🎧 Listen to this episode
Gus Pelogia has been doing SEO for nearly 13 years, and he has done it in most of the shapes the job comes in — content, then years of link building, then agency work in Ireland, and now senior SEO product manager at Indeed. What makes this episode of The Unscripted SEO Interview worth your time is that the in-house-at-scale version of the job turns out to be almost nothing like the version most SEOs practice, and Gus is unusually blunt about what actually changed.
“They absolutely do not care”
The agency years taught him the lesson that later made him useful in a product org: the person paying you is not interested in your craft.
“They don’t care about the canonical tags out of place or how fast this page is loading. They want to know are we selling more or are we getting the leads that we want? They absolutely do not care. It’s like there’s a plumbing problem in your house. All I want is my house not to flood. If you’re using a golden pipe or a plastic pipe — whichever is the easiest that will make a smaller mess in my house. That’s what they want.” — Gus Pelogia
He’s careful to say this isn’t about hiding anything. It’s about knowing which level of detail actually satisfies the person asking.
What an SEO product manager actually does all day
Gus didn’t seek the product role — he joined Indeed as an SEO content manager, and when the person doing product work left, his managers handed it to him. What they’d noticed was that he was good at explaining things simply and selling the benefit of an idea.
The rhythm of the job now looks like this: at the start of a quarter he picks three or four big ideas — internal link plays, new page types — and each one starts as a written document, not a ticket. Three pages, ten pages, whatever the idea needs. Then he walks it around.
“I’m promising in this document that engineering can do this and they might take somewhere between one and three months… So if you tell me now that this is not possible or that what I’m saying is wrong, now is the time for you to say it — because you’re going to be held responsible later for executing this.” — Gus Pelogia
That does two jobs at once. It surfaces the objections while they’re still cheap, and it gets every champion — SEO, UX, engineering — invested before the idea is ever announced. By the time he stands up and says “we’re building an internal link module,” the leaders have already seen it and already have their own reasons for liking it.
The question that knocked him flat
About six months into the product role, Gus presented to a company leader: here’s what we shipped, and traffic has doubled since I started. Fifteen minutes of good news. Then she asked him one question.
“How do you know any of these things are happening because of the things we did? Maybe traffic was going up anyway. Maybe the demand for these jobs is up. I have no proof that the work you did is actually what is driving this growth.”
His description of that moment is the most honest thing in the interview: “it’s just like the floor opens and I just fell down.” He’d walked in proud and was on the ground a few hours later, wondering whether he was about to lose his engineers.
What he did next is the actual takeaway. He found the teams inside Indeed that already had SEO testing capability, started working with them, and stopped making contribution arguments altogether.
“Then I could go back and say, this was 3% of our traffic increase. This was 5% of our conversions increase. And then it was a completely different conversation.” — Gus Pelogia
Now the pitch for the next initiative doesn’t contain the word “crawl” at all:
“With internal linking, we can get a growth up to 10% incremental. You don’t need to keep explaining why links are important and how Google views them and the crawling that happens between pages and the anchors — a lot of people don’t care about any of this. I’ve done this, gave us a 10% lift. I have another similar play to do. 10% lift? Yes, I do want a 10% lift. How much is it going to cost?” — Gus Pelogia
He also runs the MVP deliberately small — one country first, purely so the number exists — then expands. And when he skipped that step, it came back to bite him: nobody remembers the test you ran six months ago.
The higher the audience, the simpler the artifact
Gus credits his manager, Sam Brennan, with teaching him to compress. He’d write a ten-page PRD; Sam would turn it into a one-pager. So now he does both — the full research exists and he can defend any question from it, but what travels upward is short.
“They are looking at the SEO thing, the PPC thing, the branding, the radio. They’re just looking at too many things… The higher the goal, the simpler you have to be. They don’t live in your world.” — Gus Pelogia
Where AI search actually hurts — and where it doesn’t
Gus is refreshingly unwilling to pretend he has this figured out: “I don’t really have a proper answer yet.” His working read is that overall traffic declines and the keyword-and-blue-links model that told us how big demand was stops being reliable. But he doesn’t think a marketplace like Indeed is structurally threatened — you’re not going to apply for a job from inside a chatbot. What changes is acquisition, not the destination.
