwanna try.
Have you used it?
So I, sitting in a, I went toAtlanta, had Atlanta AI week
last week.
So there were three days ofspeakers and sat and listened
to, um.

(09:20):
Different ways they're using it.
Um, I think marketing isactually one of the most
effective uses of AI right now.
Right?
Customer outreach, customer, theability, and, and I was just
watching something before we,uh, before we sat down the
number.
I, I don't, I'm not sure Ibelieve the number that he gave,

(09:40):
but the number he gave was like394% increase in lead conversion
based off of using bots torespond to a lead.
As soon as it comes in within aminute.
If they, if they respond withina minute, as soon as the lead
came in, they're responding.
They're using chat, chatfunctions either through, uh,

(10:04):
messaging or through one of the.
WhatsApp or something like that,running it through a bot to try
and basically convert the leadto the next step.
And um, they said they wereseeing like a 394% conversion.
I, there was somebody last weekthat was presenting that had
another.

(10:25):
Appointment bot.
So I think he was talking, hewas focused on real estate
leads, so it's a kind of a very,maybe it's possible in that
market.
But last week there was a botthat, um, I'm not gonna say the
name of the company, but thiscompany was basically bringing
in lead conver, doingappointment setting as soon as

(10:47):
the lead came in.
So following up via, via emailor via a bot on the site
directly to scheduleappointments immediately so that
the lead wasn't sitting andgetting stale while the customer
went off and started thinkingabout something else.
And they, they were saying thatthey were having a 97%, uh.

(11:08):
By by engaging right away, theywere seeing a 97% increase in
lead conversion.
So have you started playing withany of those things?
Are you seeing that, are youhearing that in the industry
yet?
What, what, what's your thoughtson that?

Katie Smith (11:22):
Yeah, so I was at a conference last week too.
It must be conference season.
Um, and we were talking about,so I have a question for you
after this, but Sure.
So we're seeing that, and, andthat can be.
A huge resource to yourcustomer, right?
So I really, when I think aboutmarketing, I try to think about.
What's the best thing for mycustomer?

(11:44):
Because if I think about what'sthe best thing for me, what's
the best thing for my company?
I miss things or I can do thingswrong.
So I always try to have the viewof how is this gonna help my
customer?
And this helps your customerbecause they don't have to wait.
They can respond.
You know, somebody is respondingto them quickly and that's
really what people want and theydon't want, you know.

(12:04):
Decision fatigue or the extrawork of having to go back and
check in.
Um, if we can eliminate thatwith a custom chatbot, that is
actually helpful, that would begreat.
Um, I think it's good for, forspecific things like, okay,
let's help you book anappointment.
Um.
That can be really helpful ifyou have a chat bot for asking

(12:26):
questions.
You know, that takes a lot oftime and development, but one of
the things that we talked aboutwas, do you disclose that it's
ai?
Because you don't have to, but

Kevin Metzger (12:38):
you do the answer.
The answer is you do always,because as soon as you don't and
the AI makes a mistake and youhaven't disclosed it, then
people are gonna get upset.
I, I mean that's universal atthis point.

Katie Smith (12:51):
Yep.
Yeah, I think that's part, thattransparency I think is really
important.
I think the other thing that weneed to be careful of with those
is that, so what we're seeing islike AI is now getting trained
on itself, and so we have, wehave to always be looking out
for it to become more and morehomogenized because that's what

(13:12):
we don't wanna do with ourcompanies.
So I like, I'll know.
Because I use, um, I use theseLLMs a lot and I know what
they're doing.
The first thing they present me.
It's like it's got all these, ithas these specific symbols and
it, and the structure's thesame, and I have to say, okay,

(13:33):
let's change up the structure.
Something like that.
But other people who don't, Iknow immediately that it's AI,
what they've written, becauseit's exactly what AI gave me.
So we have to be aware that eventhough.
We are, you know, using it withour specific prompts and our
specific company information.
Everything is just kind ofgetting homogenized.

(13:54):
So you have to check in onanytime.
AI is not, is kept coming faceto face with your customers that
they're still holding on.
They haven't started to likehomogenize your brand voice,
haven't started to like makethings flat.
I think that's one of thebiggest.
Downfalls we can have with theway the AI is working right now.
So you just have to be vigilantabout that, that you're not kind

(14:18):
of, it's not slowly likebringing you back to midpoint.

Kevin Metzger (14:22):
Yeah.
Yeah.
I think one of the other thingsthat it's, it's interesting
you're talking about.
Using AI to for brand identityand how it might bring your
brand identity down.
But the other thing that reallyis about to, I think, start
exploding now, all these ais aredoing search and they're making

(14:43):
recommendations.
So now how do you optimize yourbrand for ai?
And I think that's a sciencethat we are about to really
start getting into andexploring.

Katie Smith (14:54):
I feel like it can, it's an opportunity to, to do
better for people to get, Imean, how many times have you
done a Google search and thenyou're like, this isn't really
what I wanted, but you can getso specific with what you want.
AI to do with your questions,and because it knows you, it's
more of like a two-wayinteraction.

(15:14):
So I'm really excited to see,um, I'm not gonna be the one
figuring that out, but I, youknow, I really love following
people like Chris Penn and, um,some of the other, like B2B.
Marketers, you know, RandFishkin is always interesting to
hear what he has to say aboutwhat's going on with search.
So I love to just stay abreastof what, what folks are doing

(15:38):
and then be able to bring thosebest practices back.
But I think it's a prettyexciting time.
I.

Kevin Metzger (15:44):
No, it's definitely, definitely is, and
thank you so much for showing ushow AI can be a major asset when
used strategically and ethicallyto maintain authenticity.
That's a wrap for our three-partseries with Katie Smith.
You can connect with her onLinkedIn or find out more about
her work at Wild PathConsulting.

(16:05):
If you love these episodes,please subscribe, rate, and
share them with your team.
It helps us keep bringing youactionable insights.
We'll see.