What a real contractor audit finds.

The findings

1. Technical score: 42 / 100

Largest Contentful Paint over four seconds on mobile. Uncompressed hero images, render-blocking third-party fonts, and no structured data of any kind. The site was not broken — it was simply slow enough and vague enough that Google had little reason to prefer it.

2. Google Business Profile: incomplete, wrong primary category

Filed under a broad contractor category rather than "Roofing contractor," which is the specific one the map-pack winners in that city were using. Services list mostly empty, description under 200 characters, six photos, none newer than eleven months. Profile freshness is now a visibility factor — this one had been idle long enough to slip.

3. Reviews: behind on both volume and recency

The three businesses holding the map pack averaged materially more reviews, and more importantly had recent ones. Review recency now outweighs raw count: a business collecting a few reviews a week beats one with a much larger, older pile. This company was collecting them sporadically, only when a customer thought of it.

4. No high-intent pages

One "Services" page covering every service, no page combining a specific service with a specific town, and no project pages. Nothing on the site could rank for "metal roof repair" plus a city name, which is exactly the kind of search that ends in a phone call.

5. Invisible to AI assistants

No structured data, thin third-party directory presence, and no mentions in any "best roofer in {city}" roundup. Google's AI leans on Business Profiles for local answers; ChatGPT and Perplexity lean on directories and listicles. This business was weak in both.

The projected outcome

After a rebuild, the technical score moves to 96 — that part is measured, because we can score the rebuilt site directly. The rest is modelling from the audit findings, and we label it that way deliberately:

We are not going to present a projection as a banked result. When we have enough completed engagements to publish real before-and-after numbers with client permission, this page will say so and the label will change.

What we would fix first

1. The primary category, same day. Highest-impact change available and it costs nothing.
2. The rest of the profile — services, description, photos, attributes — inside the first week.
3. The review engine, wired into their job flow so it runs continuously instead of in bursts.
4. The site rebuild, for speed, structure and the service-plus-city pages that were missing.
5. Directory and listicle presence, which is what makes AI assistants name you.

Note the order. The website is fourth, not first — which is the opposite of what most agencies lead with, because the website is the thing they are selling.