Navigating AI in Performance Marketing

The days of micromanaging every bid and budget are over. AI now makes many campaign decisions faster and more efficiently than humans can. But where does that leave performance marketers?

In this 30-minute session, we’ll look at the new reality: less manual control, more algorithmic decision-making, and a major shift in what performance teams need to focus on.

What you’ll learn

You’ll walk away with a clearer view of how creative, data, and measurement must evolve to meet the moment. And get clear answers about:

• How performance marketers can stay relevant as platforms automate more decisions
• What skills matter most when you’re managing outcomes, not tactics
• How to brief, test, and evaluate creative for specific KPIs, beyond ROAS
• Why structured data inputs are more powerful than clever campaign architecture
• How to align creative cadence and message with platform realities and funnel intent

---

Webinar Transcript

Webinar Transcript: AI in Performance Marketing

Speakers:

• Mohamed Hamad (Host)
• Sanjay Mayar (VP of Growth and Marketing, Imajery)

Mohamed Hamad: Hello, hello, hello everybody and welcome back to Third Wednesday Webinar. I’m your host Mohamed Hamad from Third Wunder. And today we have a special guest, Sanjay Mayar, who is VP of Growth and Marketing at Imajery.

Mohamed Hamad: I’ve known Sanjay for years. He’s been an exceptional influence in my understanding of digital ads and paid advertising and performance marketing. And we’ve got a lovely chat today talking about how AI has influenced performance marketing in the last while.

Mohamed Hamad: But before we get started, if you’d like to introduce yourself. Sanjay.

Sanjay Mayar: Yeah, thanks for inviting me. Mo. Good to, good to chat with you. So I’ve been in performance marketing for about 20 years now, give or take, in one shape or another. I’ve worked and led the marketing departments of different startups in Montreal, from online dating to fintech to luxury travel.

Sanjay Mayar: For the last ten years, I’ve been involved with imajery, so helping to become a leading performance marketing agency in Montreal. Very proud of the work we do.

Sanjay Mayar: Very happy to have done some great work with some great clients, built an outstanding team. Yeah, happy to chat with you today, and geek out about performance marketing.

Mohamed Hamad: Yeah, I mean there’s a lot to geek out on and there is as with everything AI and perform, you know, it’s, it’s infiltrated pretty much every scope of anything digital these days. And one of the things around performance marketing, looking at the last announcement from Google and what they’re, you know, they’re, they’ve been injecting it for the last while from like things like machine learning, influencing how how the ads are being displayed to different audiences and all of that.

Mohamed Hamad: But now it’s taken a step further and I just wanted to get your feeling of like how it’s changed the way you work or how you deal with ads in general. Like where does, where do you, where do you even start with that?

Sanjay Mayar: Yeah, it’s been it’s been transformative and it’s funny because at least to, to performance marketers, it’s been a huge part of how we make decisions, how we show value to clients, how we structure campaigns, how we invest, client ad spend.

Sanjay Mayar: When ChatGPT came out about a couple years ago, and then AI became the hot topic of conversation, it almost seemed like AI took over overnight as the major force.

Sanjay Mayar: But anyone who’s been on social media and has their content feeds Shown to them. Anyone who’s done a search on Google, anyone who uses Netflix, Amazon. We’ve all been experiencing AI in the form of machine learning for years.

Sanjay Mayar: Right. Anytime there’s sort of curated content or decisions made about how do we match the right thing, whether it’s an ad or a show, with the right person in order to get them to take the right action. Right? So the way performance marketers we used to do that was a lot of number crunching and a lot of experimentation.

Sanjay Mayar: So we build out our audiences, we’d find the exact right creative, we test the heck out of everything. We’d make, you know, decisions quickly with the amount of information we had. You know, we find something good and then we scale it out.

Sanjay Mayar: So AI has had, you know, for the last five to seven years, taken over a lot of that manual work for us. And our role has shifted from like an army of number crunchers and a lot of people that can, that can make decisions with the limited amount of data we have and then take action with the limited amount of targeting we have to really about how do we feed the maximum amount of information, to the algorithms to feed the machine learning so that it can make the best decisions, that we used to do more quickly, more efficiently, more effectively, and at scale.

