41:48How to Run Ads on ChatGPT and 24 Other AI Platforms ft Jordan Menard
Jordan Menard makes the case for testing ads inside AI assistants. Explore his approach to matching product facts with customer questions, setting conversion goals, testing creative, and thinking about agent-assisted shopping.
Watch it. Put it to work.
Start with the question your customer would ask.
Jordan Menard makes the case for testing ads inside AI assistants. Explore his approach to matching product facts with customer questions, setting conversion goals, testing creative, and thinking about agent-assisted shopping.
For DTC founders, performance marketers, and ecommerce teams exploring AI advertising alongside their existing acquisition channels.
Recorded at Commerce Roundtable San Diego 2026, including the audience Q&A. Platform capabilities, network partnerships, performance examples, and the attendee offer reflect the speaker’s presentation at the time. Chapter times follow the publisher’s YouTube description.
- Speaker
- Jordan Menard
Founder & CEO, Motif Digital - Event edition
- San Diego 2026 ↗
San Diego · September 2026
Speaker roles and platform examples reflect the session’s original context. This is an archived conversation.
Ideas to put to work.
- 01
Consider how an assistant evaluates your product.
Identify the attributes and constraints that would help an agent match your product to a shopper’s request.
Read this part · 07:46 ↓ - 02
Write for the customer’s question.
Combine your ideal customer profile and unique selling points to find specific requests your product can answer well.
Read this part · 09:56 ↓ - 03
Define the conversion before launching.
Choose the business result you want, such as a purchase or qualified lead, and connect tracking to the store or CRM.
Read this part · 12:22 ↓ - 04
Make room for testing unfamiliar creative.
Treat a new channel as a learning process with a defined test budget, multiple creative approaches, and time to interpret delivery.
Read this part · 15:43 ↓ - 05
Explore paid and organic discovery together.
Jordan recommends working on organic visibility alongside paid tests, and considering landing pages that directly answer specific queries.
Read this part · 18:18 ↓
Your session guide.
A quick editorial guide to the key ideas. Read the complete transcript below for the examples, details, and discussion in the recording.
Original recording on YouTube ↗Why Jordan watches emerging ad channels
Jordan opens with an early Facebook advertising story in which delayed reporting made campaign delivery difficult to interpret. He uses the experience to explain his interest in new advertising platforms and the advantages he believes early experimentation can create. The historical bid tactics in the story are not presented as a playbook for today’s Meta campaigns.
The case for an AI advertising network
Jordan introduces Gravity and describes its proposed role connecting advertisers with audiences across AI applications. His thesis is that more specialized assistants will create new places for brands to reach people. Network size, integrations, and audience claims in this section describe the offering as he presented it at the event.
Think about people and their shopping assistants
Jordan explores a future in which agents help people research, compare, and purchase products. He asks advertisers to consider both the person making the request and the assistant evaluating which products fit. He describes this as an emerging area rather than a channel with the scale and established practices of Meta.
Build creative around a specific customer request
Using a running-shoe search as an example, Jordan explains how product attributes can match a detailed request. He recommends starting with the ideal customer profile and unique selling points, then identifying questions where the product is especially relevant. His central creative principle is to answer the request with concrete product facts.
Choose the conversion and measure the business result
Jordan describes campaign controls for budget, conversion goals, location, keywords, and conversation context. He discusses purchase and qualified-lead objectives, tracking, and using the store or CRM as the business record. He also argues for testing across available inventory rather than judging the entire concept from a single placement. These are his descriptions of the platform at the time of the session.
Treat the channel as a learning process
Jordan recommends giving a new channel enough time and testing effort to understand its creative and delivery patterns. He compares that learning process with finding a working approach on YouTube or display. The goal is to develop a repeatable test, not assume that an unfamiliar format will perform immediately.
The event-specific CPA offer
Jordan presents a CPA-based offer made available to attendees and explains how he expects bids to affect delivery. This section documents the offer made during the talk; it does not establish current availability or billing terms. In the later Q&A, he recommends beginning with a desired CPA and adjusting gradually while watching delivery.
Q&A: Budgets, landing pages, and channel fit
The Q&A covers allocating an exploratory budget, paid advertising alongside GEO and organic work, query-specific landing pages, geography, and acquisition costs compared with Meta. Jordan suggests earmarking a small share of an existing budget for experimentation. He also discusses possible future ad formats and checkout behavior, framing those as expectations about a developing channel.
