On the latest episode of the Change Agents podcast, Alter Agents Chief Strategy Officer Devora Rogers is joined by Gaz Alushi, Global President of Measurement and Analytics at Whaler, a global creator marketing agency. Gaz walks through how he brought traditional media measurement discipline to the creator economy, the early-career lesson that convinced him honest, uncomfortable data always wins out, what’s next for creator best practices following Alter Agents’ joint Return on Creators research, and why he’s bullish on AI as a facilitation tool but skeptical of it as a replacement for human insight.

Unpack your title for us — what is Global President of Measurement and Analytics at Whalar? What is Whalar, and what do you actually do?

Whaler is a global creator marketing agency. The fundamental cornerstone of our business is connecting creators with brands. When people talk about creators, they often think of influencer marketing, and we’re very deliberate about differentiating the two. Influencers, I always say, are guns for hire — people who’ve built a social media presence specifically for the purpose of selling things. There’s no loyalty, no true community there.

Creators are different — they’re everyday people who are passionate about something specific. It could be skincare, home renovation, gardening, tracking down discontinued candles from Bath and Body Works across the country — really anything you can think of — and they build communities around those interests. What we do at Whaler is find the right creators for the right brands, to help them expand their reach and help this next phase of marketing and advertising move into the mainstream.

My role in that is what I like to call the chief nerd. I work with brands and creators to prove impact on the bottom line — did this drive sales, did this drive affinity, did this drive awareness — and to show how we can keep moving the needle, and just as importantly, what didn’t work as well as what did.

Tell us how you actually go about measuring that.

It’s funny — when I first joined Whaler in 2022, it was on the heels of what I jokingly call my tour of duty across analytics and social media. I’d just come from Snap, and before that a very long stint at Facebook, before they changed the name — it’ll always be Facebook to me. What I found, and this was the most mind-blowing thing for a lot of our brand partners, was that creators are media, and they should be measured the same way.

When I entered the creator economy in January 2022, EMV — earned media value — was the North Star metric. Everyone was playing this game of arbitrary math, black-box calculations. Any marketer, any brand leader, knows a black-box calculation is a recipe for disaster if you can’t validate and verify how you’re driving success. So it was quite literally bringing the industry back to basics — measuring awareness, consideration, actual call to action — and proving that the marketing funnel is alive and well, but that creators have contracted it.

Traditionally, a CPG product takes about four to six weeks from the start of a campaign to get someone from awareness to actually buying the product, whether online or off the shelf in a store. Creators have condensed that to about one to two weeks. It’s pretty wild how the calculus has changed. So going back to your original question — how do we do this — it’s quite literally taking the calculus that already existed in media measurement and reweighting those variables so creators can be incorporated and compared against every other activation. The crux of it is making sure creators aren’t measured in a silo.

Rather than invent something entirely new, you went back to the gold standard of media measurement and applied it to creators. Was that a hard sell internally?

It’s so funny, because it was mind-blowing for a lot of people. I remember one of the first projects I wanted to do was a media mix model to show that creators are part of the media mix, and our founder Neil was like, “that doesn’t sound sexy.” Well, depending on who you ask — MMM is pretty sexy, in my opinion. But I think everyone expects new math for this.

My experience at Facebook really conditioned me here. When I joined Facebook in 2012, the number one question was, “what’s the value of a fan? What’s the value of a like?” And the honest answer was kind of a womp-womp — nothing, really. It’s all about views, reach and frequency. Social media is media — that’s why it’s called social media, and creator content appears on social media. So it kind of turned into: let’s not overcomplicate this. There’s a lot of real complexity in this work, but we shouldn’t introduce complexity where it’s not required.

What is the insight in your career — a research study, an analysis — that made you and your team look at something through new eyes?

