Plugin Visibility Index · Run 02

Do two AI engines recommend the same WordPress plugins?

Asked the identical question in ten categories, Perplexity and Google AI Mode name a median of 53% of the same products and disagree on the leading pick in four of ten. That is about twice the disagreement one engine shows with itself.

What was measured, and against what baseline?

One prompt per category — “best WordPress [category] plugins in 2026” — put to Perplexity and to Google AI Mode. For each answer I recorded which products were named and in what order, matching against a fixed candidate list per category so the comparison is mechanical rather than a judgement call.

A raw disagreement number on its own means nothing, because these engines are not deterministic. So the number that matters is the comparison against a baseline: how much does one engine disagree with itself? Run 01 answered that by putting identical prompts to Perplexity twice. That within-engine figure is the yardstick.

The gap between 75% and 53% is the part that is really about the engines being different rather than being random. It is not enormous, and it is not nothing. Roughly: a quarter of the variation you see between two engines would have happened anyway if you had just asked the same engine twice.

Where do the two engines agree, and where do they not?

Note that agreeing on the winner and agreeing on the field are different things. Tables and LMS both share a top pick and still overlap under 45%: the engines agree who wins and disagree about almost everyone else. If you are the category leader that is comfortable. If you are third, which engine someone asked decides whether you exist.

Which plugins are invisible on both engines?

Ten of the 52 tracked plugins — 19% — were named in no open-ended answer on either engine. Not ranked low. Not mentioned.

All-in-One WP Migration is the one worth staring at. It has over five million active installs. Across six Perplexity answers and one Google answer it was named exactly once — in the answer to “UpdraftPlus vs All-in-One WP Migration”, where the prompt supplied the name.

That distinction matters more than it looks. The engines are not ignorant of the product; asked directly, Perplexity produced a detailed, accurate comparison of it. They simply never retrieve it when the buyer has not already thought of it — which is every discovery question that matters commercially.

Does a dominant source on one engine carry over to the other?

No, and this is the sharpest result in the run.

Booknetic is the single most-cited domain in the entire Perplexity corpus — 67 attribution marks, one citation in ten, spread across five categories, as Attribution 01 sets out. As a product it led two of Perplexity’s four booking answers and appeared in all four.

Google AI Mode did not name Booknetic at all in the same booking question.

Whatever earns a publisher that position on one engine does not transfer. A visibility strategy validated on a single engine is validated for that engine only, and a monthly report drawn from one engine will show a picture that the other engine does not share.

Do both engines make the same factual mistakes?

Also no. Perplexity named Restrict Content Pro in all three of its membership answers, twice in third position, once calling it the best lightweight option — a product that has been sunset. Google AI Mode did not name it, and named Ultimate Member and BuddyBoss instead.

So stale recommendations are not an inherent property of AI answers. One engine’s index had caught up with a product’s end of life and the other’s had not. For a vendor whose product was acquired, renamed or discontinued, that means the correction has to be checked per engine, not once.

What can this run not tell you?

Ten categories is a small sample for a median. The per-category overlap figures are single observations and should be read as illustrations of a spread, not as measurements of those categories. The headline comparison — 53% against a 75% baseline — is the number I would defend; the individual rows are not.

The two engines were also queried two days apart, and product matching used a fixed candidate list, so a product neither engine has ever been observed naming could be missed by both the answer and the list. Neither of those is likely to move the headline, and both could move a single row.

Google AI Mode exposes only a handful of inline source links and keeps the rest behind a panel, so the attribution-share method used in Attribution 01 does not port across. Cross-engine source comparison is the obvious next run and this one does not attempt it.

---

Which engines can see your plugin?

Send me your wordpress.org slug. I will run your category on both engines and tell you whether you are missing from one or from both — and which of the two problems that makes it. Free, no call, usually within a day.

---

Newer in this index

Run 03 compares the two engines’ source lists rather than their answers. They agree on 53% of the products and only 15% of the domains — and Google took YouTube as a source in eight categories where Perplexity took it in none.

---

Newer in this index

Run 04 supplies the baseline this page was missing. Repeating an identical query returns an identical source list — a cache. Change one character and a fifth of the sources move. Change engine and seven-eighths do. The cross-engine gap is about four times the noise inside either engine.