The Sources AI Actually Trusts About You
Fix your website perfectly, and an AI assistant can still describe your business wrong. This surprises people every time, and it shouldn't, because it's the same lesson search engines taught two decades ago and everyone forgot: your own site is one voice in a room full of other voices talking about you, and you don't get to pick who else is in the room. It's also the reason asking ChatGPT what your business does can turn up something wrong even after you've fixed everything on your own site.
Where a grounded answer about your business actually comes from
When an AI assistant is asked about your business, it doesn't read only your website. It draws on whatever it can find and cross-reference: your site, yes, but also review platforms, business directories, news mentions, industry listings, LinkedIn, forum threads, and anything else that's been indexed and looks relevant. Then it synthesizes an answer from whatever combination of those sources seemed most trustworthy and most relevant to the question asked.
That synthesis step is where good businesses get quietly misrepresented. A directory listing you set up once in 2019 and never touched again is a source. A review site's auto-generated summary of your business category is a source. A local news mention from before you pivoted your offering is a source. None of these are malicious, and all of them are competing with your current, accurate homepage for the privilege of shaping what an AI says about you, and the current, accurate one does not automatically win.
The three categories worth knowing
What you control directly. Your website, and anything you can log into and edit. This is the highest-leverage category because you can fix it immediately, and your website specifically is exactly what the previous post in this series covers. Your social profiles belong here too, in the sense that you control them, but they need a caveat serious enough to earn their own section below.
What you can influence. Directory listings, review platform profiles, business databases. You can usually claim, update, or correct these, but it takes deliberate effort, because nobody reminds you they exist or that they've gone stale.
What you don't control at all. Old news coverage, forum discussions, third-party mentions. You can't edit these, but you can outweigh them, by making sure the accurate, current story is so consistently and clearly told everywhere you do control or influence that it becomes the dominant signal by sheer weight and consistency.
The walled garden problem
Here's the caveat the category above glosses over. Instagram, TikTok, and Facebook aren't built to be read from the outside. They're built to keep visitors, and their attention, inside the platform, and one consequence of that design is that most of these platforms restrict or block the automated systems AI companies use to read the web. A meaningful share of what gets posted also sits behind a login wall that a logged-in human can see and an outside crawler simply cannot.
The result is a trap that's easy to fall into precisely because it feels productive. A business posts consistently, keeps its bio current, answers comments, and does everything right by social media's own rules, and an AI assistant asked about that business can still be working from information that's a year out of date, because the current, accurate version only exists somewhere the AI was never able to read. You did the work. The platform just didn't pass it along.
So the earlier category needs a split. Your website is both controllable and, once Article 2's fixes are in place, generally readable by the systems now answering questions about you. Your social profiles are controllable but frequently not readable, no matter how current you keep them, because the barrier isn't your effort, it's the platform's walls.
The practical fix isn't leaving social media, it's not treating it as a substitute for your own site. If a fact about your business only lives in an Instagram post, for most AI systems it may as well not exist. The same update, restated as a line of ordinary text on your own website, becomes available to a reader that the platform was never going to let in.
Why this isn't a one-time fix
Search engines taught an entire generation of businesses to obsess over their own website and largely ignore everything else, and that habit is exactly wrong for how AI assistants actually form answers. A perfect website sitting next to five stale, uncontrolled, or simply forgotten sources is still a business that gets described inaccurately, because the AI doing the describing doesn't know which source you'd vote for.
The practical implication is an audit, not a redesign: find out what's actually out there across the sources that matter, fix what you control, update what you can influence, and know honestly what you can't touch and should simply outweigh. That's a different exercise than making your homepage prettier, and it's the one most businesses have never done, mostly because until recently nothing was reading across all those sources and synthesizing an opinion from them at once.
Something is now doing exactly that, every day, for every business your customers are curious about. Yours included.
Mapping what's actually out there, across every source that shapes your AI answers, is the audit I run for clients. Get in touch if you'd rather map it than guess at it.
Rosemarie Withee has spent thirteen years helping operations teams get real work out of their software, first Microsoft 365, now AI. She’s written six books for Wiley and builds AI products at Portal Integrators.