The GTMnow Podcast

Ads in ChatGPT Are Coming. What B2B Marketers Should Do Right Now | Keith Delany, CEO Primer

• GTMnow

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 45:43

LinkedIn CPMs just hit $800. AI is flooding every channel with content. Outbound is dying. So where should B2B marketers actually spend their ad budget right now?

Keith Putnam-Delaney, co-founder & CEO of Primer, joins Sophie to break down the complete state of B2B paid advertising, what's broken, what's working, and where the real opportunities are hiding in 2025.

What you'll learn:
- Why LinkedIn & Google search are hitting a ceiling (and what to do instead)
- How to get 80% match rates on Meta and 70% on Reddit for B2B audiences
- The only moat competitors can't copy: your targeting
- Why you need to feed CRM conversion data back to ad platforms
- Holdout groups, attribution, and what actually proves paid ROI
- The truth about ads in LLMs (ChatGPT, Perplexity, Grok)
- B2B influencer marketing: the most undervalued tactic right now
- Founder lessons from 6 years and 3 pivots building Primer

Chapters:
0:00 – The paid advertising crisis no one's talking about
2:40 – AI's real impact on ad costs
4:06 – Google Search & AI Overviews: what still works
5:04 – LinkedIn CPMs: $20 to $800 in 18 months
6:39 – Why B2B brands need to be on Meta, Reddit & Instagram
7:01 – What Primer does (80% match rates explained)
9:29 – The device fragmentation problem & attribution
11:16 – Holdout groups: the statistically proven approach
13:30 – Which channel to start with (and minimum spend)
22:18 – Paid ad creative: feed the algorithm with variants
25:14 – B2B influencer marketing & thought leadership ads
30:59 – Why you MUST push CRM data back into ad platforms
33:31 – Ads in ChatGPT & LLMs: what's coming
35:53 – Founder lessons, mental health & iteration

Connect with Keith Putnam-Delaney: 
Co-founder, CEO of Primer
Twitter/X: https://x.com/kcpdelaney
LinkedIn: https://www.linkedin.com/in/keithputnamdelaney/

Host: Sophie Buonassisi, SVP at GTMnow
Follow Sophie: https://x.com/sophiebuona
LinkedIn: https://www.linkedin.com/in/sophiebuonassisi/

About Keith Putnam-Delaney:
Keith is the co-founder and CEO of Primer (https://www.sayprimer.com/), a B2B paid advertising targeting layer that helps brands unlock high match rates on Meta, Google, Reddit, and beyond. Former brand marketer turned data-driven growth leader.

GTMnow is the media brand of GTMfund, sharing go-to-market insights from the top 1% of revenue operators.

Visit us on: https://gtmnow.com
Follow us on LinkedIn: https://www.linkedin.com/company/gtmnow
Follow us on X (Twitter): https://x.com/GTMnow_
Follow us on YouTube: https://www.youtube.com/@GTM_now
Follow us on TikTok: https://www.tiktok.com/@gtmnow_
Follow us on Instagram: https://www.instagram.com/gtmnow_/

The GTMnow Podcast
The GTMnow Podcast is a weekly podcast featuring interviews with the top 1% GTM executives, VCs, and founders. Conversations reveal the unshared details behind how they have grown companies, and the go-to-market strategies responsible for shaping that growth.

Visit gtmnow.com for more episodes and other interesting content. 

SPEAKER_01

CPMs over the last 18 months have gone from like $20 on average to as high as like 800 we've seen with some of our customers, which is insane.

SPEAKER_00

Keith Putnam Delaney, co-founder and CEO of Primer.

SPEAKER_01

I really want a dev tool basically built on an agentic platform to make producing integrations way faster. Nobody creates an account on Reddit or Meta with a work email, you get like a 2%, 10% hatch rate. We get incredibly high, like 80% on Meta, 50% on Google, like 70% on Reddit match rates for our customers. And then that kind of unlocks a lot of interesting possibilities.

SPEAKER_00

There's a lot of talk about ads in Elements. What is your perspective on all this does?

SPEAKER_01

OpenAI just launched theirs from my perspective. I'm excited for it.

SPEAKER_00

What other kind of tactical tips would you give somebody if they were thinking about running or they are running paid advertising? Keith, welcome to GTM Now.

SPEAKER_01

Hey Sophie, thanks for having me.

SPEAKER_00

Yeah, it's great to have you here. We're gonna cover everything paid, and there's a lot to cover, but let's start off with the high-level structural changes because we always say, you know, with AI, there's a lot changing in terms of from the investment lens, where value accrues. With product being easier to build, that value shifts over to distribution. We're seeing the same kind of structural changes happen to specific areas. What's happening with paid?

SPEAKER_01

I think the the really interesting thing about paid is in the wake of all these incredible changes in AI, is that there's actually a limited number of channels to capture people's attention, right? AI has not produced a new social network. It hasn't produced something that people are sort of scrolling through for content. What AI has instead done is it's sort of glutted the existing channels with more content. Um, so you have essentially a supply and demand problem. You have oodles and oodles of demand and limited supply. And that's causing a crazy increase in costs on a lot of digital advertising channels. Um and yeah, I I think it's one of the unsung stories of the AI sort of revolution we're going through.

