Website builder research
How we build the samples
Our reports describe platforms detected on discovered local-service websites. They are historical observations, with an explicit count of websites whose builder could not be identified.
What one website means
An industry report counts a website once within that industry. The same website may legitimately qualify for several service categories. The combined report counts each eligible website once across all industries; it does not average the industry percentages.
We combine ordinary versions of the same website address. Separate customer sites on shared website-building services keep separate identities. The unit is a website, rather than a count of business locations, individual contractors or legal companies.
We use the most recent usable homepage capture for a website. If equally recent captures contradict one another, the website is held out until the conflict is resolved. Ambiguous builder clues within a single otherwise usable page stay in the sample as Unknown.
Which sites qualify
The homepage must make the service a primary or co-leading business focus. Evidence can come from the business identity, page title, main heading or prominent service descriptions. A passing mention in a review, footer or long list of unrelated services is insufficient. A house-cleaning business can qualify while also offering office cleaning; a carpet specialist mentioning a customer's home does not automatically qualify.
We require a successful homepage capture with usable content. We exclude directories, social and booking profiles, parked or expired domains, empty pages and genuine access challenges. A contact form's CAPTCHA badge is not itself an access challenge; saved pages can be rechecked to correct that distinction.
The combined report shows the candidate-to-eligible counts and final exclusion reasons. Relevance corrections are based on website evidence. Cases with ambiguous service identity remain for further review.
How a builder is identified
We look for platform clues in the saved homepage and website address, including an explicit platform declaration, recognizable platform resources or a customer address hosted by a builder. A third-party tracking script alone can be misleading, so stronger evidence takes priority. Unresolved conflicts remain Unknown.
Unknown means no reliable builder identification. It is never treated as proof of a custom-built website. Other identified platforms requires positive evidence for another platform. WordPress includes detectable self-hosted and WordPress.com sites; this research does not establish their hosting plan.
Every share uses the entire eligible sample, including Unknown. We show exact site counts, round most shares to one decimal place, and display a nonzero share below 0.1% as <0.1%.
Capture dates and review dates
A capture date is when the homepage evidence was saved. A review date is when that evidence and the sample were checked. Reviewing an April or May capture later in the year does not make it a current website observation.
Each report shows its homepage capture range and evidence review date. The combined report also shows the distribution by month, helping explain a broad range where only a few sites were captured later. A current website may have changed since its saved capture.
What these results can support
This is a nonrandom sample of websites found during local-service research. Discovery sources, search coverage, geography, reachable pages and detectable platform clues can all introduce selection bias. A larger sample reduces neither that bias nor uncertainty about unidentified platforms by itself.
State, province and city counts describe stored discovery-location labels. They are not independently verified business addresses or service footprints. Search matches may include Canada or other countries, so these counts do not establish a United States-only cohort.
The reports support statements about the archived sites we checked. They do not establish national market share, current usage, conversion performance, customer satisfaction or which builder is best. Small platform groups deserve particular care: a difference of a few sites can change their apparent order.
Public evidence examples illustrate the detection method and are clearly labeled as a small sample. Downloadable platform counts cover the full aggregate sample; a download of example evidence covers only the examples shown.