Two Lists Named 37 GEO Experts in 2026. None of Them Publish in Arabic

We checked both published expert lists in this field. Across 37 names, nobody publishes in Arabic.

That would be a curiosity rather than a problem, except that we have measured how differently AI engines cite Arabic. The two languages barely draw from the same sources at all.

Neither list did anything wrong. Both state their method and both drew from where this field publishes, which is English, so the gap you are looking at sits in the field rather than in their arithmetic.

Key takeaways

• Profound’s A-list started from 102 practitioners and named seven. Zero publish in Arabic.
• A separate list ran 1,000 prompts across five engines and named 30. Zero publish in Arabic.
• Five names appear on both lists, so the field is more concentrated than 37 suggests.
• In our own Saudi healthcare panel, Arabic and English answers to the same question share a mean source overlap of 0.149. Fourteen of 40 pairs share nothing at all.
• Ask “who is the best” in Arabic and institutions almost vanish: one of 121 Arabic citations came from an institution.
• Commercial pages take 69.3% of Arabic citations against 58.2% of English ones.

What do the expert lists actually contain?

Two lists, built independently, four months apart, using methods that share nothing with each other.

Profound published its 2026 A-list on 26 July 2026. It started from 102 practitioners and scored them on four weighted signals: original thought leadership, demonstrated results, peer and AI recognition, and educational impact.

Seven cleared the 80th percentile on all four.

Those seven are Kevin Indig, Mike King, Josh Blyskal, Aleyda Solís, Chris Long, Jason Barnard and Lily Ray, and you will recognise most of them if you follow this field in English.

The second list ran 1,000 prompts across ChatGPT, Claude, Perplexity, Grok and Gemini, then cross-checked the recurring names against published work. It named 30 people across three tiers.

Add them together and you get 37 names, of which five appear on both lists. That overlap tells you the field is narrower than the headcount suggests.

One of the 37 is Egyptian-born. Areej AbuAli founded Women in Tech SEO and is listed for building a 12,000-member community rather than for Arabic-language search work.

She publishes in English, like everyone else on both lists.

So the precise claim is narrow and checkable: nobody on either list publishes research in Arabic, and no agency on either list is based in the region.

Why would that matter if the work transfers?

It would not matter at all, if Arabic AI answers were built from roughly the same sources as English ones. Expertise calibrated on English would carry straight over to your market, and the language your practitioner writes in would be a detail.

That assumption is testable, and we tested it. It is the same discipline we described when we answered Seer’s 27 RFP questions in public: measure the thing rather than cite somebody who measured something adjacent.

Our Arabic versus English citation study asks the same healthcare question in both languages, in the same market, and compares which sources each answer cites. It is an interim release built on 87 captures across 49 distinct questions, so every interval on it is deliberately wide.

The headline number is the overlap between the two source sets.

Across 40 matched pairs, the mean overlap between the Arabic source set and the English one is 0.149, and fourteen of those pairs share no source at all.

So a practitioner who knows exactly which sources ChatGPT favours in English knows very little about the pool your Arabic-speaking customer sees for the same question.

What changes when the question is asked in Arabic?

The composition of the answer changes, and not merely the ranking of sources inside it.

Institutions largely disappear. Across our Arabic captures institutional sources take 9.1% of citations, against 24.0% for the same question type in English.

Ask specifically who is best and the effect sharpens to something close to absolute, because one of 121 Arabic citations on that question type came from an institution.

Commercial pages fill the space they leave, taking 69.3% of Arabic citations against 58.2% of English ones. The Arabic answer your customer reads is more likely to be assembled from someone selling them something.

This is not one vendor behaving oddly, because ChatGPT drops its Arabic institutional share by 14.2 points and Google by 14.9. The direction replicates across engines while the size does not, which is why you are getting both numbers here rather than an average.

Does ranking first in Google carry you into the Arabic answer?

Less reliably than most buyers assume, and this is the finding with the most of your money attached to it.