SPEAKER_00

And with these rising costs, how should people be thinking about their acquisition motions?

SPEAKER_01

Yeah. Well, what's really interesting, you know, if you again think about AI and go to market more broadly, it's also created a glut on outbound. Uh, and everybody's talking about the decrease in effectiveness on outbound these days. Um, so it's sort of like, all right, what do you turn to? Um, I think a lot of people are looking at community and kind of event field of field marketing and events again, which is cool. And then on paid, I think a lot of people are left kind of scratching their heads a bit, um, uh, which is too bad because there definitely are opportunities. Um, you know, Reddit is seeing a surge in B2B demand lately. Uh, we are seeing more and more uh brands interested in trying to get Meta to work for them. Um, some very brave B2B brands are exploring TikTok. So there's definitely opportunity there. It's just that LinkedIn and search have diminishing returns at the moment, it seems like for most of the data that we look at. Yeah.

SPEAKER_00

Yeah. You nailed on the head where a lot of people are scratching their heads and you named a few channels there. Can we actually go through each of the channels and kind of break down what has changed now with AI and what people should be thinking about with each channel? Like what's effective, what's not. Yeah. Where should people be spending their dollars?

SPEAKER_01

Totally. Um, so with AI, let's let's talk search because it's always kind of like the bread and butter default channel that marketers go to because it's high intent, right? Somebody's searching for something, they want that thing. Um, the AI overview is sort of killing search, um, slowly but surely, especially unbranded search. Um, AI answers are popping up the majority of time uh in something like Google and more and more in Bing, too. Um, so what are people doing? What's still working? They're investing the money in competitor keywords, right? So they're bidding on anybody and everybody that that's a brand name that might be related to them. That's working. It's still actually proven to be quite effective. Um, it may continue to work for a while, but it also will suffer from diminishing returns as everybody else starts doing it.

SPEAKER_00

I have to say, yeah, I can't tell you every time I search for a brand, I'm scrolling down six different ads before I actually get to the organity of that brand.

SPEAKER_01

And it's really interesting. Google's like, no, we're not gonna do like when people are searching for brands, we're not gonna do AI answers because they need to still make money to fund all those CPU costs. Uh so that's that's search kind of in a nutshell on LinkedIn. Uh wow. I mean, LinkedIn is just like CPMs over the last 18 months have gone from like $20 on average to as high as like 800 we've seen with some of our customers, which is insane. Um, so a lot of people are looking for efficiencies there. There's still really cool um uh kind of ad formats that LinkedIn is playing with. Uh Thought Leadership, probably being the one that's getting the most attention right now. Their CTV placements are actually still pretty effective. Um, from a uh and the neat thing about C CTV is you can't skip it, uh, which is kind of fun. Um so those are those are the kind of the two default channels I would say for B2B marketers. I think one of the things that a lot of uh folks don't realize is that there is the opportunity and possibility of unlocking more traditionally B2C channels. Um I think Clay just rolled out their ads product. I think it's a really interesting sort of um signal that uh that they see opportunity in helping users kind of stitch together, helping their customers stitch together identity between kind of like the LinkedIn level and that, you know, B2C Instagram, you know, me at night sort of doom scrolling.

SPEAKER_00

Yeah.

SPEAKER_01

Yeah. Yeah, yeah, yeah.

SPEAKER_00

All humans at the end of the day.

SPEAKER_01

Totally. Totally.

SPEAKER_00

And human behavior may look different B2B to B B2C, but truly it's just humans.

SPEAKER_01

Yeah. Yeah.

SPEAKER_00

Now I'm asking you all these questions about channels specific to paid because you are yourself an expert in the area. For anyone unfamiliar with primer, can you give a little bit of context to what you're building?

SPEAKER_01

Sure. So um my background wasn't always in paid. Um I uh I used to be more of a kind of a brand marketer. Uh, and then I got really intrigued by data and targeting, because at the end of the day, marketing doesn't really work if you can't get it in front of the right audience. Um, and what we've built essentially at primer is uh a targeting layer. Um, so one of the main problems with kind of trying to tar target people on Meta or Google or TikTok or Reddit is they don't have those B2B sort of filters built in natively, right? Um, so most folks try to upload a list of work emails, right? Right. And because nobody creates an account on Reddit or Meta with a work email, you get like a 2%, 10% match rate. Um, we can change that. And we do. We get um incredibly high, like 80% on Meta, uh 50% on Google, like 70% on Reddit match rates for our customers. Um and then that kind of unlocks a lot of interesting possibilities, a lot of not only cheaper inventory, but still really effective channels. Um then we're doing some cool things on the measurement front. Um, but yeah, I think what we're really excited about is nobody really owns the B2B paid acquisition space uh right now. Um, and we see an opportunity to build something powerful there.

SPEAKER_00

Well, and definitely not across D2C channels, like you mentioned, and being able to actually bring the identity of the B2B person over to those channels. That's a huge gap. Um, a huge gap right now.

SPEAKER_01

Aaron Powell Totally. Um but people are let like I said, Clay's moving into the space, we're in the space. Yeah. Um I think people see that the economics of these sort of two default channels aren't really working and they're looking for new opportunities. And where the money goes is where the startups will go.

