Your customers have started asking an assistant before they ask Google. When they do, one of two things happens: you are in the answer, or a competitor is. There is no second page to be on.
What is not true is that there is a new discipline for fixing this. Google’s own guidance is blunt about it — “optimizing for generative AI search is optimizing for the search experience, and thus still SEO” — and most of what is sold as AEO is either ordinary technical SEO with a new invoice, or a file nobody reads. We wrote up what the evidence actually shows, including the parts that are inconvenient for a company selling this service.
What we actually do
Make sure assistants can reach you. Crawler access is the one genuine prerequisite, it takes an afternoon, and a surprising number of sites are blocking the agents they want to be read by. We check your robots.txt against the current crawler list and confirm from your logs who is really visiting.
Make your facts agree. This is the work that matters most and almost nobody sells. If your site, your LinkedIn page and your directory profiles disagree about what you do, when you started or where you are, an assistant cannot resolve which company you are — so it describes a competitor it can pin down instead. We find every source that describes you and reconcile them.
Get you into the sources answers are built from. Assistants quote directories, roundups, review sites and community threads. If your category’s “best of” lists do not include you, no amount of on-site work will put you in the answer. We identify which sources matter for your category and what it takes to be in them legitimately.
Structure content so an answer can be lifted from it. Clear questions, direct answers, real figures with sources. The one peer-reviewed study in this field found that citing sources, adding quotations and including statistics raised visibility, while keyword stuffing did not.
What we will not do
We will not send you a monthly report of your “ranking in ChatGPT”. No vendor can measure that: answers differ by user, by session and by day, and nobody outside OpenAI has the data. Anyone showing you that chart is showing you a sample of one, dressed up.
We will not claim credit for a rise in AI referral traffic without a control. Assistant usage is growing fast enough that untouched pages show the same curve, and the best-known case study in this category turned out to be mostly that.
We will not build you an llms.txt and call it a strategy. We maintain one, and we will set yours up — but the largest published study found that 97% of valid llms.txt files received no requests at all in the month measured, and Google’s guidance says plainly that you do not need one.
We ran this on ourselves first
In September 2026 we audited our own site with the same method, and it found real problems: our founding year was given three different ways across three sources, our schema resolved us no more precisely than “Ontario”, and our own machine-readable file described services we do not sell. The full findings and what we changed are published, including the parts that were embarrassing.
That is the honest version of this service. You get a written baseline, a list of findings in priority order, and a clear line between what we fixed and what the market did.