Experiment

Do AI tools favour first-hand experience?

We tested whether recommendation queries cause AI platforms to rely more heavily on first-hand experience than traditional publisher or brand-owned content.
blimpp-reddit-dtc-mistakes-ai-seo-hero

Why we ran this experiment

We kept seeing Reddit, forums and review-led content appear prominently in recommendation responses. We wanted to know whether this was a repeatable pattern or simply anecdotal.

Our hypothesis

query requires judgement or recommendation.

What we found

First-hand sources accounted for 58% of citations across
recommendation-style prompts…

Results by AI tool

AI tool First-hand experience Brand content News / editorial Other
ChatGPT 62% 18% 12% 8%
Claude 54% 24% 14% 8%
Perplexity 59% 21% 13% 7%
Gemini 56% 26% 11% 7%

The evidence

A representative result from ChatGPT illustrates the pattern.

blimpp_observation-hero

Key takeaways

  • First-hand evidence matters most during consideration.
  • Brand-owned content remains strong for factual queries.
  • Third-party presence becomes increasingly important…

Experiment details

TEST PERIOD 

1–7 September 2026

QUERY TYPES

Discovery, Need-State, Validation

AI TOOLS

ChatGPT, Gemini, Perplexity, Google AI Overviews

CATEGORIES

5

QUERIES TESTED

50

CITATIONS ANALYSED

300

Limitations

The experiment used a fixed prompt set and was conducted during a single seven-day period…

Next steps

We will repeat the test across additional verticals and isolate whether source preference changes by platform.

  • First-hand evidence matters most during consideration.
  • Brand-owned content remains strong for factual queries.
  • Third-party presence becomes increasingly important…

Have a question?

Interested in running a similar experiment?

We’re always testing new ideas. Tell us what you’d like us to investigate.

Got a hypothesis?

Have an idea for an experiment?

We’re always testing new hypotheses. Tell us what you’d
like to see us explore.