The short answer
To get your company recommended by ChatGPT and Perplexity, you make the facts about your company reachable, structured, unambiguous, and current — so that when an assistant retrieves and synthesizes an answer, your entity is the one it can identify and confidently name. No reputable source lets you buy a citation; what you can do is win on data quality. This guide is the practical companion to what AI visibility (share of model) is, translated into steps a marketing team can run this quarter.
Assistants differ in their plumbing — Perplexity leans on live web retrieval and shows its sources, while ChatGPT blends trained knowledge with real-time search — but the levers that earn a mention are the same across both. Work them in order.
Step 1: Make your data reachable
An assistant can only recommend what its retrieval system can reach. Confirm your most important pages are crawlable and that you are not blocking AI crawlers you actually want, then make sure the canonical facts about your company live somewhere a retrieval system trusts — not only inside a JavaScript-heavy marketing site that machines parse poorly.
A neutral, structured directory is one of the highest-value places to be reachable, because it is precisely the kind of source assistants pull from when comparing companies in a category. Claiming or adding your entity is free; start at list your company, or, for a service business, list a local service business.
Step 2: Publish structured, machine-readable facts
Assistants and the search engines behind them read structured data — expressed as JSON-LD — more reliably than prose. Google's own documentation explains that structured data describes a page's meaning in a machine-readable way, which is exactly the comprehension an assistant needs before it will name you.
Publish, at minimum:
- An
Organizationprofile with your legal name, description, logo, and primary category, following the schema.org Organization vocabulary. ProductandServiceentities for what you actually sell, each with its own clear description.- FAQs as structured
FAQPagedata, since assistants frequently lift answers verbatim from well-formed Q&A.
On this site, authoring those facts publishes them as validated JSON-LD automatically. The Local and Brand plans describe what each edition lets you control; correcting the facts is always free.
Step 3: Disambiguate your entity
The most common reason an assistant won't recommend a company is that it cannot tell which company you are. Entity disambiguation fixes that. Build a sameAs graph that links your profile to your own domain plus authoritative references — Wikidata, Crunchbase, and your LinkedIn company page. Wikidata is especially important because it feeds the knowledge graphs behind many assistants; the Wikidata introduction explains how to get an entity listed.
Add alternate names and common variants ("Acme", "Acme Power", "Acme Inc.") so a model resolves typo'd and shortened forms to the same entity. For brands with a parent company or multiple regional arms, encode the parent organization and locality so models can disambiguate by location.
Step 4: Build topical density, not a single mention
Models weight sources by how thoroughly they cover a topic, so one stray mention is weak signal. Aim to appear across a cluster: your directory category page, comparison surfaces, and related guides. A backup-power brand, for example, gains from being present alongside peers like Goal Zero, Jackery, and Anker SOLIX in the backup generators category, because density across the cluster compounds retrieval weight for every page in it.
This is also why category-specific work pays off; see AI visibility for generator brands for a worked example of building density in one vertical.
Step 5: Keep it fresh and signal updates
Freshness is a real ranking and retrieval factor, and Bing-powered assistants (Copilot, ChatGPT search) reward fast re-indexing through IndexNow. Date-stamp your content, update it on a schedule, and use a publishing surface that pings IndexNow on every change. A profile that was accurate two years ago but lists a discontinued product is worse than no profile, because it teaches the model to describe you incorrectly.
Step 6: Measure, then close the gaps
Treat this as a measurable program, not a one-time cleanup. Sample the buyer queries for your category across ChatGPT and Perplexity on a repeated schedule and record three things: whether you are named, how prominently, and whether the described facts are correct. The diagnostic guide walks through running that audit yourself. Wherever accuracy is the gap — wrong service area, stale pricing tier, missing product — that is the cheapest and fastest win, because correcting facts is free and immediately changes what the model reads.
A note on what you cannot and should not buy
You cannot buy your way into an AI recommendation on a source worth trusting, and you should be wary of any vendor that claims otherwise. The value of a directory to a model depends on its neutrality; the moment placements are for sale as "citations", models learn to discount the source. On this site, citations, the organic authority score, and the verified tier are never for sale. Paid plans let you author and amplify your own information — and any amplification is clearly labeled and capped — but they never alter the facts or the organic ranking. That discipline is what keeps the recommendations, and therefore your mention, worth having. When a recovery surge hits after an event like a hurricane, the companies that prepared their data are the ones assistants name.