The assistants
Each one reads your site differently — and keeps changing.
Being readable to one doesn't mean being readable to all of them, and what works this month can stop working next. We check each of them separately, and keep checking.
OpenAI
Answers by reading the live web as it goes. It favours sites it can actually read — ones that let it in, publish what they offer in machine-readable form, and give it something it can confirm: a barcode on a product, an address and hours on a local business, a licence or a named team on a firm. A vague page gets passed over for a competitor with a clearer one.
Answers from search, from product feeds and from the business listings Google already holds. Whether you're sorted into the right category and properly described is what decides who gets named — the same housekeeping that has always driven Shopping and local results, now read by the assistant instead of a person.
Microsoft
Doesn't yet answer these questions in a way we can measure honestly, so we don't pretend to score it. What we do instead is check you're ready — the feeds, the identifiers, the structured detail — so you're not starting from zero on the day it does.
What each one wants from you depends on what you are. A shop is judged on catalogue data; a local business on premises, service areas and hours; a software company on plans, documentation and comparisons; an advisory firm on credentials, named expertise and proof of outcomes. We resolve which of those you are before we grade you — see how it works.
We ask through the assistants' official developer interfaces, never by scraping the chat apps. Deep dives on each assistant are on the way; in the meantime, the free check shows how each one reads you today — and which of them can't.