SPEAKER_00

And what what does that actually do for any kind of measurement? Because I think that's a huge constraint for a lot of people that are, you know, behind the concept of, okay, we're humans, we spend time on all these different platforms. We're all scrolling different various things um at night.

SPEAKER_01

Yeah.

SPEAKER_00

But it's so hard to measure. I know. It's so hard to measure. And especially because, you know, we might scroll something on Instagram tonight, but a lot of the time a demo, for example, might be booked on desktop in the daytime. So that you've kind of got the the channel fragmentation and expansion, uh, whichever is more suitable there. And then you've also got, it sounds like the device fragmentation too.

SPEAKER_01

Yeah. And the device fragmentation is sort of uh not something I think marketers give enough credit to. Um, you know, even on LinkedIn, the majority of your engagement on LinkedIn is on mobile these days, not on desktop. Um, so LinkedIn is the easiest to attribute because it's still you still people click go on desktop, they're at work, they scroll through LinkedIn, they click on the ad, they go all the way through and they sign up, which is great, ideal. I wish everybody did that. Um but they don't. Um so the challenge, and there is no easy answer. If somebody had nailed attribution, I mean, attribution is where so many startups have gone to die. Um so attribution is tricky, um, but I do think like it's just a process of triangulation ultimately. Um uh my friend Pranav over at Paramark uh is doing some cool things on the incrementality and media mix modeling side of things. Um, I think that's where a lot of the future lays for the sophisticated paid teams and for earlier stage marketers. Um, you really have to kind of not just look at first touch, last touch. You really have to sort of embrace um causality. Like you have to do holdout groups, you have to even something simple. Like, hey, we turned up the dial on Meta. Um, we saw a ton of leads come in in May. We turned it off in June. We saw you know uh a significant drop. That's that there's a strong correlation there. Is you know, let's all right, meta works for us. Let's keep investing. It doesn't tell you how much to invest, but at least it gives you signal.

SPEAKER_00

Yeah, directional data. Is holdout groups the best way for people to do that?

SPEAKER_01

100%. Yeah. Hold out groups are the statistically proven way to do it, right? Um, you basically have a control population, doesn't see this ad campaign and exposed population that does. You compare the conversion rates, all other marketing touch points being equal between the two groups, and you see if there's a statistical lift. And like honestly, it's sort of uh going back to the good old days of like madmen advertising.

SPEAKER_00

Yeah.

SPEAKER_01

That's what they that's what they do, you know. Yeah, um, Zoom is super famous for their uh they never did any digital advertising, right? Only out of home and kind of sponsorships, and they would just compare markets.

SPEAKER_00

Yeah, right.

SPEAKER_01

And they would look in Google Analytics and just compare Nashville versus Louisville, you know.

SPEAKER_00

100%.

SPEAKER_01

Yeah.

SPEAKER_00

And a lot of companies are doing that now and and for out of home, just as it's grown, right? That is the best way. And it's funny, we've just calm overcomplicated things with attribution when in reality it's more the directional side, like you said.

SPEAKER_01

Totally. Yeah. And and it's just it's hard to build a mod everybody wants a model that's perfect.

SPEAKER_00

Yeah.

SPEAKER_01

And it just doesn't exist.

SPEAKER_00

Yes, 100%. Well, you mentioned something that's super important, statistical significance, even if you're doing holdout groups. What kind of size of company does this make sense to actually scale towards to have that kind of end that you are hitting statistical significance? Yeah. Who is this for if I've been kind of listening to this?

SPEAKER_01

Yeah, totally. Um, I mean, for the marketer who's trying to, you know, run a campaign to 500 companies or people, it's not going to be for you. Um, but uh you can do it with relatively small sample sizes these days. Like even uh, you know, a 5,000 person audience given enough time, um, you can get to StatsIg.

SPEAKER_00

Yeah. That's yeah, pretty doable for a lot of companies.

SPEAKER_01

Totally.

SPEAKER_00

And what about um channel? Let's say you said the starting place is 5,000 people. What about a channel starting place? If you're you know advising any kind of companies and marketers building, is there a channel that they should be starting with, or is that perhaps nuance to each company?

SPEAKER_01

Um I still think the easiest place to test is LinkedIn. Right. Yeah, just because they've got the filters built into their product, right? You can define your ideal customers right there and and try to go after them. Um so it's it's definitely the easiest place to test. One of the things that I'm I'm really hoping that we'll we'll we'll get into as a as a company in the next year or so is the the barrier to entry to testing advertising is just very high, right? Because the the cost of experimentation is really high. Yeah. Um that's one of the things I think about a lot is if I was a a you know kind of solo founder or seed stage startup, how could I how could I be comfortable making this investment um uh and expecting that it won't just be, yeah, it won't just be lighting $20,000 on fire, which to me might be a lot of money at that, at that stage. For sure. Um it's something that I think we we're gonna have some cool features around at some point.

SPEAKER_00

Ooh, very cool. We'll have to revisit this when it comes out. Super exciting. And but you're right, it is a hard barrier, kind of higher barrier to entry. And totally maybe just as we talked about at the beginning, certain areas become a little bit more commoditized, if you will. That is one area that kind of retains that uh barrier to entry, is paid.

SPEAKER_01

Yeah.