Sanjay Mayar: And so when we started, certainly in imajery, but just the performance marketing in general, when we adopted that mind shift from how do we train people to be really good number crunchers to how do we restructure, rethink and feed the right data to AI across Google, you name it, that was a big game changer for us.

Sanjay Mayar: That was huge.

Mohamed Hamad: Okay, so you mentioned that it’s helped you deal with large sets of.

Sanjay Mayar: Data.

Mohamed Hamad: Taken the burden of number crunching to a certain degree. But in what other ways has it taken over? And what is it that you still want to maintain overview and oversight on?

Mohamed Hamad: It feels like there’s a fine balance between just complete automation and complete manual work. And where does that line lie in your side of things?

Sanjay Mayar: Yeah. So ultimately what we want to do is be able to drive the most amount of whatever the client wants, sales, revenue, you know, lead whatever, for the, for the least amount of investment.

Sanjay Mayar: And we want to do that consistently so we still own, we, as in humans will always own, the, the, the accountability. Right. If, if we’re, if what we’re doing isn’t working, that’s on us.

Sanjay Mayar: We can’t believe the tools. So we and then in order to, to to be sure that we’re accountable, we need to ensure that we’re the ones making the decision about how things are deployed. So we make calls on strategy, we make calls, on where the investment is going to be.

Sanjay Mayar: You know, are we pushing on YouTube? Are we pushing on meta ads? Do we explore TikTok or you know, do we throw everything to shopping ads? Right. So you know, we’ll have to make calls on where we deploy that investment, who we want to see those types of ads, what messaging do we want them to get.

Sanjay Mayar: We look at their behavior when they interact with the ads and they come to the website, and then we decide where in that process things need to be optimized. Do we need to fix the audience? Do we need to solve for creative, do we need to rethink the actual tactics themselves?

Sanjay Mayar: So AI has been instrumental in distribution. So taking our media plan and our strategy and then distributing it to the right people in a way that’s cost effective, but also, at scale and at reach, it’s also really effective in being able to make decisions a lot more quickly than we can about, you know, who should see what, how should we change our bids.

Sanjay Mayar: But it’s not a substitute for for sort of really getting into the core business objective and then turning that into a media plan to business strategy and then executing it there. So that’s the, that’s the first differentiator.

Sanjay Mayar: The other big differentiator is recognizing how the things that used to make us good performance marketers, you know, 10 years ago, seven years ago, aren’t the same things. So whereas we used to sit down with a client or sit down with with the business leader and then try to decide, you know, based on demographics, based on past purchasers, based on all this other criteria, who should see the ad and what should the ad say?

Sanjay Mayar: Now we can let AI make those decisions for us a, lot cleaner and a lot faster. It makes sense. Just because I happen to be, male, over 40, living in the suburbs, doesn’t mean that my behaviors and my actions are going to be, you know, aligned with every other male who’s over 40 who lives in the suburbs.

Sanjay Mayar: Right? Like, we’re not this, we’re not this homogeneous group that just acts and thinks the same way, right? So demographics are important. They’re not the, they’re not the main predictor of what action I’m going to take. Right? There’s all these other variables and Factors that could lead a, that could lead machine learning into making a decision about what my behavior will be.

Sanjay Mayar: Right. And a person or an army of people can’t look at all that data and then decide in real time, you know, with a degree of confidence, what action I’m going to take, whereas AI can. So really it’s been about acknowledging, and then leaning into that shift in, in in decision making.

Mohamed Hamad: So what you’re saying is AI’s got an insane amount of data that has been collected by the different platforms. So it can to a certain degree make some more informed decisions, better than, let’s say some like a, a performance marketer that might be limited in, in view and scope of what certain demographics in niche areas, you know, could be.

Sanjay Mayar: Yeah, in a nutshell. Right. So, so you know, we used to try to identify certain dimensions as predictors of behavior. So age, income, location, device, the, the, the cohort or the people that that people hang out with the content that people consume.

Sanjay Mayar: So if we want to target someone who wants to buy a car, let’s say there are a lot of signals that that person will give out before they, before they go and purchase a car. Right? They’ll do their research, they’ll go to different car dealerships, you know, they’ll read out, they’ll read up about you know, different models, fuel types, whatever it is.