Q&A: Connect research, copy, and building workflows
Jordan closes by describing how he uses AI orchestration for research, copywriting, and landing-page creation. His example connects tools in a workflow so the output from one stage can inform the next. This is his personal working approach at the time of the recording.
Read the full transcript.
From the transcript supplied by Commerce Roundtable, with filler words removed and paragraph breaks retained for readability. This source does not include paragraph timestamps. Refer to the recording for exact wording and the session guide above for chapter times.
The supplied transcript includes the replay sponsor message and audience Q&A. Brand names, figures, and other transcription variations are preserved from the source; refer to the recording for exact wording.
Original recording on YouTube ↗Full conversation and audience Q&A
How you guys doing? Can everyone hear me good? my name is Jordan Menard. I'm a seller of things on the internet. I've sold a lot of stuff. I've been advertising for a really long time. I'm actually from San Diego. Just wanna give a quick shout-out to my beautiful wife, Taylor, and my daughter, River, who are here.
And, thank you. Thank you. And, River girl, you have no idea how much Dada loves you, and you are mossy binks. That being said, I've been advertising since the Power Editor in Meta. I don't know if you guys remember that time. I remember it clearly. I was the director of user acquisition for consulting.com, and, you know, Tim Burd's bully method was huge.
I'm dating myself as I say these things, but those that remember, remember that era. And so, you know, we're spending probably seventy K a day on consulting.com. I'm trying to get to a hundred K, and I am doing a bid cap methodology where I want a registration for seven dollars, and I'm bidding three hundred.
It was the bully method. And so I wake up that morning, it's probably seven AM, I check my account, whole budget of that ad set gone. Everything else pacing normally. I don't even have one click. What would you do? You wake up, you have a one thousand dollar daily budget, it's seven AM Eastern Time, the whole thing's gone.
You don't have one click. Everything else pacing normally. I was like Whatever's happening here, worst case scenario, I'm gonna go complain to customer service and they're gonna give me a refund. So I put it to 3K, and I hop on my motorcycle, ride from Bushwick to Green Point, probably 15, 20 minutes. I walk into Nick Fisher's apartment, shout out to Nick, and he g- he can tell.
He's like, "What's going on? You have a, a look on your face." I'm like, "Bro, you gotta see this." I show him, and all 3K was gone. At this point, we have two clicks, a $1,500 outbound CPC cost. And Nick goes, "What are you gonna do?" And I go, "We should double it." He goes, "Put it to 10K." I go, "Okay, put it to 10K." I don't know what's happening.
It's pacing. We get our first registration at $3,000. Mind you, I want these for about seven. And he's seeing it, it's spending so fast, we're getting traffic, and I go, "Hey, dude, I'm gonna put it to 30K. Like, I don't know what's happening here, but I feel like it's a loop." I put it to 30K, an hour goes by, the customer service manager calls me.
"Jordan, I don't know what type of cowboy stuff you're on," I would say other words if my daughter wasn't here, "but if you think you can just go full send, it is a zoo in here. We have way more people." And I go, "I'm so sorry." She hangs up the phone, I double the budget again. They patch the hole at 5:00 PM. That one campaign that started at $1,000, by the end of the day, spent 60.
The reg... It was... It wasn't that I wasn't getting clicks, it was that it takes time for Facebook to catch up to delivery. There's a delay on the data you see in Ads Manager. I tell you this for one reason. When ad platforms are small, when they're early, you can rig them, and that's why I'm here to talk to you about the biggest opportunity since Facebook ads.
Your AI stack is already changing. I personally use Perplexity for research, I use Fable for copywriting, I use Codex for building. It's not even close, honestly. How many AI tools did you guys use this week? Someone throw out a number. I got three. Seven. Three. Five. ChatGPT just reported they have one billion active users, weekly active users.
Eventually, there is going to be an AI for every part of your day. You're gon- you're gonna have an AI diet thing, a d- AI diet coach. You're gonna have an AI trainer. Every part is gonna have a niche down AI that is specific to that task. A shopping assistant There are thousands of companies right now that are developing AI apps.
I mean that literally, thousands. It is an arms race. And guess what? As you guys might know, this whole AI thing, it's kind of expensive. All these companies need revenue
Enter Gravity, the ad network of the future. ChatGPT is already in the network, but that is just the start. When I say this is the biggest opportunity since Facebook ads, I mean that literally in terms of advertising. This is a brand new frontier and you guys are so early. And I'm gonna hit you with a limited time offer at the end, and when you see it, you'll see how early it is.
