Local Businesses And The AI Recommendation Problem
Record the conditions alongside the results: which assistant, which model version if visible, whether web access was on, the date and the run number. When a result changes sharply, the conditions log is usually what tells you whether the world changed or your setup did.
Where you serve several towns, resist the instinct to claim the widest possible area. A stated coverage radius that you genuinely honour is more useful than a list of thirty places you would only travel to reluctantly, because the specific claim gets quoted and the vague one does not. Being the obvious answer within a tight radius produces more work than being one of many possibilities across a county.
There is a specific failure that catches out otherwise well marketed companies. An assistant clearly knows things about them, cites a page that mentions them, and still declines to recommend them, or worse, confuses them with a similarly named business in another country.
Performance and Score Based Models Both sound aligned and both create problems. Payment tied to mentions creates pressure to shape the prompt set toward questions you already win, which is measurable improvement that means nothing.
Accuracy Beats Coverage The most common real defect is not missing markup, it is markup that disagrees with the page or with the rest of the web. A founding year in your schema that differs from your about page. A logo URL that returns a 404. A contact point nobody monitors.
How Identity Fragments Fragmentation is rarely deliberate. It accumulates through ordinary business activity: a rebrand that was applied to the website but not to old directory listings, a legal name that differs from the trading name, an office move recorded in some places and not others, a founder's profile that lists a different company spelling.
Make Sure It Can Fetch You Check that your robots.txt permits the relevant crawler, and check your server logs for what it actually receives. Bot management products frequently serve challenge pages to legitimate retrieval agents, which produces total invisibility with no error anyone sees.
That transparency makes it the best available proxy for how retrieval based answering behaves generally. Here is what the citation pattern reveals, and what a brand can actually do about it. get recommended by ai
Fix the Prompt Set and Never Casually Change It Your prompt set is the instrument. If you adjust it between runs you are measuring your own edits, and any trend line you draw afterwards is meaningless.
Where Third Party Coverage Fits Even after all of the above, most citations in a commercial category will point somewhere other than your site. That is not a failure of your optimisation, it is how the system weighs self interested sources.
Where Analytics Can and Cannot Help Referral traffic from assistant domains does show up in analytics, and it is worth segmenting into its own report. Treat the numbers as a floor rather than a count, since some assistants strip referrer information and some traffic arrives looking direct.
One structural decision saves a lot of trouble later. Keep the raw answers in plain text files named by date, assistant and run number, rather than pasting them into a document that gets reformatted. Six months in you will want to search across every run for the first appearance of a competitor or a source, and a folder of plain files supports that while a slide deck does not.
One overlooked cost is your own time. Every engagement in this field needs somebody inside the business to confirm figures, approve crawler changes and answer factual questions, and a plan that assumes this is free will stall. Budget a few hours a month explicitly and name the person, because the alternative is an agency waiting on answers and billing for a month in which little shipped.
A retainer describing ongoing optimisation and strategic guidance with no countable deliverable is a subscription to a relationship. It may still be worth having, and you should know that is what you bought.
The shortlist is shorter than a conventional local results page, which raises the stakes on being included. Being fourth on a map still gets calls. Being fourth in a recommendation that names three businesses gets none.
The risk is scope drift into activity that is easy to report and hard to value. The protection is to have the retainer specify countable units: prompt set runs per month, listings audited, corrections submitted, pages published or rewritten, outreach attempts made.
What to Do About llms.txt and Similar Files Proposals for machine readable files aimed specifically at language model consumers appear periodically. Adoption is inconsistent and support varies by provider, so treat these as low cost and speculative rather than as a requirement.
Assertions with nothing behind them are weaker than silence, because they introduce a detail that fails verification. The pattern that works is reciprocal: your site names the profile, the profile links to your site, and some independent source associates the two without either of you being involved.