Sanjay Mayar: Right. So all that behavior is shown is shown online. So traditionally the way that performance marketers would go and target that those people who are interested in the car is try to identify all the signals based on the user, based on how the user’s acting and you know, whatever other sort of fixed criteria they have, location, age, whatever else, AI can do that way better, way faster, way smarter than, than any person.

Sanjay Mayar: Right. And it looks at those signals that we would typically have and dozens and dozens more, in order to build a profile of someone who has a high intent to purchase is in research mode, you know, or, you know, it’s just a car enthusiast that happens to like read magazines.

Mohamed Hamad: Yeah, yeah, no, that’s super interesting. Okay. I mean like, you know, there’s, there’s a lot of positives to how AI can be implemented. But looking back at like say let’s say Google’s I.O. a couple of months ago where they unveiled a whole slew of new tools and new features within their, their ads platform and they, it seems like they’re trying to make it look like the promise of AI within ads is that you can just offload everything to the AI and it’s going to create the creative and the copy and you know, choose the audiences and all of that.

Mohamed Hamad: And like how much of that is hype and how much of that is something that is useful? I mean obviously we don’t want to rely on AI for everything but when it comes to let’s say the creative like where do you go there?

Sanjay Mayar: Yeah, great, great question. It’s it’s you know Google especially will talk on both, out of both sides of their mouth where on one side you know they’re promising us you know performance marketing is going to be like a self driving car where all you have to do is just give us money and sit back and we’ll do the driving.

Sanjay Mayar: And on the other hand they’re trying to you know, encourage us to adopt these incredibly sophisticated tools that you know really you need tons of experience, tons of exposure to and training to in order to master.

Sanjay Mayar: Right. So it’s really, it’s a, it’s a hats off to them for being able to, to push out both those narratives at the same time.

Mohamed Hamad: Oh absolutely.

Sanjay Mayar: Yeah. Creative, creative is, is, is the most critical piece. And I, I love that you brought it up because we certainly an imajery we really feel it’s often overlooked or often treated as a secondary consideration and it’s absolutely the primary and I think AI as a toolkit with respect to creative is massive.

Sanjay Mayar: I think it’s a good copilot, I think it’s a great producer. I think it’s really strong for output whether it’s copy, whether it’s images, whether it’s video, whatever it is it allow us to iterate it allow us to generate things at scale more cheaply.

Sanjay Mayar: It’s great for the client, it’s great for us because we can test things out. It’s no substitute for authenticity for having a strong hook for being able to tell a story. Well with creative that’s where maybe it’s a callback to the Mad Men type of marketing but that’s where performance marketers have an opportunity to step in, continue to manage machine learning and AI in platforms because it’s here but at the same time own the creative process and being able to speak to that both for cost effectiveness but also for the quality of creatives Use AI as a tool but we call it shots and with the Client, on creative.

Mohamed Hamad: Yeah, I mean across the board, AI’s promise is that it’s going to take over everything, that it’s going to create all your social media assets and all your content and write your blogs for you. And there is this dangerous message out there that you know, you just can let it do its thing without oversight and there is that garbage in, garbage out, situation that happens.

Mohamed Hamad: But to your point, if you have someone in the driver’s seat that knows what they’re doing, they can use it as a copilot, as a tool, to speed things up.

Mohamed Hamad: I want to kind of step back out of the AI conversation there and there is a conversation around privacy and attribution in ads. I mean, you know, AI brings in a whole different dimension, but we’re going into a new world around performance marketing with you know, with with the with the world going cookie less at some point.

Mohamed Hamad: I’m not sure exactly when that is. There’s this, it’s always being delayed. But yeah, where do you, where do you see that going and what’s your views on it?

Sanjay Mayar: It’s like this. Nobody wants to be no platform, and certainly no advertiser wants to be the privacy, bad guy. Right. So back in the day, meta got slapped with a lot of violations of misusing users data.

Sanjay Mayar: Google got slapped with it as well. I don’t know if I’m going to get in trouble with this. I remember talking to one of my Google reps, back in the day about, I don’t know, remarketing or something. We were asking about why can’t we do this?