I’ve been doing this for over twenty years now, and it goes back to year one of my first job out of school, at a market research firm. The CEO there, Dale Lersh, was an amazing researcher, and I learned so much from her. We had done some concept testing for a tequila brand that really wanted to go after men aged 25 to 34 — that was their sweet spot. And everything in the data was just tepid. Not hated — just tepid, like it existed, sure. I panicked. This was my first year, and I thought, did I mess up the sampling? The fielding? The tabulation?

I brought her a preliminary report, and she asked, “did we only talk to men?” I said no, we’d captured women and other age groups too. She asked why that wasn’t in the report. So we went back and found that women actually liked the content better. She told me: sometimes the client’s brief is their fantasy, and you need to present them with reality.

That extended to significance and non-significance, too. Oftentimes people see there isn’t a statistically significant lift between groups, or between control and exposed, and assume that’s bad. She’d say, sometimes that’s actually really good news. In this case, the brand manager had already bought a multimillion-dollar campaign and mostly wanted reassurance. But when we presented to the senior leader, Dale told them they were disenfranchising an audience that actually liked the product. There was a real moment of reckoning in that room, and in the end, they changed their strategy — they made the campaign more inclusive instead of narrowing it to the original target.

Ever since then, I’ve looked back at my career and the number of leaders I’ve brought negative results to — results that contradicted something they believed internally — and I’ve never once had a leader get angry. I’ve never had a budget pulled. I’ve never seen anyone lose their job over it. There’s always been gratitude — great, now we know, now we can pivot and start doing the right thing. That’s the core insight: the truth will do it for you a hundred percent of the time. When I see people trying to game final results to make the client’s outcome look better, I always think that’s a short-term win at best, because the reality is still there in the data — that the target audience doesn’t like it, or there’s a better placement, or, worst case, you should stop selling this product altogether and save yourself the production costs.

Can you share anything about what you’re working on right now?

Right now is exciting because we did work with Alter Agents, combined with research we’d done with Kantar and Nielsen, and we called it Return on Creators. It showcased how a lot of brands are thinking about creator activations the wrong way. The work we’re doing right now is building out best practices and playbooks across different subverticals — what works great for beauty, what works great for travel, what works great for fintech.

That helps us bring strategies to life when we brief creators — here are the tactics to deploy, make sure the brand is present by this second mark, make sure you’re actually using the product, and so on. What that does is actually free up the creator’s creative voice — we let them do what they do best. So the work right now is taking what we did with Alter Agents and scaling it globally, applying empirically validated learnings across thousands of campaigns, scoring all the content so we can say, that was a cute idea but don’t ever do that again, it’ll torpedo your brand — or, you need to crank this up by a factor of ten if you want to meaningfully increase your profit margin.

Are brands still skeptical about the impact of creators, or are people on board now?

It’s no longer a question of whether creators work — I have so much déjà vu here, because when I started at Facebook, it was “no one buys anything because of social media,” and now it would be insane for a brand not to have a social media presence. Nobody’s saying, “I’m not doing Instagram, I’m not doing TikTok, I’m not doing YouTube.”

People are there with creators now. The real challenge is preconceived notions — people picture “those people who do dances on TikTok” and think that’s not right for their hair care brand, for instance. And they’re right, that’s not what we’d use for that. So the conversation has shifted from why to how.

Are there still categories or behaviors where people resist buying online, even now?

It’s interesting — one of the things I enjoy geeking out on is the micro insights. Nobody’s going to buy laundry detergent online, people used to say — but pet food, pet care, household care products, that’s a different story. Then you look at grocery delivery, and there are still some people who say, no, I’ll buy my own meat and produce, but you can bring me the frozen or refrigerated foods through automated delivery. So there are still some barriers, but by and large, it’s not a foreign concept anymore to want something brought to your home immediately and to order it online.

You’ve been really effective at getting insights a seat at the corporate table. What’s your approach, and what would you recommend to other researchers trying to do the same?

