How We Increased Our Client’s Revenue Using AEO & GEO Strategy

We kept seeing the same pattern with clients. Rankings held steady, sometimes they even climbed, yet traffic slid and revenue followed it down. Nothing in the old playbook explained it. The pages were winning the search and still losing the visitor.

The cause sat above the blue links. AI summaries and chatbots were answering the question before anyone reached the site. So we rebuilt how we approach visibility, around answer engine optimization and generative engine optimization, and the revenue picture started to recover. This is the strategy we run, the order we run it in, and the evidence behind why it works. No invented numbers, just the method and the research that holds it up.

The problem we kept running into

For two decades the deal was simple. Rank high, earn the click, convert the visitor. That chain started breaking at the first link, and the break was structural, not seasonal.

Gartner expects traditional search engine volume to fall 25 percent by 2026 as people move questions to AI chatbots and virtual agents [2]. We watched the same behavior inside Google. Pew Research Center found that when an AI summary appears, users click a traditional result only 8 percent of the time, against 15 percent when no summary shows, and they click a link inside the summary just 1 percent of the time [3]. Our clients’ answers were feeding the summaries while almost nobody reached their pages.

There was a quieter upside hiding in the same data, and it shaped our whole approach. The clicks that did come through after an AI summary carried higher intent. Someone who reads the answer, then still chooses to visit, is closer to a decision. So we changed the target: fewer raw sessions, more qualified ones, plus a brand that shows up by name inside the answer where buyers first form an opinion. That mix is what we found protects revenue even as click counts drop.

What we realized: AEO and GEO are not the same job

Early on we treated the two terms as synonyms, and the work suffered for it. They solve different problems. The way we now explain it to clients: SEO gets you into the room, AEO gets you read aloud, GEO gets you believed. Our GEO vs AEO vs SEO comparison goes deeper, but the table below is the working version we use.

Read it left to right and the dependency is clear. AEO earns mention-level visibility. GEO earns reasoning-level authority. Neither works if the SEO floor underneath is shaky, because most answer engines still pull from pages they already find and trust. We stopped trying to skip steps once we accepted that.

The exact sequence we run for clients

Sequence mattered more than effort. The clients who saw revenue move were the ones who let us fix the order rather than chase the shiny part first. Here is the order we follow every time.

Skip a step and the next one underperforms. AEO formatting on a page no engine trusts is wasted polish. GEO evidence on a page with no structure rarely gets extracted cleanly. We learned that the layers reinforce each other only in order.

What moved the needle for us

We did not guess our way to this. The strongest validation came from a controlled study by researchers at Princeton, IIT Delhi, Georgia Tech, and the Allen Institute, which tested content changes across 10,000 queries to see which ones earned citations inside generative engines. A handful produced real, measurable lift, and they matched what we were seeing in client work.

The same study validated the effect on Perplexity, a live engine, with gains up to 37 percent [1]. In practice, we now aim for one specific data point every 150 to 200 words, each tied to its source, rather than a single stat near the top.

One finding kept us honest. The effect of each tactic shifted by topic. A move that lifted visibility in one client’s domain did less in another. So we test rather than assume, and we treat any tactic list as a starting hypothesis. The hands-on execution of this is what our generative engine optimization services are built around.

In day-to-day work this looks ordinary, which is the point. Before a page goes live, we check it like an editor and a skeptic at once. Does the opening answer the question in plain words? Does each claim carry a source a reader could verify? Would an AI quoting this page misrepresent anything? Pages that pass those three checks are the ones we see surface in answers. Pages that fail them sit quiet, no matter how many keywords they hold.

The mistakes we made early, and stopped

Knowing what to drop saved as much budget as knowing what to do. A few habits carried over from classic SEO produced nothing in AI search, and some actively hurt. We learned several of these the hard way.

The pattern is hard to miss. Every failed tactic tries to game volume. Every winning one adds verifiable substance. That single distinction predicts client outcomes better than any tool we have used.

There is a mindset shift buried in all of this. The old SEO instinct was to produce more, faster, and hope the algorithm rewarded the volume. AI search punishes that instinct. It rewards the page that earns trust, not the one that shows up most often. Once a client internalized that, the work got simpler. Fewer pages, more care, every claim backed. That is a harder brief to fill than a content calendar full of thin posts, but it is the one that actually moved revenue.

How we build entity authority for a client

Generative engines lean on consensus. They trust facts that appear consistently across many trustworthy sources, then synthesize from there. So our work is partly off-page, not just page-level. We are teaching the wider web a stable story about who the client is.

