We rebuilt Steele Services' website from their own job records and search data
Steele Services is a commercial fire protection contractor in New Jersey, working across NJ, PA and DE. They came to us for a rebrand and a new website: one that reads like the established regional operation they are, and that brings in work through search.
Most website projects start with a design and a list of services drawn from the client's vision. This one started with the company's own records, and that difference shaped every decision after it. Here is the process in four steps, and why it benefitted the client tremendously.
1. They asked for a rebrand and a new website
The brief was direct: look like the premium shop they already are, retire the small-shop presentation, and build for search. The old site ranked for the company name and nothing else, so there was no search equity to protect and no reason to keep any of it. We started from a blank page with a new logo system, a new design, and a site plan that was driven by the data.
2. We pulled their job records and ran the search data against them
The client gave us access to the software their office runs every job through. We pulled their data via an API: every job, every site, and every problem a technician wrote up across five years.
This info shows what the business actually is. It outlined the percentages of types of jobs they are executing, along with hundreds of logged findings, in the technicians' own words, of what fails inspections in their field.
Then we ran SEMrush, the keyword tool, seeded from those service lines, to learn what buyers type. The findings reshaped the plan. Searches with a town or state attached barely exist in this field, so the local pages in the original plan were not worth building, and local visibility would come from the Google map listing instead. Buyers search by service, in plain words: pages named for the service they want earned clicks, while pages aimed at an industry or a brand earn almost none. And the tool's own labels understate how many of those searches are purchases, so we judged each phrase by what it costs advertisers and how it is worded rather than by its label.
3. The data drove the site map and the content plan
The site map came out of the two records together. The work the company does most now leads the site, and the other services sit in proportion to it. Services with real field volume but no page of their own got one, and pages in the original plan with no demand behind them were dropped before they were built. Where competitor data showed buyers searching for something the plan had missed, a page was added. Each service page now targets one phrase with measured demand behind it, in the words a facility manager types, or is reached from a page that does.
The content plan works the same way. A script crosses the logged findings with an index of 46,284 keywords and ranks each topic by how often the crew sees the problem and how many people search for it. A topic has to score on both. The first article it produced, on fire extinguisher inspection tags, went live the same day, and the technicians now send photos and notes from the field through a phone form to feed the next ones.
4. Real photos
The client had over 4,000 photographs from real jobs. We scanned the whole album against the site's needs, stripped the location data from every file, kept out anything showing customer paperwork or another company's tags, and gave each page a hero image of the service that page describes. Were there gaps? Sure. We flagged these and are still actively working to source them in the near future.
The copy follows a curated voice skill we created through manually reviewing the initial drafts: plain words a property manager uses, no scare framing, and no claims they cannot back. Automated checks run before any copy change goes live and flag an unsupported claim, a duplicated line, or a retired phrase. Reporting claims in our first draft came out from these checks.
What the client got out of it
- A site shaped like the business. Inspection leads because inspection is the work, so a buyer sees the company as it is.
- Pages only where work and demand meet. Three service lines gained a page. A page and a matrix of more than two dozen pages that would never have earned a visit were never built.
- Content no competitor can copy. The sections on what fails inspection come from more than nine hundred of their own findings, and the article plan is ranked by their own field information.
- Target phrases that are measured, not guessed. Each page competes for a phrase buyers actually type.
- Nothing on the site the company cannot back.
- Real photographs, published safely. Location data stripped, customer paperwork and other companies' tags excluded.
- Copywriting rules that outlast the web build. The rules and the checks live with the site, so all future content goes through the same QA checks and reads like the same company.
- A launch that works on day one. The site, steeleservices.net, went live with a contact form that delivers, analytics, and markup that search engines and AI assistants can read.
- A pipeline for what comes next. Field notes from the technicians and a ranked topic list turn into future articles with proof of demand behind each one. We built an automated skill that drafts these articles from the notes and the ranked list, and a weekly text that asks the crew for new material, so the library keeps growing on a schedule with every draft reviewed before it goes live.
Do you need an expert who can diagnose your business and match your website content to what it actually does? That is the work we do: start with a conversation, or with the free preview and see what we build from what is already public.