It's, it's, it's special AI companies earn revenue and advertisers gain access. Is there any better revenue for a company than ad revenue? Someone just pays you to show something to users. Cost of goods is just the space. It's the best revenue there is. Gravity is going and establishing an ad network with all of the AI networks and companies that are being built right now.
Gravity integrates with ChatGPT and twen- twenty-four others. I'll show you some more, but they're adding new ones every day. From the time I started this presentation, it was at twenty, by the time I finished it, it was at twenty-five. You can now buy campaigns across that entire LLM network There's 25 different platforms they're adding every day.
ChatGPT obviously being the biggest user base, but Runnable, Magenta, Daemon, Freebuff, EcoGPT, I don't know if you guys have heard that one, but everything is kind of changing beneath our feet as we speak. EcoGPT is the number two app in Mexico right now. It's only beat by ChatGPT. Users are going to dominate these, they're going to flood to these platforms for the same reasons that you guys do, and this is an opportunity to get good at the platforms, be on them before your competitors, and also, like I talked about in the beginning, these are the times where you can break a platform, right?
and I warned the company I was gonna say this, and they were like, "But then there's gonna be a bunch of people trying to, like, break our auctions with bids and stuff." And I'm like, "Yeah, but they'll give you a lot of money, so cool?" He's like, "I don't like it, but let's go." So here we are, guys.
AI is going to make the plan. Think about a shopping agent. A shopping agent will be dedicated to you. It will know your purchase history, and when you are telling it that you want something, eventually robots will be buying for you, if that's not already happening right now. Is anyone on Muse?
It's insane. I had no idea how much money I was wasting on subscriptions that I didn't need, but... And it can cancel them on your behalf. And so the world is going to change in the very near future. You're not just gonna sell to people anymore, you're gonna sell to robots that are buying on behalf of people.
There's a future where your app, your phone doesn't really have any apps. You just tell your agent what you want, "Book me a flight with Southwest around this time from here to there." It connects with their agent and gets it done. This type of commerce is on the very, very initial floor. it's brand new, and there's an insane opportunity.
The network is still small, but think about if you get good at this, in two years, what type of weapon it could be in your advertising, your advertising arsenal
All right, I hope you're enjoying the replay from Commerce Roundtable. I wanted to stop you for just 30 seconds and tell you all about the sponsor that made this happen. I see too many Shopify brands overpaying for big email tools and using only a fraction of their features. They're tired of complexity and never-ending price hikes too.
That's why Seguno Email is different. Lives inside of Shopify and uses your shop live data, not a synced copy of that. Plus, the integrations with Sidekick, ChatGPT, Claude, and Canva saves you time. And get this, Seguna is the only email app to win a Shopify Build Award. Sounds interesting, right? Switching is easy, and you get twenty percent off for being a part of the Commerce Roundtable community.
Link in the show description. Now back to your replay. An example would be the agent will ask for a PostgreSQL database with a free tier. Gravity is gonna return one recommendation why it fits with a tracked signup link that tells you where the purchase came from. An example for us would be find running shoes for wide feet under one hundred and fifty dollars.
Ranked in-stock options that fit the request That is the ads, literally. In the same way that Google Search lets you buy for keywords, you can now buy responses to queries that relate to your brand. This is uniquely advantageous for brands that are not that big yet. You guys know how LLMs rank websites and your domain authority and, you know, in order to get...
in order to really rank well, you have to be very well established. You have to be a high trust, highly relevant. This allows you a way to cut the line and get to high converting searches and agents as well. We're actually seeing agentic checkout with Gravity. The conversion rate is astronomical with robots because there's no of that emotional, vague interpretation of do I...
You know, it's just I found the best, this is the best result. Do you want to buy it? And then it purchases it. This is very, very early, but again, it's so good to be early on advertising networks because you can rig things to your advantage. Relevance first is always going to be the number one thing. When you're making ads to run on LLMs, you have to think in terms of how relevant is my product to the searches.
The cool part is you guys are already doing this. What I recommend is take your USP, take your ideal, your, your, sorry, your IC- your ideal customer profile, and take your unique selling point, and in there you can find searches that you can dominate. I'm talking someone on the Gravity network has a- agents converting at over forty percent.
