July 1, 2026 · 3 min read

What our first real scans taught us about local business websites

Our first real users arrived from paid search clicks and scanned their own businesses: contractors, barbershops, cleaners, food service, professional services — from the US and well beyond (.ca, .mx, .br, .uk domains showed up in week one). We never publish anything about an individual scanned business, but the aggregate patterns were loud enough to write down.

Scores spread wide — roughly 38 to 84 — but the reasons barely varied. The same handful of signals scored weakest almost every time, and none of them are design problems.

Signals that scored weakest across our first scans Service-area clarity Availability cues Pricing expectations Trust placement Contact consistency most often weak least often weak
Qualitative ranking from our first cohort of real scans. Bars show how often each signal was a site's weakest, not exact counts.

What each of those looks like in the wild:

  • Service-area clarity was the most common weakest signal. Sites imply their territory — an area code, a city in the logo — but almost never state it: "We serve X, Y, and Z." A comparison shopper hesitates; an AI assistant simply can't confirm you serve the searcher's suburb, and moves on.
  • Availability cues were nearly as thin. Hours, response time, same-day or emergency service — the questions that decide urgent hires — were routinely unanswerable from the page.
  • Pricing expectations were absent far more often than pricing itself is secret. Nobody expects an exact quote on a homepage. But "most jobs run $X–$Y depending on scope" beats silence in every comparison, and silence was the norm.
  • Trust placement was a near-miss everywhere: genuinely strong reviews and credentials existed but sat far from the call-to-action — proof and decision separated by three scrolls.
  • Contact consistency produced the strangest finds — details differing between pages of the same site. Small inconsistencies read as neglect to a human and as unverifiable data to a machine.

The meta-lesson: these are missing sentences, not missing budgets. Not one of the weakest signals requires a redesign, a rebrand, or a developer. A serving line, a pricing range, an hours block, a review moved next to the button — an afternoon of copy edits. That's also why consumers increasingly let AI tools shortlist local businesses — the assistant does the tab-comparison humans hate, and it rewards sites that answer plainly. Bain's AI-search research suggests those answer-driven journeys are growing, which means the missing-sentence tax compounds.

It also explains something that surprised us about our own funnel: high scorers didn't feel safe either. A site scoring 83 still usually had one weak signal quietly filtering it out of somebody's shortlist. "Good website" and "verifiable business" are different tests.

See which sentences your site is missing — the scan is free, takes about a minute, and shows quoted evidence from your own pages: agentsitescan.com/scan.

References