Buyers now ask an assistant before they ask your site.
SaaS discovery moved to comparison and recommendation, which are assembled answers. If a system cannot state what your product does, you are not in the shortlist.
The recurring pattern
5 failures · sector-specificThese are patterns, not a diagnosis. Which of them applies to a specific site, and in what order they are worth fixing, is what a reading against the 100-point scale establishes.
Why this sector fails this way
The structural reason, not the symptomSaaS discovery moved before most SaaS sites did. The shortlist is now routinely assembled by a system answering “what should I use for this”, and that system builds its answer from what it can read and corroborate rather than from what the marketing site asserts. A category claim made only on your own homepage is, to that system, an unsourced claim.
The structural weakness is usually the split between documentation and marketing. The documentation holds the specific, answerable, high-intent material; the marketing site holds the authority and the links. Where those live on different subdomains, in different systems, behind different rendering, neither accumulates what it needs — and the competitor whose docs are simply crawlable becomes the citation instead.
Where the work sits
Two of five services · two specialismsAEO
When a system answers instead of linking, is the answer yours?
GEO
Can a generative system state a fact about this brand and defend it?
Next.js SEO
Rendering strategy per route, metadata inheritance, and what actually reaches the served HTML.
AI search optimisation
Extraction and corroboration, for a category bought after a question is asked.
Common questions
Answer first, then the reasoningShould we publish comparison and alternative pages?
They are frequently the highest commercial intent queries in the category, and most sites have no page for them. The constraint is that the comparison has to be accurate and maintained: a stale competitor page is worse than none, because it is checkable.
Our docs rank and our marketing site does not. Is that a problem?
It is a problem when the two are not connected, because the documentation answers the question and nothing routes that reader to the offer. It is not a problem to be solved by removing the docs from the index.
Does rendering really matter for a marketing site?
It matters most for the systems doing the assembling. A marketing site behind a client-side render is read inconsistently, and the check is a comparison of served HTML against rendered DOM per template, not per page and not a score.
Build visibility
Seven questionsAtlas reads the site before the first conversation.