SPEAKER_00

So when we think about it at the macro level, you know, where are those areas and gaps and opportunities, and you call that a few different channels, including community, including paid. Like where does paid sit in terms of kind of where people are going? Are you seeing an uptick? Are you seeing people pull spend out?

SPEAKER_01

No, I think I think we're seeing people double down on paid, um, just be in part because outbound is, I mean, paid is the is the top channel for for most companies. Um it takes a lot of from and I should say for most um marketing teams and and kind of you know go-to-market teams. You know, products, obviously you have you, I mean, there's there's product growth loops that are really powerful. Um, but uh I think you know, paid is basically outside of product growth loops, the the biggest area of investment. Um and as outbound really decreases in effectiveness, um, people are turning money into pouring more money into paid because they can still make the their ROI pencil, even though it is getting more expensive. Um yeah, it's and what I think as a result though of that increase in cost is they're turning to the channels where it's cheaper to advertise and where they can still get that ROI.

SPEAKER_00

Um that's which are the Reddits, like you said.

SPEAKER_01

The Reddits, the meta, you know, the Instagrams. So people the places where people like where they're at and then and it's sort of kind of an obvious thing, but it's like people that's where people are spending their time. You should be where they are uh to begin with. Even if um even if LinkedIn wasn't so expensive, you should probably still be on Instagram because that's where your audience hangs out. Yeah.

SPEAKER_00

Definitely. No, it it makes sense. Are you seeing a certain kind of split between channels that people diversify across, or is that quite variable from company? Just as you know, marketers are thinking about their channel diversification.

SPEAKER_01

Yeah, I think um it's somewhat persona dependent. Yeah. Um, so Reddit, right, isn't for everyone. Um, you know, we've seen match rates on Reddit as high as 70%, 80% for IT and engineers, and then four percent for legal or procurement, right? Um, because it's a smaller community uh and and not everybody's on there. Um I think the same go like it's actually really interesting to think about the channels where people aren't on LinkedIn, you know.

SPEAKER_00

Yeah, 100%.

SPEAKER_01

Like doctors, not on LinkedIn, educate you know, people in education, not on LinkedIn. Um so those are some of the things that I think are are actually like kind of quite interesting as it really varies by persona. Um, but the majority, for better or for worse, and I think it's for worse, um, the the place where we see people continuing to dial up investment is on search. And I just think it's kind of foolish. Like, I I really do. I don't think I they they're like, oh, well, we're paying $8,000 for it, you know, for a good for an opportunity, you know. And it's like we can still afford that, but it's like sh but is that the right question to be asking yourself? Like, is there not something a much better, more cheaper, higher volume way to get that? Yeah. Mm-hmm.

SPEAKER_00

Yeah, why why are people spending more and doubling down then if it's only getting more expensive?

SPEAKER_01

Um, oof, that's a great question. Uh fear, I think. I'm trying something new. The the cost of falling on your face, you know. Yeah. Um, you know, sp like again, that lighting money on fire. It's hard to know. Paid typically takes a fair amount of experimentation to get right. Um I'll just I'll just say where I uh I I sort of teased it earlier, but I'm just saying like what I think is hugely missing in uh in paid is benchmark data. Right. It's like understand like no why it it boggles my mind that there isn't uh a company out there that doesn't aggregate and anonymize segment by persona and then say, okay, here's what's working if you're going after cybersecurity companies. Here's what, or you know, uh here's what's working if you're going after HR. Um, and sort of helping like provide that data to help steer. Now, does it lead to like a reversion to the mean? Yes, but yeah, like and then do will the best marketers always kind of still take a first principles approach? But I think it de-risks a lot of the the investment if you can learn from your peers in a way that's uh that's uh not compromising, I guess is the right word. Yeah.

SPEAKER_00

I mean, it would be hugely beneficial, and there's benchmarks for so many things. So what's the barrier to creating paid benchmarks? I can immediately what pops in my head is the fragmentation we've been talking about and people not actually having the data around what is effective because of the B2B event.

SPEAKER_01

If we keep growing, we'll have the we're starting to produce the data. Yeah. Yeah, we're we're working on kind of our first set of those. Um we're just kicking off some fun data science around that. We've finally kind of hit the scale around certain personas uh where we can we can produce that. Um, and we want to build that into the product. And that is what I think helps that maybe the marketer who's spending a million dollars a month on Google actually decide to put some of that money into Meta. Maybe that founder who's trying out marketing for the first time to like, you know, press the on button.

SPEAKER_00

I guess.

SPEAKER_01

Yeah.

SPEAKER_00

Yeah, definitely. I love it. Well, we're excited to see that um and then for companies that you're working with as you're growing and scaling, what we talked about generally what size of company or stage of company makes sense to do paid advertising. What stage of company does it make sense for uh them to look at primer?

SPEAKER_01

Typically um we currently, right? I think it's gonna change. Um but currently we serve companies with you know that are already investing about $10,000 a month on average in in ad spend. Um it's sort of that's sort of the litmus test we found for like, all right, we've got emotion, it's working, how do we and it's enough where it's we can make it more efficient, basically. Um I think at that point, what we've found is anybody spending under that is gonna turn it off, turn it back on, turn it off, turn it back on. And as a software business, we don't like that.

SPEAKER_00

So certainly not ideal.

