The honest starting point
A July 2026 survey of 45 studies on getting content into AI answers found that being relevant to the question and showing up prominently are the most repeatable factors. Generic tricks carry over poorly from one AI engine to another, and gains shrink once competitors copy them. Its bottom line: no technique has yet shown a stable, long-term effect across platforms.
So anyone promising a guaranteed spot in ChatGPT is ahead of the evidence. What the evidence does support is below, with the Algotally tool for each. And because none of it is guaranteed, Algotally measures what follows each change.
What the evidence supports
What others say about the business
A 2025 study across many industries found AI search leans heavily on independent sources (directories, reviews, press) over a business’s own site and social posts. For local businesses this is usually the biggest lever.
In Algotally: the directories and lists AI cites in each market, and which ones the business is missing from; a listing kit with copy to paste; a review request kit for the review sites AI reads there; and pitch drafts for local outlets and “best of” lists.
Being readable, and correct
AI repeats what it reads. A website that blocks automated visitors can’t be read at all, and wrong hours or a closed location on a directory travel straight into answers.
In Algotally: a website check on every report and, on Growth and up, a check of what AI says against the business’s real details, with a kit to correct what it gets wrong.
Clear facts on the business’s own site
Plain statements of what the business does, for whom, where, and at what price are easy for AI to find and repeat. Vague marketing copy isn’t.
In Algotally: from the business profile, a pricing section, FAQ answers and schema markup for the site.
Each engine on its own
Engines differ in which sources they use and how often they search the web, so a result on one says little about another.
In Algotally: Perplexity, ChatGPT and Claude are each asked with web search on. On Agency, results can be read engine by engine.
Regular search still matters
Engines that search the web draw on what search engines find. Good ordinary SEO is still the foundation.
In Algotally: we don’t measure Google rankings. We measure what AI recommends and the sites it reads to decide.
What the evidence doesn’t support
- Guaranteed placement. AI answers vary from one ask to the next. Nobody can promise one.
- Keyword stuffing, and rewriting pages just to be “citation-friendly”: the survey found such rewrites can make pages harder for AI to find.
- Special “AI files.” Algotally generates an llms.txt file because it costs nothing, but there’s no solid evidence yet that it changes what AI recommends, so we don’t sell it as a fix. We measure it like any other change.
Why we measure instead of promise
Ask an AI assistant the same question twice and you’ll often get a different list. So we ask the same buyer questions every week, in each market, and report how often each business is named, with the number of answers behind every figure (how it’s measured).
On Growth and up, when a step is done we compare the weeks before with the weeks after, and only call it a change when it clears a noise test. That shows what followed each step, for each business.
On the Agency plan, those results are pooled across all subscribers: which kinds of change have moved AI mentions, and plans ordered by them.