AEOAEO for B2B SaaS Meet with an expert

Methodology: how sources are chosen and numbers are cited

AEO for B2B SaaS does not run its own studies. It reads published research and platform documentation, checks the figures against the original, and explains what they mean for a B2B SaaS website. This page sets out how.

Which sources are used?

Sources are used in this order of preference:

  1. Platform documentation, such as Google Search Central and the crawler pages of OpenAI, Anthropic and Perplexity, for how a feature or crawler works.
  2. Published studies with a stated sample, where the publisher says how many companies or accounts were analyzed. Where a study leaves out its period or how its sample was chosen, the page says so.
  3. Published benchmark tables without a stated sample. These are labeled as benchmarks from their publisher, not as study findings.

Opinion posts are not used as evidence for a number.

Some AEO questions have no published study with a stated sample. That currently applies to how many B2B buyers use AI search, and to whether particular page formats cause AI tools to cite a page. Pages on those subjects say so and give no figure.

Which sources does the site draw on?

Every study, survey, benchmark and platform page cited anywhere on the site is listed here. Topic pages name their sources in the text and link to this list. Articles carry their own source list as well.

Studies, surveys and benchmarks

Platform documentation

How is a number cited?

Every figure is tied to a named study and publisher: in the sentence, beside the table, or in the source line at the top of the page. The sample and date are given where the source states them, and each article lists its sources in full at the end. Where a study quotes a figure from somewhere else, the original source is named or the figure is left out.

Figures are copied exactly. They are not rounded, averaged across studies or adjusted.

What are the limits of the data?

Much of the published research on AI search comes from SEO software vendors and agencies analyzing data they have access to. That makes it a convenience sample, not a random one. The sites studied may be larger, or more US-based, than the market as a whole, and most studies cover Google only.

Pages say so wherever such numbers appear. Three further rules apply:

  • A modeled projection is never described as an achieved result.
  • A comparison between two groups of companies is not described as a before and after.
  • A benchmark is a range to compare against, not a target every company should hit.

How are examples and arithmetic handled?

Two kinds of number on this site are not study findings, and both are labeled where they appear.

  • Arithmetic applies a published rate to an example, such as a percentage change worked out from two published rates. The inputs are shown.
  • Invented examples illustrate a method, such as a sample robots.txt file. They are described as invented.

How often are pages updated?

Every page shows the date it was last updated. A page is revised when a source it relies on publishes a new edition, when a platform changes the feature it describes, or when a reader points out an error.

How are corrections handled?

If a figure on this site does not match its source, the page is corrected and the updated date changes. Substantive corrections are noted on the page.

Talk your AI search visibility through with an expert

A free consultation. Bring your AI search, crawler or content question and brainstorm the next step.

Meet with an expert →