Sanjay Mayar: You guys have this amount Google, you guys have this amount of info. You can do this, you have this platform. Why can’t we link the two together? We were asking about some sort of targeting capability and the rep looked at us and said just because we can do something doesn’t mean we will.

Sanjay Mayar: The amount of information that these ad platforms have about all of us, relative to the level of targeting, that they’ll allow advertisers to do is wild. So absolutely, data helps us make better decisions.

Sanjay Mayar: Ad platforms know that better than anybody. There’s this, there’s this real conflict between users wanting privacy, but at the same time, you know, wanting to avail themselves to free services.

Sanjay Mayar: So you know, how do you square the circle of wanting to use Google Maps but also wanting to keep your privacy, you know, wanting to, to do on platforms like Instagram and Tick Tock, but wanting to maintain your privacy. Right. And so I don’t think it’s really been solved.

Sanjay Mayar: I think it’s a, it’s a whole discussion about you know, whether it’s regulation or whether there’s a inherent responsibility for the platforms to, to, to give users better options or their users need to govern themselves. I will say from a performance marketing standpoint we need data in order to make decisions.

Sanjay Mayar: Right. Most of the data we get is anonymized. So I wouldn’t be able to run an ad specifically for Yumo. But we can do a lot with the information we have even when it is anonymized. And I agree.

Sanjay Mayar: I think the days of cookies are numbered certainly in the way that we’ve been using them for the last 20 years. Different regions have a lot of different rules. Certainly in Quebec, you know it’s it’s a, we have regulations similar to what they have in California and in Europe.

Sanjay Mayar: And it’s been a big struggle for us to be able to you know, show clients the value of this ad, led to, this action led to this purchase or lead when we live in a world that’s increasingly becoming more hostile to cookies.

Sanjay Mayar: So we’ve developed a lot of proprietary ways to be fully compliant with the law, fully compliant with ad platforms, use of cookies, but also be able to continue doing the things that clients pay us for, which is to serve the right ads to the right people, to get them to drive actions.

Sanjay Mayar: Right. And it’s a combination of of taking that data and re, exporting it back into the app platforms. So again machine learning can do a lot of this decision making. We’ve used server side tracking, to again stay compliant with cookie policies, but also be able to leverage that data.

Sanjay Mayar: Decision making. So you know, I think there’ll be more stringent cookie policies where whether you know, browsers reject them, ad platforms reject them or governments reject them. And I do think there’ll be innovators on the ad side, whether it comes from agencies or retailers or you know, the ad ecosystem in general to find ways to still do what we need to do.

Sanjay Mayar: It’s a really interesting time to be, to be a marketer, and to, and to watch this disconnect between you know, users wanting to have privacy, governments wanting to safeguard it, but machine learning being you know, Insatiably hungry for, for more data to be able to do what it does.

Mohamed Hamad: I’m wondering if the ad platforms themselves and the big companies that are collecting all of this data are to a certain degree moving towards the cookie less world where all of the tracking data that we’re used to is not available to us as the end marketer in the pipeline but it is still available to the big machine learning algorithms and the AI algorithms and they’re using that to shift the decision making over onto their side as opposed to building out their analytical tools and their AI decision making tools versus keeping us in the driver’s seat as marketers and performance marketers.

Mohamed Hamad: What do you think of that?

Sanjay Mayar: I think it’s interesting. I think whether we have data that comes to us via cookies or whether we have data that comes to us somewhere else, I do think it’s important to keep that data anonymized. Again I don’t want to have access to individuals actions it’s too much power for one agency to have.

Sanjay Mayar: So we don’t want Makes me think of how Google was faced with a challenge and a need to find a different solution other than cookie targeting in order to be compliant with European laws.

Sanjay Mayar: So they came up with this plan to deprecate cookies in GA4, Chrome, etc. And then used a combination of model data you in order to allow advertisers to do tracking, still use great tools like you know, Performance Max and AI search and Demand gen and all this cool stuff.

Sanjay Mayar: But also be compliant with what the European regulators wanted. And then this committee looked at their plan which involved taking data from, and signals from YouTube, Chrome, Google search, users, phones, Google Maps, whatever.

Sanjay Mayar: Which is what Google’s plan was to build a modeled set of data and identify users not based on cookies but based on their Google activity which is pretty vast. And then so the committee looked at this and said well that’s way more invasive than cookie tracking so we’re not going down that route.

