We wrote down every number before we touched anything

You have seen the pattern: an agency publishes a chart with an arrow going up and no verifiable place where the line started. It is easy to write because nobody wrote down the “before.”

So we did it in the other order. Every figure below was captured on one date, before launch, using checks that can be repeated by anyone — reading the live pages, running a fixed list of searches, and pulling the business’s own verified Google profile. In roughly three months we re-run the identical checks and publish both columns. If a number does not move, it will say so.

Why the zero on schema matters most

Structured data is how a website states facts in a format software can quote: this is the business, here is the address, these are the engines, this is the price, this is the warranty. With none of it present, every search engine and every AI assistant has to infer the whole business from paragraphs of prose — and when it is not confident, it names a competitor that spelled it out instead.

Three of ten. And the three wins are all searches that already contain the town name — meaning they mostly capture people who know the shop exists. The seven misses are where someone with a blown 5.9 and a credit card is shopping, and on every one of those the results are filled with national resellers and out-of-state rebuilders.

The good news hiding in that table: nothing is blocking the assistants. They are welcome to read the site — there is simply not enough specific, structured, quotable substance for them to pick this shop over a competitor that publishes engine families, tiers, prices, core policy and warranty terms in plain machine-readable form.

A 3,500-follower audience built organically around engine work is a genuine asset, and it is the one thing here that is already working. The problem is the destination: every click currently lands on a page that cannot show an engine, quote a price, or capture a lead, and shares out of that page render with no image at all.

Four phone numbers is a quiet, expensive problem. Search engines and AI assistants use name, address and phone agreement as a confidence signal; when a business publishes four numbers across four places, every one of them gets trusted a little less.

1. Website and technical audit. Every URL in the live sitemap was fetched and parsed for title, meta description, heading structure, structured-data blocks, image alt coverage, social tags and page weight.
2. Fixed tracked query set. Ten commercial-intent searches were chosen to span local, metro, regional and national intent, then run once and recorded as top-10 presence or absence. The same ten strings get re-run verbatim.
3. Answer-engine citation check. For each query we recorded whether the domain appeared among the sources an answer engine drew from, separating local-qualified queries from national commercial-intent ones.
4. Answer-engine readiness check. robots.txt, llms.txt and on-page structured data were inspected directly to see what a machine can actually access and quote.
5. Social and directory sweep. Five social platforms plus Yelp and the local chamber directory were checked for an owned presence, follower count, recency and business-information consistency.
6. Verified profile data. Rating, review volume, category setup, listed phone numbers and owner reply rate were pulled from the business’s own verified Google Business Profile with owner-level access.
7. Automated re-run. The whole procedure — the query list, the per-page audit script, the profile pulls and the roll-up arithmetic — is stored as a single reusable routine and scheduled to fire on December 19, 2026, so the follow-up is executed the same way rather than reconstructed from memory.
8. Archived evidence. Full-page screenshots of all live pages and the raw technical data were captured and archived on the capture date, so the “before” can still be inspected after the old site is gone.