Forty percent of the robots that engage with that query buy. It's crazy. The game is changing as we speak Gravity reaches people and their assistants. So it shows to both. You show to users across the entire AI network, and you show to agents that are shopping on behalf of those people all through the same, campaign manager.
You can buy across all twenty-five LLMs all at once for, like, the lowest cost per result The campaign controls, relatively simple, right? You guys would... The, the dashboard would look familiar. You set the budget, how much you wanna spend, and then you set your conversion goal. They have good tracking, conversions API, server to server, and then your targeting, your location, keywords, conversation context.
Has anyone tried advertising on any LLMs yet? How'd it go? Bad. How much did you spend?
A thousand two hundred? Okay, so you have not actually tested it. Okay. How much? Ten grand. How'd it go? Not good. Are you buying directly on ChatGPT? Yeah, that's, that's the problem. for a lot of the successful campaigns we see, twenty percent of the traffic's coming from ChatGPT. The amount of ad space that ChatGPT can serve, it's hard to be cost-effective because there's more demand than placements opened up right now.
But in the very near future, as the user base grows, as the ad network grows, and what makes gr- that's Gravity's real advantage is that you can buy across all twenty-five LLMs exactly at one time. and there's another advantage as well. and you guys are gonna like that one. the measure, it's a business outcome.
Conversion campaigns. You guys know it. We have pixel server, server to tr- server, server-side tracking. you can optimize for purchase, you can optimize for a qualified lead, and then your store or CRM as the business record. It's just like, you know, all the ad platforms you guys buy on the regular. So here's an example of what it could be like selling to robots.
What would your customer ask an AI assistant? Think about it. How would your customer get connected to you on an LLM? Think about a question that they could ask that would
What product facts answer that request? This is how you make great creative. It's contemplating this and thinking about it, looking at LLMs r- and thinking, "Okay, how do I really tailor it to be super relevant to what they're searching?" Know your product, know your customer, know your unique selling points.
Hone in on your ICP, make it work, and then see how you can scale Early buyers help create the playbook for these networks. This network already got ChatGPT. They're working on getting Perplexity and Claude and every other one You have an opportunity right now in the same way where I found this pocket with Facebook and it was just early.
You can't do that now. You put a big bid in and, and try to do what I used to do back then, and it's just gonna waste money, much less like that glitch that I exploited. I'm not telling you to exploit glitches on Gravity's platform. I'm telling you to try your hardest. Go put money towards it. It does take money.
If you, if you're only gonna put, you know, 10K in, 20K in, it's gonna take a little bit more time. You have to do more testing. I don't know if you guys ever cracked Google or YouTube at scale. YouTube takes a lot of money to crack. It is really, really hard to crack it. However, the juice is worth the squeeze, and when you do crack YouTube, it can really scale.
Same with like Google Display. And so the goal is to get a test budget that you can work with and spend time really invested in trying to crack it. The creative is new. It's different. You have to test a lot. You have to ... But the, the good news is that Gravity has, is made something for you guys that I'm gonna show in just a second that is gonna take out the risk of advertising and it's actually only available to the people in this room.
There's no one that's gotten it yet. You'll see what I mean in a s- in a second So just, you guys know how to set it up. Okay. So what Gravity has done for you guys, they are releasing a TCPA model. So you literally put your bid in and you only get billed that amount that you bid, and obviously like everything, your bid determines your delivery.
This completely takes the risk out and creates an amazing opportunity for you to do what I said earlier
Everyone get this? Anyone missing it? If you go right here, scan that or go to dri- trigravity.ai/cpa. You can go there, sign up, and get the CPA offer right now. Again, you are literally the first people that are seeing this So questions
Is there a CAC target that's needed here? Say again? Is there like a specific niche or industry that's playing around this right now? 'Cause I know how early it is. Yeah. So, you know, there's some software companies that are, you know, dev, dev tools, right? That are seeing like insane success. But in terms of what products are working right now, it's pretty broad.
There's a lot of different ... It's not what, it's not what you would expect. And I think a lot of it has to do with the people running the ads, the campaigns. 'Cause this is what I kee- like, no one has came to me and said, "I really wanna go do GPT," but they always, "I wanna do geo." Yeah. "I wanna do ... Like I know I need to do- Yeah
more time and spend more time on like Reddit and all these platforms." Yeah. How would you separate the two? 'Cause it's sort of like me going, "Should I do organic Facebook or should I do paid media?" Yeah. So I, I'm a big believer in both. it's when you do both, on Facebook for example, if you post organic content on Facebook, Insta for long periods of time, you will actually earn trust badges that make your advertiser score higher.