It’s twofold. Our co-CEO, Joe Cronk, calls me the chief accountability officer, which I think is such a smart way of framing the function we play — we’re the accountability folks, making sure you actually did what you said you were going to do.

What I’ve found is that instead of going in and puffing your chest about all the great things you have for a leader, what they react to best is: here are some failures other people have made that I’m going to share with you, here’s what we don’t know, and here’s how we’re trying to work around it. Executives love hearing that somebody else paid to learn a mistake — what didn’t work, and why. I often find that has more gravitas than simply presenting what did work.

The second part is being upfront about what we don’t know. This is part of why I love multimodal research so much — when you focus on a single channel or a single data source, it doesn’t capture everything. It’s important to remind stakeholders that the insights they’re getting right now might be siloed, might be biased, and here are the recommendations we have to make sure they’re wrapping their arms around everything they can know — and here’s the reality of what we’re making educated guesses on that we may never fully know. That’s what earns you a seat at the table, because instead of it being transactional, a one-and-done delivery, you get invited into the ongoing dialogue — measurement and analytics becomes part of the conversation, so we can flag watch-outs and dos and don’ts as things move along.

It sounds like part of being effective is educating executives and stakeholders on what’s actually knowable and what isn’t. Is that fair?

Yes, that’s so critically important. It goes back to a truism you learn in college: a little bit of good data is better than a lot of bad data. I think the industry, especially executives who aren’t research or analytics specialists, tend to think, well, we have big data, we have a massive data source — that’s lovely, but I need to explain why that’s limited and how you need to augment it.

A good example: I was talking to someone at CES a couple of years ago who worked at a major media company. They’d built this huge AI infrastructure analyzing all the content they produced in-house to identify best and worst practices, but they weren’t getting the insights they needed. I told them, that’s like analyzing someone’s existing diet and telling them what to eat more or less of — you need to vary the diet. You need to measure things you haven’t measured yet — are you incorporating content you didn’t produce in-house? Competitor content? Just because you have a massive data source doesn’t mean it’s good data. I definitely offended him when I said that.

We went through a hype cycle with big data — and it sounds like we may be seeing something similar with AI. How do you see the AI hype cycle playing out in research and analytics?

AI is fascinating because it can only capture what it has access to. It was in the news that Instagram moved away from end-to-end encryption for direct messages, and everyone was talking about the privacy angle — but I don’t think it’s about targeting a specific individual. They need to understand user behavior outside of just the feed, the explore tab, the things people are clicking on. They also need to see what people are talking about — when someone shares a post, who are they sending it to, and why. I think you’re seeing this across the industry, with a lot of companies realizing their massive data sets are actually woefully incomplete.

So I always say: yes, AI is going to do a really good job of tabulating the numbers in a given data set, but it’s unlikely to draw a deep human insight out of it. Think about your own online behavior — most people are pretty voyeuristic. You’re not commenting on every post you come across, not engaging with every link you click. So these digital traces miss a lot — and you can’t capture what’s happening in person either. There’s a real attempt across the industry to capture that analog behavior — wearables with cameras, geolocation and NFC in retail — but it’s resource-intensive, labor-intensive, and it still doesn’t capture the true insight all the time. So I think AI is amazing as a facilitation tool. I have real hesitations about it as an absolute replacement tool — we keep seeing headlines about entire pieces of infrastructure failing for 18 hours because they were left entirely in the hands of an AI agent.

Last question — of all the data and methodologies you’ve worked with, what’s your personal favorite?

Multimodal methodologies, simply because they cast the widest net. It’s the opportunity to get as close to a 360-degree view of human behavior as you can, versus just analyzing synthesized big data, or just running a one-off control-versus-exposed experiment. Control-exposed will always be the gold standard for a reason, but when you incorporate that into a multimodal approach, you get much deeper insight than you would from any single methodology. On the flip side — you basically said it yourself — when’s the last time you heard someone push hard for “big data”? That was the buzzword for years, and I haven’t heard it in going on two years now.