SPEAKER_01

Yeah. Uh uh, so that's kind of what yeah, that's a short answer to your question. Yeah. Or maybe a long answer.

SPEAKER_00

People that have always on paid systems at that scale.

SPEAKER_01

Exactly.

SPEAKER_00

Cool.

SPEAKER_01

Yeah, typically around 150 to 100 employees as well is kind of what we find.

SPEAKER_00

Yeah. That's a really helpful benchmark.

SPEAKER_01

Yeah.

SPEAKER_00

And Keith, you know, a lot of people think about paid as just direct pipeline generation, but there's so many other outcomes that come from anything that gives you data from the customer. Yeah. How do you think about the actual utility of paid? Like, how should people be thinking about their programs and what quantitative and qualitative information that they're gathering?

SPEAKER_01

Yeah. I think one of the coolest things about paid. Is that if you think about it, right, it's it's the thing that's furthest up your funnel, right? Impressions are the thing that's furthest up your funnel. So if you're if you feel confident you're reaching your ideal customers, you can get a lot of eyeballs on a lot of messages in a really short time frame. Um, so it's incredibly powerful, not just obviously like message testing, not just to drive those, you know, form submits and leads, um, but message testing that can help inform your strategy more generally, like what your website should should say. Um you can really iterate with messaging at a much faster pace than you can anywhere else. And it and that is one of the huge benefits that obviously AI unlocks as well.

SPEAKER_00

Yeah, it sounds like the the actual learning loop itself from it.

SPEAKER_01

Yeah.

SPEAKER_00

Brilliant.

SPEAKER_01

Yeah.

SPEAKER_00

And what other kind of tactical tips would you give somebody if they were thinking about running or they are running paid advertising and they're trying to kind of get more juice from the squeeze? Like what are the kind of the tried and true tips that you have?

SPEAKER_01

Ooh, tips. Um It's a great question. Like I'm trying to think, ah, God, I could go in so many different directions.

SPEAKER_00

We can go in all of them.

SPEAKER_01

Um Well, okay. I'll I'll talk about some like some things that I believe to be true. Um more like philosophically, um you should the hardest thing for a competitor to copy is your targeting. They can copy everything else. They can figure out what channels you're on, they can figure out your creative, they can figure out your messaging. Um, all of that is is is easy for them to look up and and imitate. They can't figure out your targeting. Um so that can be a unique competitive advantage. Brandon Cammy, the former head of growth at Rippling, uh, he and I used to talk about that all the time. Um uh the second thing is um go where your competitors aren't, I think. Channel wise, channel-wise. Um I always think that's that's uh really helpful is um uh, you know, or go where the the market generally is. It's sort of take like that first principles approach. I I think it's like fun when, you know, out of home, I feel like is having a resurgence. Yeah. Um, because like it was like I I think, I mean, these these channels or these tactics are cyclical, right? Everybody rushes in to do field marketing, that's gonna decrease in effectiveness. So what's the thing they're not doing right now? Um the other, the other tip I have is um, or kind of more philosophically, I guess it'll be my third point, is I think these days, um buying credibility is a huge, has to be a huge piece of your marketing strategy. Um what I mean by that is you look look at what's happened in uh B2C marketing, uh, influencer is uh it and it's already happening in B2B, but I really do think that that we're only at the beginning of what that really looks like. Um, and one of the fastest ways you can grow is buying that credibility from people who already have the audience, as opposed to trying to create it yourself. And I think also be because of some trends in B2C, people don't seem to really want to hear from brands very much these days. They want to hear from trusted people, yeah, that they feel connection to.

unknown

Yeah.

SPEAKER_00

What influence does that last point have? And those are all fantastic points. Yeah, what influence does that last point have on companies' approaches to more programmatized platform paid spend? Like you're talking about meta, we're talking about LinkedIn. Yeah, totally. When we talk about the creator space, which you're right, I think is like the early days of B2B. Yeah, what does that actually do to somebody sitting in the seat being like, where do I allocate my money?

SPEAKER_01

Okay, so there's vehicles for accelerating that with paid. Like those those tools exist. Um, for example, on Meta, this has happened for a long time, where if you get a an influencer can share their uh basically give you access to their brand, and then you can promote their content. You can basically boost their their content. Which you can do on LinkedIn. And you can do that on LinkedIn, and then LinkedIn came out with that uh six to 12 months ago. Yeah. Um, and I don't I honestly don't know, don't see many brands doing that yet. And I that promoted thought leadership, that's not from someone within your own company, that's from someone outside your company. Um, I think that that is probably that's good, that's a hugely undervalued tactic. Yeah.

SPEAKER_00

I like it. I like it. That's interesting about meta too, because I know a lot of marketers are working that on the LinkedIn side or wanting to do more of that on the LinkedIn side, but haven't, you know, dumbled down enough on LinkedIn itself if that's where they want to start. Right. But certainly haven't crossed the chasm to Meta.

SPEAKER_01

No, and I I don't think like I don't think many B2B marketers are doing it on Meta at all. Um I do think that's just a big opportunity though.

SPEAKER_00

Yeah.

SPEAKER_01

Yeah. Yeah.

SPEAKER_00

So one thing you mentioned, Keith, was the actual targeting is something that your competitors can't copy. So if we talk about defensibility overall and paid, what happens to the creative side?