And the clicks that do survive are better ones:
“Once people reach your website… it’s such a much bigger effort to do a click after you did some AI search, after you found some answers. If you still want to click through to a source, I think you’re much more ready to perform an action than before.” — Gus Pelogia
The sharper problem is measurement, and he has a personal example. He booked a holiday to Málaga through ChatGPT. His criteria were genuinely hard to search — under two hours’ flight, direct from Dublin, in March, with less rainfall than Dublin — and traditional search meant comparing rainfall across cities and airline routes by hand. Two or three interactions with a chatbot replaced what would have been traffic to ten or twenty pages. Then he went straight to the airline’s site and booked.
“All of this happened in a search environment. The credit was just direct… Our attribution is fucked, but we need to be part of this one way or another to understand why it’s happening and how we can be part of it.” — Gus Pelogia
Demand gets created somewhere you’re not looking
Both of us landed on the same conclusion from different directions: the demand SEOs harvest increasingly originates off-platform. Gus pointed at agencies like Rise at 7 deliberately generating demand through TikTok and then building the page that catches it — and at a retail brand whose Christmas commercial featured a piano, after which a specialist e-commerce site saw piano sales climb without ever knowing why.
Jeremy’s version is a cautionary one: a brand that spent roughly $100,000 on an influencer campaign for pink pants while holding about ten pairs in stock. They sold out, went out of stock on a suddenly-hot keyword, dropped out of the results, and effectively funded a competitor’s quarter. Nobody had told the person running the store.
“This is not even SEO, right? This is just having the business sense… You create a page that is optimized for that search and maybe you start building that page a while before, and you find ways to get links to it to make sure it’s ranking high. So you get to cash in on this demand that you’re about to create.” — Gus Pelogia
When the SEO win is the wrong call
One of the products Gus works on is Indeed’s career guide, which has a feedback mechanism job seekers actually use. That feedback, plus UX interviews run against little more than a Figma file, has killed features that would have been good for search.
“We’d think, oh, this is a great feature, we’re going to do this for SEO, it’s going to have a great impact. And people were like, why is this in here?… There is no SEO reason alone that is going to put this in this place.” — Gus Pelogia
He’d just had the same experience as a user, trying to find out whether a hotel gym opened on weekends and wading through walls of text and obvious internal-link anchors to get there. “Someone has been SEOing this website.” The balance he’s after: rank for it, and then don’t punish the person who arrives.
His one tactic: let ChatGPT write throwaway Python
Asked for something actionable, Gus gave the least glamorous and most repeatable answer — he’s spent the last six months having ChatGPT write small Python programs he runs in Google Colab, despite not knowing how to code.
His worked example: he needed to know when a set of pages started losing traffic, and every page peaked and dipped in a different month, which makes the usual tool workflow miserable. Two or three prompts produced a script that took a spreadsheet of URL, month and sessions and returned each URL’s best and worst month. From there he could go look at what actually happened — and for a lot of those pages, nothing had: no ranking change at all, just AI Overviews and other SERP features crowding the space. As he put it, nobody’s winning.
“Anything that is just a boring task where you’re checking things, you can ask ChatGPT to run a program for you… I don’t know how to code pretty much anything. I click on run, it asks me to upload the sheet and gives me the results. If you have a boring task to do, maybe ChatGPT and Google Colab can do the boring part for you.” — Gus Pelogia
That reframes the anxiety nicely. The value of an SEO was never that we could sit and reconcile CSV exports by hand — it was knowing which question was worth asking.
Connect with Gus
Gus Pelogia is a senior SEO product manager at Indeed, based in Dublin. He’s mostly on LinkedIn, a little on Bluesky, and writes sporadically at guspelogia.com. He’s a regular at Brighton SEO — and made a good case for going in person, including a Lazarina Stoi talk on machine learning for SEO that he only half-followed in the room, went back to the recording for, and would otherwise never have known was a thing he could do.
Listen to the full episode on Castos. For more conversations like this one, explore the Unscripted SEO archive, or try the keyword clustering tools at SEO Arcade.
