What We Learned Running SEO, AEO and GEO for One Client

For nine months we ran SEO, AEO and GEO in parallel for a single client. Same brand, same budget pool, same content team. Three disciplines, one company, three completely separate scoreboards.

We did it deliberately. Every agency deck in 2026 now has a slide that says "we do SEO, AEO and GEO" as if they are three products you can buy off a shelf. Almost nobody can tell you, from real work on a real account, where the three actually overlap, where they diverge, what the incremental cost of each one is, or which you should run first if you could only afford one. We wanted to answer those questions with evidence instead of a slide.

So this is the field report. The client was a mid-sized Indian D2C brand in a considered-purchase category - the kind of product people research before buying, across Google and, increasingly, inside AI assistants. We anonymise the brand here, but the numbers, the splits and the lessons are real. If you are deciding whether to spread your budget across all three disciplines or concentrate it, this is the piece I wish someone had handed me before we started.

If you want the textbook definitions of how the three disciplines differ before reading our experience, our SEO vs AEO vs GEO explainer is the reference page. This post assumes you broadly know what each one is and want to know what happens when you run them together.

The short answer

Running SEO, AEO and GEO together for one client taught us five things. First, roughly 70 percent of the work is shared - one well-researched, well-structured content foundation serves all three, and treating them as three separate content pipelines is how budgets triple and work starts cannibalising itself. Second, AEO is mostly an extension of good SEO - schema, answer formatting and question research applied to pages you are already building - so its incremental cost is small. Third, GEO is a genuinely different game - it depends on off-site entity signals and third-party presence you do not fully control, which makes it the slowest and most expensive of the three. Fourth, each surface needs its own scoreboard; rankings live in Search Console, answer-box presence does not, and AI citations live nowhere unless you build the tracking yourself. Fifth, if we could run only one, we would still run SEO first - not because it matters most, but because the other two stand on top of it. The whole programme cost about 1.4x SEO-alone, not 3x.

Why we ran all three instead of picking one

The honest reason we ran all three was that the client's buyers had visibly split their research across surfaces. We could see it in the assisted-conversion paths, in the sales team's call notes ("they said they'd seen us mentioned by ChatGPT"), and in the growing gap between brands that rank on Google but are invisible in AI answers. A meaningful slice of the category's high-intent research had moved into answer boxes and AI assistants, and the brand was getting clicks on Google while being entirely absent from the AI surface where a competitor was getting named.

That is the situation that justifies running all three: not because three acronyms sound thorough, but because your buyers are genuinely making decisions across three different surfaces, and you are present on only one of them. If your buyers all still start and finish on a blue Google link, you do not need GEO yet - you need better SEO. We checked that assumption first, with data, before committing budget. The single most common mistake we see is brands buying a three-discipline programme on fear of missing out rather than on evidence that their buyers have moved.

What actually overlapped (and what didn't)

The most useful thing the nine months produced was a clear map of where effort was shared versus discipline-specific. Here is the split as we actually experienced it.

The shared foundation was the same thing it has always been: genuinely useful content, researched properly, structured clearly, published by an identifiable expert. The pages that won featured snippets were the same pages that ranked. The pages that got cited by ChatGPT and Perplexity were, almost without exception, the same pages again. There was no separate "AI content" that worked while the SEO content didn't. There was good content, and good content paid out across all three surfaces. That alone is the most reassuring finding for anyone worried they need to rebuild everything: you mostly don't.

Where the disciplines genuinely diverged was the thin layer on top:

• SEO-specific work was the classic stack: technical health, crawl and index control, internal linking, and earning links. None of this directly helps an AI assistant cite you, but all of it is load-bearing for the foundation everything else stands on.
• AEO-specific work was formatting discipline: tight, liftable answer passages near the top of each page, the schema changes that actually move AI answers, question-led headings drawn from People Also Ask, and structuring content so Google could extract a clean answer. It is cheap because it rides on content you are already producing.
• GEO-specific work was the genuinely new and genuinely expensive part: making sure the brand's entity was consistent everywhere it appeared, building presence on third-party sources the models trust, publishing an llms.txt file and keeping content machine-readable, and continuously monitoring whether AI assistants actually named us. This is the layer most teams underestimate.

Where the budget actually went

The number everyone wants is the cost multiplier. Running all three did not cost three times running one. It cost roughly 1.4 times what an SEO-only programme of the same ambition would have cost, because the expensive part - the content foundation - was shared.

Here is the rough allocation of incremental effort across the nine months, expressed as share of total programme hours.

Two things in that chart surprised even us. AEO was astonishingly cheap - barely a rounding error on top of the content we were already producing, because it was mostly disciplined formatting and schema. And GEO cost more than AEO despite, at least early on, returning less measurable value, because its work is structural, off-site, and never really "done." You build entity consistency once and then maintain it forever. That maintenance tail is the part nobody budgets for.

The hardest part was measurement, not execution

If you take one operational lesson from this, take this one: the execution was the easy part. The measurement nearly broke us.

SEO has a mature scoreboard. Rankings, impressions, clicks and conversions all sit in Search Console and analytics, and everyone on the client side knows how to read them. AEO is half-measurable - you can track featured-snippet and People Also Ask presence and AI Overview appearances, but the click behaviour around zero-click answers is murky, and you are often winning visibility you cannot fully count.