SPEAKER_01

Man, well, I think a lot of creative um LinkedIn is just so bad, but uh at at this, like they don't have nearly any of the tools that that Meta Meta's probably the best, Google's second best. Um uh LinkedIn and Reddit are the worst. Um uh on on Reddit, though, I'll actually give like quick aside. The key for Reddit is just um making your con making your ad content appear native, um, which is very which is actually easy to do if you're thoughtful about it. Um I think the the where where ad content on Reddit fails is when it just it's like you're in the middle of a subreddit and it's just like ad, you know, people just glaze over it. But if it looks like it's actually a thought, a thoughtful post, um, people will engage with it. Um shout out to Jason Laritzin at Confluent for kind of that playbook. Um in general, though, uh I think what's happening with Creative is um you're unlike on Meta and Google, is uh the the dynamic optimization is is huge. So you just basically have to feed the algorithm enough variance and enough mix of different formats and messages, and it will it will end up picking the winner. Um and it's getting better and better at that.

SPEAKER_00

Yeah.

SPEAKER_01

LinkedIn's got a long way to go though.

SPEAKER_00

So would that be almost that kind of moat in a way is your variant?

SPEAKER_01

Yeah, I mean your variants of it. Your ability to iterate variant and produce variants, particularly uh video, um uh is really, really important. Um that that is a moat uh or or it's something that it's a it's a muscle, I would say, that is important to develop. Uh and there's lots of cool tools out there to help with this, like Opus Pro, you know, something I use a lot. Um Riverside and yeah, there's tons. There's tons. Cap cut, all of it. And that that's only going to get better and better and better with AI. Yeah. And easier and easier and easier. Um the uh the key is that that uh that can be a very big competitive advantage and then um yeah, and then targeting. Yeah.

SPEAKER_00

How do you feel uh about AI videos? You know, you work with so many different companies that are iterating on all their different video variations. What's happening there? Have any like really doned into pure AI um variations?

SPEAKER_01

I haven't seen much of it yet.

SPEAKER_00

Okay.

SPEAKER_01

Yeah. I really haven't. Um I don't know. I feel like it has to happen at some point. Um I s I see a lot more like kind of AI edited, like, you know, AI B roll mixed in. Um yeah. But I haven't really seen a ton of like purely AI output content. Yeah.

SPEAKER_00

I'm curious what that does for paid, both on the the video and the image side. Uh just as we see a flood and maybe a flood of different variants that algorithms can run and test, if that will take away maybe the targeting does remain that final mode.

SPEAKER_01

Could be. Um, it could be. And and that being said, I mean, I I do think the the algorithms are also trying to take away aspects of targeting. Um, but it is still something you you have some power over in addition to the messaging. The the last thing I'll say is like if you're not pushing conversion data back into these platforms, um uh you're really in trouble to get success.

SPEAKER_00

Tell us more about that.

SPEAKER_01

Well, all right. So if if the out if ultimately if all these ad platforms are basically saying trust the algorithm, trust the algorithm, trust the algorithm, uh, what is the algorithm gonna optimize for? It's gonna optimize for whatever signal you feed it, right? So it's gonna say, all right, um you want form submits, great. I don't care how good those form submits are, though, right? Because I'm just the algorithm. I wanna, I wanna serve your ads as much as possible. Um so you have to feed it basically CRM conversion data back into the ad platform in order to say, hey, no, no, no, no, you have to optimize for this, this thing that's actually valuable to me, not the junk. Um, and the Andromeda algorithm is actually getting the the meta out ad algorithm is getting better and better and better at that, at a crazy clip. For as much as meta seems to be uh losing in some spaces, it's still really fucking good at at their uh the yeah. And I think a lot of people, a lot of B2B marketers like dip their toe into meta, test it, oh, it doesn't work, then they come back. But like if you haven't tested it in a while, it's worth another try.

SPEAKER_00

Yeah. And that's one thing I want to double down on because you mentioned earlier, it's really hard to get it right initially. It takes time. Yeah. So for companies, marketers, founders that are trying to scale to new channels, like what should they expect? What kind of like time horizon does it take to actually get that right and it to click? And how much money usually does it take to? Oh man.

SPEAKER_01

Um I actually don't know if I have a great answer to that question.

SPEAKER_00

Um can also be super variant.

SPEAKER_01

Yeah, that's what I mean. It can really, really it can be really, you know, kind of again, like target market dependent. Um I do think that there's uh like probably a floor of investment that you have to make. Um and then probably more than anything, it's just coming in with good test design, um, you know, proper A-B tests, um, uh knowing what metric you're trying to optimize for, um, and then giving yourself actually not not you giving yourself, but you getting maybe enough rope uh to not hang yourself, to to basically prove that you um to to run enough experiments. Um yeah.

SPEAKER_00

Super helpful advice. What about um I have to ask you, you're remiss if I didn't. There's a lot of talk about ads in LLMs. What is your perspective on all this?

SPEAKER_01

Does DVD, right? I mean, Perplexities had their um their kind of ad program in beta, obviously. Open AI just launched uh theirs. Um it's I don't know, from my perspective, I'm excited for it, but that's because I this is what I do. Um I mean, and if you think about how much money Google has made off of this for so long, I mean, it's going to it's it's gonna be really interesting to watch it unfold. Um and it like I'm I'm really interested in um uh in how well like actually what it's gonna look like. I haven't had the opportunity to test it yet.

