What AI assistants actually recommend

A public index of which products ChatGPT and Perplexity recommend when buyers ask buying questions — measured weekly, published with the method.

This week: round 11, 19 October 2026

Perplexity answer overlap, −3.7 pt since round 10

Scanned 19 October 2026. 135 samples from Perplexity, round 11. AI Customer Support for Ecommerce; short ink tick: round 10; 3 days since previous round. All figures measured. Methodology v2026-09-29.1.

Round 11 landed this morning. Perplexity now agrees with itself on 62% of our AI-support questions, down from 83% a week ago; ChatGPT slipped to 77%, from 82%. One engine is still making up its mind.

Categories, round 11

Ecommerce Email & Retention

ChatGPT answer overlap, +4.0 pt since round 10

Perplexity 69%, −3.0 pt

Widest engine split in the top 5: Brevo, ChatGPT 45% vs Perplexity 27%.

Product ChatGPTPerplexity Gap
Omnisend 59%56%
Klaviyo 48%55%
Brevo 45%27% 18 pt

Scanned 19 October 2026. 204 samples across ChatGPT and Perplexity, round 11. Answer overlap per engine; short ink tick on each arc: round 10; gap in percentage points from unrounded rates, shown where 10 or more; 3 days since previous round. All figures measured. Methodology v2026-09-29.1.

Full table

AI Customer Support for Ecommerce

Perplexity answer overlap, −3.7 pt since round 10

ChatGPT 77%, −2.6 pt

Widest engine split in the top 5: Help Scout, ChatGPT 10% vs Perplexity 27%.

Product ChatGPTPerplexity Gap
Gorgias 50%53%
Zendesk 24%30%
Tidio 29%22%

Scanned 19 October 2026. 270 samples across ChatGPT and Perplexity, round 11. Answer overlap per engine; short ink tick on each arc: round 10; gap in percentage points from unrounded rates, shown where 10 or more; 3 days since previous round. All figures measured. Methodology v2026-09-29.1.

Full table

65 question pages and 246 product pages, all free. How we measure

One email a week. One chart, one story, what changed.

Three minutes to read, every number labelled.

How the index is measured

Latest articles