GEO had no scoreboard at all until we built one. There is no Search Console for "how often does ChatGPT name you." We had to construct the measurement from scratch: a fixed set of category prompts, run repeatedly across ChatGPT, Perplexity, Gemini and Claude, with the brand mentions extracted and rolled up into a citation rate and a share of voice. This is the discipline of AI citation tracking, and it is the single biggest gap between agencies that say they do GEO and agencies that actually do.

The reason this matters is political as much as technical. Three months in, the client looked at Search Console, saw GEO had not moved their rankings, and asked - reasonably - whether the AI work was a waste. It was not. It just lived on a dashboard Search Console will never show. The brands that conclude "AI search isn't doing anything" are almost always looking at the wrong scoreboard. Build the AI-surface scoreboard before you start, capture a baseline, and you can show movement that the rankings dashboard structurally cannot.

What each discipline actually moved

By the end of the engagement, here is what each layer had contributed, in plain terms.

SEO moved qualified traffic and revenue. This remained, by a wide margin, the largest source of measurable business outcomes. The blue links still did the heavy commercial lifting, and anyone telling you classic search is dead is not looking at a real account's revenue attribution. Our take on whether SEO is dead is anchored in numbers like these.

AEO moved visibility and defended against zero-click erosion. As more queries resolved inside the answer box, AEO work meant the brand was the thing being quoted rather than the thing being skipped. It did not produce a clean, separately attributable revenue line, but it defended share of attention on queries that were increasingly resolving without a click - and on a few high-intent queries, being the answer-box source measurably lifted branded search afterward.

GEO moved being in the consideration set. This was the slowest and the most strategic. Over the nine months the brand went from never being named by AI assistants in its category to being one of a handful of names that came up consistently. That did not show up as a traffic spike. It showed up in sales conversations - prospects arriving already aware, already shortlisting, because an AI assistant had named the brand when they asked for options. For a considered-purchase category, getting into that AI-generated shortlist is worth more than its traffic numbers suggest, which is exactly why it is so hard to justify on a traffic dashboard.

If we could only run one

People ask this constantly, so here is the honest answer from the work rather than from a positioning slide.

We would run SEO first. Not because it is more important than AEO or GEO in some abstract sense - the future clearly belongs to a blend - but because of dependency. AEO and GEO both feed on the foundation SEO produces. The well-researched, well-structured, authoritative content that earns rankings is the same content that earns snippets and citations. You cannot do good AEO or GEO on a weak content base; you can do good SEO and get a large share of the AEO benefit almost for free.

If you have done SEO well and want the next increment, add AEO - it is cheap, it rides on what you have, and it defends you against the zero-click shift that is eroding everyone's click-through. Add GEO when you have evidence your buyers research inside AI assistants, when your category is considered enough that being in the AI shortlist matters, and when you can fund the off-site, ongoing entity work it actually requires. For most brands, that sequence - SEO, then AEO, then GEO - is the right order of operations, and it is also, conveniently, the order of cost.

The brands that should invert this are the ones whose buyers have already moved: certain B2B and high-consideration categories where the research genuinely happens in an AI assistant first. If that is you, GEO earns a bigger early seat at the table. But that should be an evidence-based exception, not the default - and you can find the current adoption numbers in our AI search statistics for 2026 to check where your category actually sits.

The mistakes we made so you don't have to

A few things we got wrong in the first three months, in case they save you the same lessons:

• We initially staffed it as three workstreams. That was a mistake. It created duplicate content briefs, conflicting recommendations, and a budget that ballooned toward the 3x trap. Collapsing it into one content team with three optimisation lenses cut cost and improved consistency immediately.
• We measured GEO too late. We started the AI-citation tracking a month into the engagement, which meant we had no clean baseline and spent weeks arguing about whether things had improved. Build the scoreboard before you touch anything.
• We over-invested in GEO tactics that did not stick. Some of the early off-site work chased volume rather than the specific sources LLMs actually trust. Presence on a source the models ignore is wasted effort. Concentrate GEO work where the engines genuinely retrieve.
• We under-communicated the surface split to the client. Because the three surfaces have three scoreboards, we needed a single combined view that showed all three together, or every monthly review turned into a debate about which dashboard was "the real one." A unified report fixed it.

If you are weighing whether to run these disciplines together or want a partner who has actually done it on a live account rather than in a pitch deck, that is precisely the work our AI SEO services and core SEO team do every day - and our broader thinking on choosing the right partner covers what to look for.

The bottom line

Running SEO, AEO and GEO together for one client was worth it - but not for the reason the three-acronym slide implies. It was worth it because the disciplines share a foundation, so the marginal cost of covering all three surfaces was modest, and because the client's buyers really had spread across all three. The work was 70 percent shared content, with three thin layers on top: SEO for the fundamentals and traffic, AEO as a cheap defence against the zero-click shift, GEO as a slow, strategic play for the AI consideration set.

The discipline that will catch most teams out is not execution; it is measurement. Three surfaces need three scoreboards, and two of those scoreboards do not exist until you build them. Build them first. Then run one content engine, not three. And if you can only afford one discipline this quarter, run SEO - because everything else is standing on it.

Want a candid read on whether your buyers have actually moved to AI search, and which of the three disciplines deserves your next rupee of budget? Talk to us - we will tell you the honest answer, even when it is "you don't need GEO yet."