SPEAKER_00

Yeah. When do you think ads will be live in ChatGPT?

SPEAKER_01

If you had to give a guess, um we should get uh Tommy Fink in here from OpenAI.

SPEAKER_00

Um that's right.

SPEAKER_01

Text them right now. Um I don't know. I would guess I mean s every so much over there seems up in the air at the moment, but in the next six months seems reasonable. Yeah.

SPEAKER_00

That's very soon.

SPEAKER_01

Yeah.

SPEAKER_00

That's very soon.

SPEAKER_01

They gotta make some money if they want IIPO, you know. Yeah, monetize those free users.

SPEAKER_00

100%. And I I mean I said Chat GPT, but it goes for every LLM Microsoft.

SPEAKER_01

Yeah, what what the that one platform I haven't mentioned is X, which is just excuse my language, dog shit. Um from an advertising kind of infrastructure standpoint.

SPEAKER_00

We need that asterisk on there, donate on X otherwise.

SPEAKER_01

But like, yeah, but like like you know, ads and GROC and all of that. I think I think that's a natural extension if they can figure that out. Yeah. I know they brought on a new CEO to help with the monetization.

SPEAKER_00

So we'll see. It will be very interesting to see. Well, we'll revisit that once we've got ads and LLMs and Access uh potentially running more paid opportunities.

SPEAKER_01

Definitely.

SPEAKER_00

Well, great. We've talked a lot about paid advertising itself, but you know what we haven't yet covered is yourself as a founder. Love to hear, you know, now you're over six years into building and you've got incredible operating experience also. If you look back in time, what have been some of you know the biggest lessons that you wish somebody had told you earlier in your founder journey that you would then pass along to someone else?

SPEAKER_01

I think everybody did tell me that's the problem. I just didn't listen. Um, no, I mean I I joke that this is the most expensive MBA anybody's ever uh paid for. Yeah. Thank you to my investors, the GTA fund. Um pleasure. Yeah. Uh I mean, it's been there's so many, you know, statements about being a founder. Um you know, one of the ones I love a lot is uh, you know, failure is a symptom of iteration, right? We have failed so many times. We launched a product, got great traction, you know, pivoted it, got great, you know, went down to nothing to like make it a you know subscription product, got it back up, you know, went down to nothing, and then went into the dark for nine months and rebuilt and pivoted again. And it it that that many times at bat, um, it takes that, you know, that grit that people talk about, that um that belief, that like irrational belief that this thing needs to exist. Um, which, you know, paid advertising, not everybody would argue needs to exist.

SPEAKER_00

But uh But I mean, it kind of does. It kind of if you like it's kind of one of the few things that actually has to exist. Because if you have these platforms that have advertising potential, like you need to be winning on them. You need to be winning.

SPEAKER_01

Yes, yes. It does like it, I know. Um I meant more like, you know, existentially. But uh resent. But so it's like a weird thing to feel deep, so deeply passionate about. Like this is something this is something I geek out about. Um, but I uh yeah, I don't know. It's just the everybody talks about the roller coaster. All those things, all the cliches are absolutely true, but you don't really realize they are until you go through it.

SPEAKER_00

Yeah. I guess told yourself at like that hardest nine-month period.

SPEAKER_01

Not to take it out on my partner.

SPEAKER_00

Yeah.

SPEAKER_01

That was really important. Um, it took me a long time to realize that uh I was the one responsible for my own happiness. I could I couldn't seek affirmation or sort of like, you know, sympathy from somebody where that it was just unfair to ask it from them, right? Like I should be able to give that to myself was kind of like that. It was a big unlock that helped me outside of work, um, uh be a lot have a healthier relationship with my work. Um, that was big. Uh it's okay to have panic attacks. It's normal. Um, get help, like talk to people, get help. Uh, that's really important. Um, yeah. I think I and that's one of the things that's changed a lot in Silicon Valley in the last, you know, five to ten years, is like the embrace of mental health with found for founders. I think that's really cool. Um and then yeah, just recognize those moments where um it it can be just like a a simple fork in the road. You could say, all right, I'm done. What does that mean for me? Or I I want to keep going, what does that mean for me? And just, you know, kind of analyze those two options and pick the one that's right. I I decided to keep going. Um and uh for better or for worse, uh, seemingly for better. Um, but it there's costs to keeping going too that you just have to pay in terms of uh your relationships, your your the opportunity cost potentially to your career. Um if it does because there is still no guarantee that if you even if you do decide to keep going, that it's gonna work out.

SPEAKER_00

Right. So yeah, a lot of things to weigh. But you're absolutely right. It's a very kind of beautiful time to see people being so much more open about their mental health for founders and just for overall humanity. It's a cool evolution that we're seeing and we're continuing to see.

SPEAKER_01

Amen.

SPEAKER_00

And as you've kind of gone along the founder journey, have there been any particular books or things that you've read that have really helped you in whatever capacity you found?

SPEAKER_01

Um, you know, the books that have helped me the most, if I'm being totally honest.

