Present in 24 of 84 answer receipts.
Public AI recommendation scoreboard
When a buyer asks AI what to buy, which brands become the answer?
Hobby Answer tests real buying questions across ChatGPT, Claude, Gemini, and Perplexity. We show who gets recommended, which sources shape the answer, and the evidence gap your team can act on.
One to three buyer questions, a dated receipt, and one named gap. No card. The paid Benchmark is the decision-ready analysis.
How the scoreboard works
From buyer question to evidence-backed action.
Follow one illustrative buying question as it becomes a four-engine recommendation pattern, a source map, and a focused plan your team can act on.
Illustrative Benchmark excerpt
Thornwood Miniatures: visible in specialist answers, absent from the beginner buying moment.
16 buyer questions
4 AI engines
84 dated receipts
Jul 8 test date
Illustrative example—not a real brand ranking.
Every brand, result, excerpt, and source pattern below is fictional. It demonstrates how a Hobby Answer result is organized and interpreted.
Actively recommended in 10 receipts.
Usually appeared after competitors.
Official pages appeared in 6 receipts.
Highest-value named gap
“What is the best beginner terrain paint set?”
Observed
Across eight repeated receipts, Thornwood appeared once. RivalCo Terrain appeared seven times and received a strong recommendation six times. Five answers cited a retailer guide or creator tutorial.
Reasoned hypothesis
Thornwood leads with individual products, but lacks a buyer-facing beginner bundle page, comparative fit guidance, and consolidated FAQ evidence.
DECISION
Build the beginner-proof and comparison evidence first. Then rescan the same questions.
Where the answers found proof
Retailers and creators framed most recommendations.
Why it matters: YouTube, Reddit, retailer guides, and reviews can supply the language and proof that AI answers cite or repeat. The owned site still needs to make the buyer case clearly.
Prioritized 30-day plan
Turn the finding into a controlled test.
- Publish the answer.Create one beginner terrain-paint hub with bundle guidance, product fit, and a clear buying path.
- Strengthen the proof.Add approved product facts, FAQs, and two independent demonstrations.
- Measure movement.Rescan the same questions and compare recommendation position, citations, and competitor share.
What the scoreboard measures
Evidence a client can use—not a vanity score.
Share of answer
How often the brand appears across priority buying questions.
Recommendation strength
Whether the answer merely mentions the brand or actively recommends it—and in what position.
Source pattern
Which owned pages, retailers, creators, communities, and reviews support the answer.
Gap closure
Whether published changes improve the same questions on a controlled rescan.
Transparent by design
The receipt matters as much as the score.
Every paid Benchmark includes the exact prompts, test dates, engine and model information where available, run counts, faithful answer excerpts, source URLs, scoring definitions, and material limitations.
Read the full methodology →- Observed factWhat the answer said, cited, recommended, or omitted.
- Reasoned hypothesisOur labeled interpretation of why the evidence pattern may exist.
- LimitationHow model, date, geography, wording, personalization, and source availability may affect the result.
Your buyers are already asking
Find out whose evidence AI trusts before the buying decision happens without you.
Request the Benchmark
Free scorecard for a first signal. Benchmark for the exact prompts, repeated runs, dated receipts, competitive source map, and prioritized plan.
