Verified against ChatGPT · 2026-08-13
Plan where to place genuinely citable facts so an LLM has a real reason to mention your brand
Builds a citation-building plan focused on placing specific, verifiable facts and data points in places an AI model would actually draw from, rather than a generic PR or backlink outreach list.
The prompt
Ready to copy — highlighted parts are example details you can swap.
I want a plan for increasing the odds our brand gets mentioned or cited when people ask AI models questions in our space — not a generic backlink or PR outreach list, specifically a plan built around what would make an AI model have an actual reason to reference us. OUR BRAND AND WHAT WE ACTUALLY KNOW We've processed 50,000 freelance invoices through our platform and could report an actual average payment delay by industry, but haven't published this analysis anywhere yet. WHO CURRENTLY GETS CITED IN THIS SPACE A well-known industry association's annual report gets cited repeatedly for freelance payment statistics, even though it's two years old. WHAT WE HAVEN'T PUBLISHED YET Internal dashboard showing payment delay by industry and by client company size, currently only used for our own product analytics. START FROM THIS PRINCIPLE: models cite sources that contain a specific, checkable fact, statistic, or original data point tied to a query — not sources that are simply well-known or well-optimized in the traditional SEO sense. Given that, do the following: 1. Identify what specific, genuinely original data or facts we actually possess (from the brand and expertise details given) that nobody else in this space has published in a citable form yet. Do not invent a statistic we don't have — if the input doesn't describe an actual data point we hold, say so and note that this is the gap to fill before a citation plan can work. 2. For each real data point identified, name the specific query or question it would be the best possible answer to, and where it should be published so it's structured as a directly citable answer (a labeled stat with source and date, not folded into a longer narrative). 3. Compare against what's currently getting cited — is it winning because of a genuinely stronger fact, or just because of a first-mover advantage or brand recognition? Say which, since only the first case tells us we need a better fact, not just any publication. 4. Flag anything in our unpublished assets that looks like it could become a citable data point once it's published, and what minimal packaging it would need. OUTPUT: numbered plan following the four steps above, ending with a one-paragraph honest assessment of whether our current data actually supports a strong citation play in this space or whether more original research is the real prerequisite.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Asked generically for a citation-building or GEO plan, GPT-5.1 tends to produce a repackaged backlink-outreach or digital-PR checklist, because that's the closest well-represented pattern in its training data for "getting mentioned online" — explicitly starting from the principle that models cite specific checkable facts rather than well-known sources redirects the model away from that template and toward reasoning about what your brand actually possesses that's citable, which is a fundamentally different and narrower question. The instruction not to invent a statistic the brand doesn't actually have is the load-bearing accuracy control in this prompt, because a plan built on a fabricated data point isn't a plan at all — it's a suggestion to publish something false, and the brief this prompt operates under specifically requires never presenting invented statistics as fact; forcing the model to say "this is a gap, not a plan" when the input lacks real data keeps the output honest even when the honest answer is less satisfying than a confident list of tactics. Requiring the model to distinguish whether current citations win on a genuinely stronger fact versus first-mover advantage or brand recognition matters because these two situations call for different responses — if the association's report is cited mainly because of prior brand recognition, a strategy of "give AI models something better cited" can genuinely displace it, but if it's cited because no competing data exists, publishing something is the prerequisite over publishing something more optimized. The closing honest-assessment requirement counters the natural pull toward ending on an upbeat, actionable-sounding note even when the actual input reveals the brand doesn't yet have a strong citation play — which is a real and useful answer a stakeholder needs to hear before investing in a plan built on a data gap.
What you get back
1. Real data point: average freelance payment delay by industry, drawn from 50,000 processed invoices — not currently published anywhere. 2. Best-fit query: 'average time to get paid as a freelancer by industry' — should be published as a standalone, dated, labeled statistics page, not folded into a blog post. 3. Current citations (the industry association's 2-year-old report) appear to win mainly on brand recognition and recency of nothing better existing, not because their underlying data is stronger — a fresher, more granular data set has a real chance of displacing them. 4. Honest assessment: the underlying data exists and is strong; the actual blocker is that it's never been packaged as a citable public asset, so the real next step is publication, not further research.
Verified against
ChatGPT GPT-5.1 · 2026-08-13
Changelog
- 2026-08-13 — Initial publish, verified against ChatGPT GPT-5.1.
Building this for real?
This is a free starting point. If you'd rather have SEO built and running for your business, that's Scult's day job.
EXPLORE SEO