SPEAKER_00

Yeah, 100%.

SPEAKER_01

Uh it's kind of like escapism. Like for me, at the end of the day, it's funny. I like, I know I should do more business reading, but yeah, but I just like at some point I just need to unplug and fall into a different world. So I'm like a big sci-fi fantasy nerd. Um that's one thing, but that's kind of a BS answer.

SPEAKER_00

Um, it's not a BS answer at all. Actually, I read um really recently something that was talking about the cost of actually not reading fiction anymore, just as people are consuming more short-form videos and reading just pure business books is the imagination and what that does for your actual creativity, for marketing, for the ad creative, for actually thinking through these campaigns, because suddenly you're served the experience visually instead of reading the words and actually envisioning what that sci-fi world looks like to yourself. So it could actually have downstream effects.

SPEAKER_01

You're making me feel good about choices then.

SPEAKER_00

30%, yeah. Yeah, there's there's uh downfalls to all these or little trade-offs to all the the evolutions and things that we are doing. Totally.

SPEAKER_01

Yeah. Um I will say though, on more of the coming back to the mental health side, I think uh Untethered Soul was really uh amazing unlock for me to kind of you know uh let go. Yeah, anyway, mindful on the mindfulness journey, I'm still very much on, but um that has helped an enormously. Um I mean, I could lift I could rattle off business books, but I actually think that those, if I'm being like really genuine, those are my two answers.

SPEAKER_00

That's the only answer we take is the genuine one. So those are fantastic answers. I love it. And um AI. Okay. Yeah. More of the personal AI side. But of course, as you're building, you've implemented some really innovative AI solutions and you're personally using it and your team is using it. I'm curious if there's been any specific use cases that you use, other than your product, of course, on the paid media side, but for yourself company building that you know are less obvious but have been really helpful in your journey.

SPEAKER_01

Yeah. I mean, I'm sort of waiting for someone to create. I I really want a dev tool that's you know, basically built on an agentic platform to make producing integrations way faster. Um, something that could, for example, look at your, you know, kind of like a uh a version of cloud code or cursor that um is more specific. It looks at your code base, understands you know how you your frameworks for producing integrations, goes off, looks at API documentation, and turns out like a V1 of an integration, um, and then helps you with the infrastructure support around that too. I think that would be fucking awesome because integrations are such a bottleneck for so many, so many uh companies. Um, and you don't want to wait, you it they're somewhat commoditized, but they take effort to build. Yeah.

SPEAKER_00

Yeah. Interesting. Um that's what they're also expected. If you don't have an integration, it's actually a huge, huge um kind of red flag in the deal process.

SPEAKER_01

Totally, totally. And it's like 10 years ago you could get people to pay for them, right? Like kind of a premium firm. Now it's like, no way. Like I'm just gonna go to your competitor that has them. Yeah.

SPEAKER_00

Exactly. It's a non-negotiable.

SPEAKER_01

Yeah. Um it's like the what's the expression, the Big Mac, the french fries on the uh apple pie. Like you don't, you don't pay for like nobody's buying a happy meal for the apple pie, right? Like, yeah.

SPEAKER_00

No, no, certainly not.

SPEAKER_01

Um and then maybe more on the personal front. And it's just because I'm dealing with this all the time. It would actually be a terrible business because they would have no recurring revenue. Yeah. But like I would love it. I'm going through a home remodel, and I would just love if there was some like AI platform to help with all of that. Because the project management is like, I feel like I've got a whole nother full-time job on the side.

SPEAKER_00

You probably do. That sounds like a real task.

SPEAKER_01

Yeah, it just sucks because like I would I would happily pay for it, but as soon as the project's over, you know, I churn out. So terrible business, but would be real cool.

SPEAKER_00

Oh, that's awesome. Well, Keith, this has been amazing. If you had to leave listeners with one piece of advice, you've shared a ton of really valuable advice, but overall advice or some kind of um quote or life motto that you kind of live by through years of experience.

SPEAKER_01

Ooh, another deep question. Um, I don't know. I think what I said I said before, um, failure is a side effect of iteration. I think that's an Elon Musk quote. Um, that is something that I it's kind of like I I it reassures me about the choices I make. Um uh it's like, you know, you don't you don't kind of get to that best outcome unless you unless you fail, um, unless you try. Um so that's something that kind of uh makes me feel okay when I make the dumb move that I do from time to time, as we all do.

SPEAKER_00

Because they're not dumb.

unknown

Yeah.

SPEAKER_00

Exactly. Yeah, when I always think of it as the same thing, just in different wording, but failure is a data point. So it's like if you don't have data, you're actually uh in the worst possible position that you could be and you'll have data on how to the success front and the failure front. But um, if it's not statistically significant, how do you iterate? Just like the paid side, truly.

SPEAKER_01

Well, and it's a weird sort of motto to to take into parenting, kind of like it's a weird way to phrase it in parenting, but I do think about that a lot with my kids. It's like you gotta you gotta let them fail. Yeah, you gotta let 'em you gotta let them try something and have it not work out. Yeah. That's my, I guess, negative wisdom.

SPEAKER_00

Very, very helpful. Keith, this has been wonderful. Thank you. Really appreciate the conversation in time.

SPEAKER_01

Thanks, Sophie.