Sanjay Mayar: Again I think this conflict between the capabilities of these behemoths of these platforms versus the public’s appetite, to balance their privacy concerns needs to be solved.

Sanjay Mayar: I think it’s going to be an evolving conversation and and yeah I think for people that are concerned about their, their, their privacy, you know it’ll be up to themselves to, to, to, to, to decide how much they want to share and their actions, you can take, whether it’s machine browsers or whatever.

Sanjay Mayar: But the amount of information that’s out there that can be actioned, monetized by platforms is, is massive. So you know, in the end I kind of think it serves a strong purpose. I like getting better ads and more relevant ads.

Sanjay Mayar: You know, if you’re planning a vacation and you go on, you know, travel sites, you know, you’re looking at flights, hotels, whatever it is this all happened to us, you’re going to get bombarded with ads for like a month, two months after about flights, hotels, rental cars, this, that, whatever.

Sanjay Mayar: Right. I would love to use that to my advantage as a consumer. I’d rather get advertisers competing for my consumer spend, showing me the best deals possible, the most relevant destinations possible, you know, the best possible flights, all that kind of good stuff.

Sanjay Mayar: And I think it’s, it’s up to retailers, brands, advertisers, not to do crappy ads with the incredible array of tools that we have. And I think more relevant ads to the right type of user, would drive a very different type of consumer reaction.

Mohamed Hamad: Absolutely. I’ve got a couple of questions, from our guests here and All right. So let’s go. How, how reliable do you find the AI responses and among the AI tools, which ones do you find the most reliable?

Mohamed Hamad: I feel like this is a very open question about pretty much AI platforms in general. But in your perspective, what do you think?

Sanjay Mayar: I really personally I use Gemini and chatgpt. I don’t know how reliable, I’d call them. It’s hard to say. I think they both have proven pretty, they’re not flawless.

Sanjay Mayar: But I have, we’ve used them to refine client pitches. We’ve used them as sounding boards when it comes to planning. We certainly use them for, for copy. So from a workflow perspective, I think it’s really good.

Sanjay Mayar: You know, we’re encouraging the different stakeholders and leaders in the team to adopt AI, again to help in their workflow. So whether it be in data analysis or CRM, or again you know, creative copy generation, we’ve used it in, in ad optimization scripts.

Sanjay Mayar: So for Google Merchant center, when we have a ton of data and we’re trying to make things, you know, we’re trying to optimize the data that we show in an ad space from gmc. So you know, we’ve incorporated AI to various degrees of success, into our workflows.

Sanjay Mayar: And I like having the different stakeholders at the agency, experiment with, different tools. So, you know, some people swear by clothed, others, especially in the paid ad side, love, perplexity. So, it’s a wide array of tools.

Sanjay Mayar: Actually, I’d be interested Mo in knowing what you think of this because, just, from offline conversations, I know that you’ve been, a strong adopter of AI in your workflow.

Mohamed Hamad: To a certain degree. We’ve used it for pretty much everything from writing code to writing copy to planning things out, analyzing, large sets of data and being able to triage things. Each one has its strengths and weaknesses.

Mohamed Hamad: I really haven’t played around a lot with Claude, mostly Gemini and ChatGPT, NotebookLM, and to a certain degree, perplexity for deep research. And each deep research does different things.

Mohamed Hamad: So it’s a lot of fun playing around with each one. And usually I’ll just run the same prompt through all of them at the same time and see, you know, and let them battle it out for which one gives me the best, response. And I take what I need from each one.

Mohamed Hamad: But I can’t really say definitively which one is better for it because they keep leapfrogging each other. At one point, one’s better than the other, and then, oh, they’ve got a new model and then next thing you know, this one’s so much better than the other one. So you play around with them. In saying that, before we wrap up, quick, fire question.

Mohamed Hamad: If someone is looking to improve their internal process around, performance marketing, the fact that there’s all of these new tools, where would be the first thing that you would recommend that let’s say they leverage these new tools to help them out, what, either scale up or, speed things up, improve their process.

Mohamed Hamad: Like, where would you start with that?

