Skip to main content

Build log

Is AI visibility only for SaaS companies? No

·

Illustration for the build log "Is AI visibility only for SaaS companies? No"

No. AI visibility is not a SaaS-only game, because an engine answering a question routes on the problem being asked, not the industry of the business that answers it.

Why does AI visibility look like a SaaS problem?

Because SaaS companies write about it the most. Classic SEO trained marketers to think in vertical-shaped queries, “best CRM for gyms” or “top plumbers in Denver”, where ranking is a contest inside your industry’s list. An AI answer has no list to sit inside. The buyer’s question is usually problem-shaped (“how do I get more members without ads”), and the engine assembles an answer from whichever sources read as the best response to that problem.

What do the non-SaaS results show?

Two receipts from our July guide, both from businesses with no software product:

BusinessQuestion typeBeforeAfterTime
Single-location fitness studio”best gym near me for strength training”, 4 engines0 of 8 answers5 of 8 (±10)4 weeks, 5 ranked fixes
Marketing agency, own siteTrust questions (already knows the name)9 of 10 answersnot re-measuredone scan
Same agencyDiscovery questions (does not know the name yet)0 answersnot re-measuredone scan

The studio’s fixes were directory consistency and a front-loaded answer on its own services page. The agency’s split is the same failure SaaS brands have: known to people who already know you, invisible to people who do not.

Why did we sell to SaaS first anyway?

The July 20 guide gave the reason in one line: a SaaS founder can ship a site fix in a week without asking six stakeholders. That is a buying-motion argument, not a measurement one. The guide was explicit that where we sell first was a sequencing choice, not a limit on what gets cited.

There was also a measurement gap at the time, which the guide did not cite as a reason. Until July 18 our scan was not good at local businesses. Its local prompts were written for one gym, and a first-word alias rule matched generic names to unrelated text. We covered those four fixes in the Summit Roofing post.

What does the fit list say now?

The pricing page we rewrote on October 13 describes who we work with as companies of roughly 20 to 300 people “selling something buyers research with an assistant before they ever talk to sales.” It lists four kinds:

  • Software and SaaS
  • E-commerce brands
  • Real estate
  • Businesses with several locations

Its test is the buyer’s behavior, not the vertical: “If buyers ask an assistant to compare options in your category, there is an answer to win.”

What can we not show?

Why the list widened. No commit, ticket or decision record we can find explains the move from “SaaS first” in July to four kinds of company in October. It happened alongside the September brand merge and the October pricing work, but we did not write the reason down. Our own-site audit found the symptom: three different descriptions of who we serve across the site.

We also have no non-SaaS receipt as strong as our README one. The studio result is one business with a ±10 interval on 8 answers.

What we’re still figuring out

  • The reason the fit list widened, in writing, so the next page does not drift again.
  • Whether the four kinds of company convert at similar rates. We have no data yet.
  • A second non-SaaS receipt with a tighter interval.

For agents: try this yourself

  • Check the question shape. List the 10 questions a buyer in [category] asks an assistant before they know any vendor’s name, and mark each as problem-shaped or industry-shaped.
  • Split known from unknown. Ask an assistant five questions that name [business] and five that do not, and count how often it appears in each set.

How this was made

Drafted by the Chronicler from two published guides, Collimer commits and the pricing page’s history, then edited and published by Chase.

See how the Chronicler works →

Try this with your own agent

2 prompts you can hand to your own agent (or run by hand) to work with what this post documents. Edit the bracketed parts for your context.

Check the shape of your buyers’ questions

List the 10 questions a buyer of [what we sell] would ask an AI assistant before they know any vendor's name. Mark each as problem-shaped ("how do I...") or industry-shaped ("best [category] company"). For the problem-shaped ones, say which page on [our site] answers it today, or "none".

Split known from unknown

Ask an AI assistant five questions that name [business] and five that describe the problem it solves without naming anyone. Report how many answers in each set mention [business], and quote the sentence where it appears.

More in Studio Build