Evidence
Partly, and less certainly than it is sold. The evidence is one benchmark paper, newer preprints and a lot of vendor reporting.
The short answer
Partly. The strongest public evidence is a 2024 benchmark: adding sources, quotations and statistics raised visibility by up to 40% on older engines. Newer data says retrieval matters most. The effect on your site is unknown until you measure it.
What ranked on 18 September 2026: The academic query returns the arXiv paper, Wikipedia and Google's guidance. The commercial phrasing returns no primary evidence. Same subject, different standards of proof.
Last reviewed · 5 sources
Aggarwal et al. tested edits to a source document against a benchmark of queries.
A 10 October 2026 preprint studied 158,099 answers. Treat it as unreviewed.
The problem is structural, not dishonest.
A cheap, defensible test beats every published average.
0 of 3 done
0%
Questions
Not in controlled public form. AI platforms were 0.32% to 1.08% of site traffic in two studies, so measure your own.
A trend needs six to eight weeks of sampling to separate from noise. Anyone promising faster is describing hope.
Rewrite your top pages so the first paragraph answers the question alone and names the entity. It costs a day and helps human readers too.
Evidence, labelled
Each claim is labelled and sourced. Google rows speak only for Google.
Last reviewed · 5 sources
Adding sources, quotations and statistics raises visibility.
Reported
30 to 40% on position-adjusted word count in a 2024 benchmark. Older engines, domain-dependent. Keyword stuffing gave little to no gain.
Engines cite pages mainly because their own search retrieved them.
Reported
Preprint on 158,099 answers. Commercial questions only, ChatGPT focus, not peer reviewed.
Results repeat from run to run.
Reported
Under 1 in 100 repeats of the same brand list. A test needs many runs per prompt.
AEO lifts revenue.
Hypothesis
No controlled public test found. AI sent 0.32% to 1.08% of traffic, so effects are small for most sites today.
SE Ranking traffic study (opens in a new tab) source 1Conductor benchmarks (opens in a new tab) source 2
Go deeper
Next step
Tell me your site and what you are seeing. I will say which points above apply to you and which do not.