Discoverability: Can AI find you?
Crawlers allowed in robots.txt, indexed in Bing (ChatGPT leans on it heavily), pages that render without JavaScript. Nothing else matters if this fails.
One page covering why this happens and the six-phase system we run to fix it, in the same order, every time.
About 78 out of 100 research answers cite an outside source rather than the company's own site. Backlinks barely move the needle here. A 75,000-brand Ahrefs study found plain brand mentions with no link at all matched AI visibility better than backlinks did. Most companies have almost nothing off-site for a model to learn from, so it falls back on whatever it can find: old reviews, a forum thread from 2022, a competitor's comparison page.
If you have nothing on any of these, the model fills the gap with whatever ranks, usually a competitor's content or a stale review nobody corrected.
Fix content without fixing entity data and the model still can't tell what category you're in. Fix reviews without fixing pages and there's nothing to quote. We run all six as one engine, the same order every time.
Crawlers allowed in robots.txt, indexed in Bing (ChatGPT leans on it heavily), pages that render without JavaScript. Nothing else matters if this fails.
Entity data, schema markup, your About page, LinkedIn, Crunchbase, all pointing at the same clean facts, so the model stops guessing.
A review push to real customers on G2, Capterra and Reddit, corrected listings, and a plan for what shows up when someone asks what's wrong with you.
Rewritten category pages, comparison pages, alternatives pages with stats and quotes a model can lift. The structure the Princeton GEO study found lifts citations 30 to 40%.
Press, YouTube, original research, comparison sites, and community threads answered as a person with a real, size-qualified opinion, not a vendor pitch.
The same 40 to 60 questions, same engines, every month. No moving goalposts, no swapped metric because one looks bad. If a number doesn't move, we say so.