One laboratory in our panel holds 355 first positions in Google, and it appeared in 45.5% of the Arabic answers we captured for its own category.

Read that carefully before you draw a conclusion from it. One organisation inside an interim panel is a signal to investigate rather than a rule to plan around.

What it does rule out is the comfortable assumption that your Arabic AI visibility arrives free with your Arabic SEO.

If you are buying AI visibility work in Arabic for Riyadh or Cairo on the strength of somebody’s Google rankings, that assumption is the one to test first.

What this does not prove

Three limits, and they matter more than usual because the sample is small.

It does not prove the people on those lists would do bad work in Arabic markets. Several of them are among the strongest practitioners in this field, and the Profound list in particular is built on published research rather than reputation.

Competence travels a good deal further than data does.

It does not prove Arabic is unique, because we measured Arabic against English in Saudi healthcare and nowhere else. The same divergence probably exists in Turkish, Thai or Portuguese, and nobody has published that either. We are describing a hole in the field’s evidence rather than claiming a special case.

It does not carry statistical confidence you should treat as settled. Eighty-seven captures across 49 questions is an interim panel, and the repeat arm we ran to measure noise returned volatility uncomfortable enough that we published it. One question returned eleven sources on one day and none the next.

My read on this

I think the absence on those lists is a symptom rather than the problem itself.

The field publishes in English because its conferences, newsletters and tooling are English.

Engines that recommend experts then read that published record and return the same names, which is the loop the second list’s own author pointed at when she noted the engines over-index on the SEO-trained crowd.

Nothing in that loop is malicious. A body of evidence about English retrieval simply keeps getting cited as evidence about retrieval in general, and your market inherits the conclusion.

What I find harder to excuse is how little Arabic data exists at all. We could only write this article because we built the panel ourselves, and it is still an interim one. A region with this much search volume should not be relying on one agency’s incomplete study, including when that agency is us.

Objections we hear

“So you are saying we should only hire Arabic-speaking agencies.”

No. We are saying you should ask where the evidence behind the recommendation came from.

When an agency tells you which sources ChatGPT cites, ask whether it measured that in your language and your market or read it in an English study. Both answers are legitimate, and only one of them should be presented to you as measured.

“An overlap of 0.149 sounds low enough to be a measurement error.”

Partly it is, which is why we published the repeat arm and the volatility rather than only the headline. What survives the noise is the direction. Fourteen pairs sharing nothing is hard to explain as measurement drift, and the institutional gap replicates across two engines independently.

“Why would institutions get cited less in Arabic?”

We do not know, and our study measures the effect rather than its cause. One hypothesis is that fewer Saudi institutions publish answer-shaped Arabic content, which leaves commercial pages as the best available source for the engine to use. We have not tested it.

“Our brand ranks first in Arabic already.”

Then you already have the input for the one test worth running. Ask your category’s main questions in Arabic across two engines, list which sources get cited, and count how often your own pages appear. Your ranking is a hypothesis about citation rather than a substitute for checking it.

The more useful question

The question is not who belongs on a list of GEO experts, because both lists are reasonable and useful for the market they describe.

The question is which market they describe, and whether it is yours.

If your customers ask in Arabic, the evidence base deciding which sources get cited to them has been measured once, partially, by us. It needs measuring again by more people than us. Until that happens, treat any confident claim about Arabic AI visibility as a hypothesis, and that includes every claim on this page.

Ask us for an AI search audit and we will run your category’s questions in Arabic and English side by side, then show you which sources each answer is actually built from.

Related guides

• Arabic GEO: How to Get Your Arabic Content Cited by ChatGPT, Gemini and AI Overviews
• Best AEO Tools in 2026: Verified Prices, and Which Ones Actually Work in Arabic
• Sovereign AI in the Gulf: What Arabic Models Mean for How Your Business Gets Found
• What AI Overviews Cite in Arabic vs English: A 16-Prompt Audit