How Long AEO Takes to Work: What Our Own 52 Days Show

Every agency has a timeline and none has a measurement behind it. Ninety days is the usual number. We decided to publish ours instead, including the part that does not flatter us: three readings, 52 days apart at the widest, on the same 8 buyer questions, and our score has not moved off zero.

The short version

We hold three dated readings on 8 Dubai buyer questions: 21 July, 7 September and 12 September 2026. Ivanooo scored 0% named and 0% cited on all three. Across the wider 62-query panel (16-18 September 2026) our Share of Recommendation reads 0.0, with 59 of 62 questions returning us absent, 2 citing us and 1 naming us.

That panel also shows what the clock is running against. The engine drew 301 citations from 178 different domains, 131 of them cited exactly once. The four platform surfaces hold 18% of those citations, the largest single source holds 6%, and our own domain holds 1.7%. In the same window the agencies that were winning turned over completely, with one falling from a 50% citation share to 0%. So the honest timeline answer has two parts: position can change in under 60 days, and earned coverage takes considerably longer than that.

Why the usual 90-day number is not a measurement

It is borrowed from SEO

Ninety days is the familiar SEO ramp. Generated answers do not behave as ranked lists do: selection happens at query time from a changing pool, which is why we saw wholesale turnover inside 52 days rather than gradual movement.

It conflates two different clocks

Getting cited once is fast and unstable. Being named consistently is slow and durable. A vendor quoting a single number is averaging two things that behave differently.

Nobody publishes the failures

A timeline built only from wins is a survivorship number. We publish our zero precisely because it is the part that would otherwise be invisible.

The two clocks

	Citation clock	Recommendation clock
What moves it	Passage structure, freshness, evidence	Distinctiveness, entity strength, corroboration
Speed of first movement	Weeks	Months
Stability once earned	Low, re-sampled constantly	Higher, but not guaranteed
Our own status	0% across 3 readings	0.0 Share of Recommendation
What a vendor can promise	Effort and cadence	Neither, honestly

The Princeton GEO research, Aggarwal et al., measured passage-level changes moving visibility by up to 40% under controlled conditions, which is evidence about the citation clock, not the recommendation clock. Controlled benchmark effects and live contested markets are different things, and the gap between them is where most timeline promises live.

What to expect, stated honestly

1. Days 1 to 7: your baseline. You learn which questions return you and which do not. This is the only fast, certain deliverable in the category.
2. Weeks 2 to 8: passage work lands. Answer blocks, quantified claims and inline sources change whether your pages are retrievable. Movement here is possible but noisy.
3. Day 60: the first honest read. Two dated measurements let you separate a real change from a re-sample. Anything judged before this is guesswork.
4. Months 3 to 6: coverage compounds or it does not. A cluster that answers a subject completely keeps qualifying for new answers as the engine re-draws them. One page does not.
5. Continuously: position churns regardless. Our data shows leaders losing everything in 52 days without doing anything wrong, so plan for maintenance rather than arrival.

What we would tell a buyer who needs results this quarter

That AEO is the wrong instrument for a 90-day revenue gap. It compounds, which means it rewards the quarter after next. If the pipeline needs filling now, the honest advice is outbound and placement, and to fund AI Search Visibility as the asset that makes next year cheaper. We say this while selling the discipline, which is the test of whether a timeline claim is honest.

Questions buyers ask

Has anything moved for you in 52 days? Our score has not. What moved was the field around us, which is itself the finding: the churn data shows a 100% turnover among leaders in the same window.

Then why continue? Because coverage is the durable half. Position churns; a site that answers the subject completely keeps re-qualifying. Topic Authority is built from coverage held over time, and time is the input nobody can shortcut.

Is 8 questions too small a panel to judge? For a timeline claim, yes, and we say so. It is large enough to disprove a 30-day guarantee and too small to prove a general law.

What would make you faster? Off-domain placement, on our own diagnosis. More than half the answers we want are assembled from platforms, so writing more pages on our own site cannot reach them.

Should I believe a case study showing 30-day results? Ask for the second reading. A single favourable date in a system with this much churn is as likely to be a re-sample as a result. The diligence method covers what to demand.

When should I fire an AEO vendor? When they cannot show you two dated readings, or when they have never reported a decline. As Firoz Azees puts it: "a measured zero is a work order; a flattering estimate is a story you tell yourself."

What is a realistic first win? A citation on a narrow, specific question where your evidence is genuinely better than the field's. Broad commercial questions come last, not first.

At Ivanooo, Firoz Azees runs Distinctiveness Engineering for the AI-answer era: measuring who the engines name, cite and recommend, then engineering the gap between listed and chosen. Start with a free AI visibility check and set your own clock running.