Sanjay Mayar: Yeah. No, that’s awesome. So the first thing I’d look at, which would be, even putting AI aside, is take a deep look at how the, you know, how the pipeline, is functioning to drive people new, users to the website and then what they’re doing, what you want them to do on the website.

Sanjay Mayar: So let’s say you’re running YouTube, ads, search ads, meta ads, TikTok ads, and you’re selling whatever it is, running shoes on your website. The first step would be to establish what that pipeline Looks like what the different KPIs or key metrics are for each of those stages.

Sanjay Mayar: And then do the work yourself to see where in that stage you need the most help. Maybe the flow from YouTube ads to brand searches to shopping ads, whatever is really strong.

Sanjay Mayar: Users are coming to your website may you know, at scale and cost effectively, but, oh, once they come to your site, you see a big drop off on product page views. So do you know, look at the data, do that analysis yourself and then use AI to substantiate what you’re seeing.

Sanjay Mayar: Right? So go in, don’t go in there blind, have your own, do your own due diligence, have your own narrative and then use AI as either a sounding board to validate what you see and confirm your diagnosis. Or perhaps it’ll see something that you don’t know can add something new to the conversation.

Sanjay Mayar: But establishing your KPIs first and then having your own diagnosis review of what the issue is and what you want to solve is really, really helpful. Don’t ask big broad questions, zero in on something you want salt specifically.

Sanjay Mayar: And then you can use AI to help you solve for that so you can show it your, you know, rather than look at all your ads and give you commentary on video length and CTR and audiences, which might be fine, you can focus AI on the problem that you want to solve, right?

Sanjay Mayar: That triage or that, that fire you want to put out. So have it look at your PDPs and you know, give it a critical analysis. Right. So is my page load time really bad? Am I am I not optimized for mobile? Do I need to add more reviews?

Sanjay Mayar: Whatever. The thing is, it can really focus your work. Then once you deploy those changes, you can then examine and review the deltas in those metrics to see if the action that you took resulted in the prescribed change.

Sanjay Mayar: If not, go back to the drawing board and use it there again. You don’t want it to write your whole script for you. You don’t want it to be a, you know, a substitute marketing director for you. You need to own that flow. But as a co pilot, as someone that can or something that can help you take specific action on a specific front.

Sanjay Mayar: It can be quite, it can be quite useful.

Mohamed Hamad: Absolutely. I think, yeah, specificity is is key to this whole workflow with AI. So everybody can connect with you on LinkedIn. I put the LinkedIn, link, your LinkedIn profile link over there.

Mohamed Hamad: But imajery also has a Free audit. You want to give us a little bit about that and what people can expect from a free audit from Imajery?

Sanjay Mayar: Yeah, absolutely. So, yeah, could I be on LinkedIn? We’d love to chat. So imajery audits, all the marketing activities that, you know, potential client, would ask us to look at.

Sanjay Mayar: So that would include GA4, your data ecosystem, looking at Google Ads, meta ads, tick, tock, email, we’ll check out your website. We’ll do a SWOT analysis, a good old fashioned, business school SWOT analysis.

Sanjay Mayar: And then we’ll come back to you with a list of really firm recommendations on what to do in the short, medium and long term. No obligation. Like most said, it’s free. And then, you know, prospective clients can use that, as they will.

Sanjay Mayar: So, you know, you’ll get to see how Imajery works, and how we make decisions, and, you know, what value we might add, if it aligns with what you want. Great. Let’s. Let’s work together. You know, if you’re not, if you don’t, for whatever reason, you’re not happy with it, you can shop it around, you can throw it to your internal team.

Sanjay Mayar: It’s no problem. We’re happy to, happy to have helped in any way. So, yeah, the free audits are dope. You should, you should, check it out.

Mohamed Hamad: Free audits are dope. You heard it from him. On that note, it was a pleasure talking to you. Lots of insights here. Thanks, everybody for joining us today and, we’ll catch you on the next one. Everybody join, connect with Sanjay on LinkedIn.

Mohamed Hamad: Check out his free audit at Imajery. The link is right there and it’ll be in the show notes. And, thank you everybody and see you next time.

Sanjay Mayar: Thanks, everyone. Thanks, Mo. Appreciate it. Good chatting with you.