This is why a strong content marketing program now doubles as AI visibility work. Every well sourced article, accurate listing, and credible mention adds a vote to the consensus the engines read. The clients who show up by name in answers are the ones who built that consensus on purpose.

Structured data supports all of it. FAQ and HowTo schema help engines parse answers cleanly, and Organization and Author markup reinforce who stands behind the content. Schema is not a trick. It removes ambiguity, so a machine reads the page the way it was meant. We keep the markup honest and matched to what is on the page, because mismatched schema does more harm than none.

Tuning for each engine, because they do not behave alike

A single optimization pass stopped covering every surface. Each engine weighs sources differently, so the same page can win in one and vanish in another. We do not build a separate strategy per platform, but we tune the edges for the engines that matter to a client’s buyers.

The common thread runs through all five. Each one wants clear, credible, well evidenced content, and each reads the wider web to decide who to trust. We tune the edges, then let the base work carry across every engine.

How we measure what Search Console cannot show

Honest answer first. The tooling for AI search is immature, and standard analytics will not show you most of it. There is no native report for how often ChatGPT cited a client. Waiting for perfect data means waiting too long. So we track proxies, and we track them consistently.

We pick a small set, set a baseline, and review monthly. The tooling will improve. The discipline of measuring proxies now is what lets us prove the work later. More of our thinking sits in our guide to AI search and AEO strategies.

How the visibility turned into revenue

Visibility only matters if it reaches money. For our clients the connection was real, but it ran through a different path than the old click-to-convert funnel. Here is the mechanism, without inflated promises.

First, the answer became the new shelf. When a buyer asks an AI tool to compare options or recommend a provider, the brands named in that answer enter the consideration set. The absent ones never get a shot. Being cited is the modern version of page one, and it happens at the exact moment intent forms.

Second, the clicks that survived converted better. Fewer people reached the site, but the ones who did had pre-qualified themselves against the AI answer. A smaller, warmer stream beat a larger, colder one on revenue per visit.

Third, consensus compounded. Each accurate citation made the next one more likely, which slowly turned a client into the default answer for a topic. That position is hard for competitors to dislodge once it sets, and it keeps producing pipeline without paid spend behind every impression.

The timeline is worth setting expectations on, because it is not instant. The AEO wins arrive first. Snippet and overview placements can shift within weeks once the structure is right. The GEO side moves slower, since consensus and citations build over months as the wider web catches up to the new signals. We tell clients to expect early movement in answer ownership, then a longer climb in citations and the branded demand that follows. Patience here is not a soft skill, it is part of the strategy.

There is a reporting trap inside this. Old dashboards undercount the work, because a mention that shapes a shortlist may never register as a session. A prospect can read a client’s name in an AI answer, sit with it for a week, then arrive through branded search. The credit lands in the wrong column unless you widen what you measure. Judge AEO and GEO on last-click traffic alone, and you will call it a failure right as it starts working.

What we would tell a team starting now

There is no single right way to run this. It turns on bandwidth and how fast the category is moving against you. A small marketing team can often run the AEO layer alone, since structured answer engine optimization is mostly discipline applied consistently. The GEO and consensus work is heavier, since it spans content, PR, data accuracy, and constant measurement.

Here is the test we give clients. If competitors already own the AI answers in your space, speed matters more than control, and a team that has run this before saves months of trial and error. If you are early in a quiet category, building the muscle in-house can be the better long game. Either way, ask any partner for specifics, not promises. Can they show how they track AI citations? Will they refuse to guarantee placement? Honest answers to those two questions tell you most of what you need to know.

Whatever route you pick, start before you feel ready. The brands settling into AI answers now are setting a baseline that later entrants will spend a long time trying to match. Waiting for the tooling to mature, or for a competitor to prove it first, just hands them the head start.

Where this is heading in 2026

A few shifts are worth planning around now rather than reacting to later.

None of this replaces the fundamentals. Useful, well sourced, clearly structured content still wins, the same way it always has. The difference is that the audience reading it now includes machines that summarize before a human ever arrives. We build for both, and the strategy holds up no matter which engine grows next.

If there is one idea to carry out of our experience, it is that AEO and GEO reward substance over scale. The clients winning AI answers are not the loudest or the most prolific. They are the ones whose pages a machine can read, quote, and trust without hesitation. That bar is higher than the old SEO checklist, and harder to fake. Which is exactly why getting it right now, before a category fills up, pays off later.