Publish proprietary surveys, benchmark reports or unique datasets that other sources need to cite when answering category questions. Original research becomes the canonical source AI engines reach for — durable citation share for years on a single investment.

1. 1

   Audit how your brand currently appears in AI engines: ask ChatGPT, Perplexity, Claude, and Google AIO the same 5 buying-intent queries.

2. 2

   Identify the citation sources the AIs are pulling from — almost always Wikipedia, Reddit, G2, and a few trusted blogs.

3. 3

   Plant content where those AIs source: get listed on G2 / Capterra with strong reviews, contribute to Wikipedia's references, get featured in independent listicles.

4. 4

   Add FAQ schema and Article schema to every key page — AI engines lean heavily on structured data.

5. 5

   Publish an llms.txt and robots-friendly content so the right crawlers can ingest you.

6. 6

   Re-run the audit monthly — AI engine results shift far faster than Google SERPs.

Stripe's "State of the Developer Ecosystem", OpenView's "SaaS Benchmarks Report", and Reforge's growth surveys are cited routinely across Hacker News, Twitter, and now AI search engines. The reports cost ~$50k to produce and generate years of inbound links, AI citations, and warm pipeline.

• ✗Calling the test too early — read results at full 7-day multiples to neutralize day-of-week effects.
• ✗Skipping the audience / segment definition — broad targeting dilutes the signal and inflates CAC.
• ✗Failing to instrument the downstream metric — clicks are not revenue.
• ✗Watch-out from the playbook: Pick stat angles that will get cited — 'how X varies by company size' or 'what % of teams do Y' beats 'how customers feel about Z' for AI ex

Established companies with survey-capable audiences or proprietary data