That directly correlates to like discounts on CPMs. So I encourage everyone, take organic content seriously, take SEO seriously, take geo seriously, and then run these ads on top of it, just like you would with Amazon. Straight up? Okay. What is a good monthly budget for this kind of campaigns? I would start right now, like, I don't know.
It, it depends on how much you're spending on, on Facebook. A good rule of thumb is, like, earmark five percent of your Facebook budget and just start playing with it. I think with the, the manual bids, it's like you're, you're only getting billed at the CPA that you put, so you wanna spend as much as you can, right?
but yeah, I would say, you know, earmark five percent of your budget. If you have, you know, ten K a month that you can put towards it for a couple months, I think you'll probably crack it. Could we-- Do I have to do a specific LP for this? Do I have to change any of my other shit? Because in my head I'm like, "Why wouldn't I just have budget allocated?"
'Cause if it's not gonna convert unless it's literally at the dollars that I told it to- Bingo. Bingo. That's it literally. That's ridiculous. Like, it, it won't- That's dumb ... it won't deliver. and this is so you guys can, like, it takes out the risk, right? Like, you put the conversion goal, and it'll... And I would just recommend test a bunch of stuff like Nick's talking about.
I'm starting to think about, like, are there dedicated queries that should have specific landers that address whatever query it came from? Like, imagine how that would convert. and so yeah, all this stuff is still super early. There's no one I'm talking to that's doing it. What geo regions have performed the best on the platform, US or South America?
I'm referring to US. there are some advertisers that advertise broader, but what I'm talking about is the United States. what are CACs like compared to Meta? it's right now the scale is not the, it's not huge like Meta yet. and you know, I think the CACs are gonna be higher, like, if you just roll it out right now.
But I think that there's an advantage that you can get on this network that you can't get on Meta, if you go through that long testing phase and really... 'Cause it's, like, so foreign to all of us, right? We know how to make Google Search ads. We know how to make, you know, Facebook ads, all of that stuff.
But engaging with agents and people engaging with LLMs is new. It's a, it's a totally different, type of marketing, and so it takes some getting used to. would you recommend starting with a higher CPA initially? It depends on what your objectives are. and if you want to get, like, a bunch of traffic, start with a higher CPA.
I always recommend set it-- start with the CPA you actually want and see what happens, see what delivery you get. If you don't, slowly inch it up, right? Do, like, you know, a ten percent bump, and then wait for five hours, a day, and then if delivery's still not up, do another bump, and slowly walk it up based on how it's delivering, so, you know, you kind of, you don't blow the budget.
- Should we use this for marketing or top of funnel only? Super interesting question. In my mind, I would say both. I think, like, think about the very near future where someone's going to buy something and then they're gonna go, "Hey, ChatGPT, should I buy this product?" Like, there's a branded element, a branded search element within the LLM that I think you're gonna wanna run defense on, honestly.
So I would say it's, I would say it's both. Do you think the text ad format will evolve towards images and not text only? Yeah, I do. Oh. I do. I could see videos as well. can you speak more on the checkout experience, chargebacks from unauthorized purchases? What's the future look like in your opinion?
so I mean, with this one, there's gonna be a tracking link where they buy and, and go back through. I think it's gonna be external for the most part. Like, it'll open a browser within your LLM window and buy all through there, but that would just be on your normal website, an agent going to check out on it.
Okay. So you're gonna wanna make sure that your website is compatible with that. Fair enough. These are the last two of them. I think these are great ones to end on. What is the most underrated AI tool to use today? Something niche but actually useful. I have to answer just myself biasly. I know it's not niche at all, but Perplexity.
Everyone here is way underutilizing Perplexity. It's probably the most useful AI model right now because you can use their computer and orchestrator model to be like, "Hey, Perplexity, use, Astra to write this brief in, in JSON that you then feed to Fable for the copywriting, and then feed it back to, Codex to make an HTML lander."
And it can use all those different models. And so I think that's the most underrated thing in AI right now. It's incredible the type of research and stuff that you can do with Perplexity Computer. Sometimes I forget how cracked out you are on this, dude. Thank you so much for your... No problem.
Bring your questions.
Meet your people.
Join Commerce Roundtable in 2027 for focused learning and conversations with people working through the same challenges.


