Know what the number is counting.
A visitor, a matched record, and a paid job are different things. Start there when you read a marketing report or a claim about performance.
Follow the record through the business.
Website traffic tells you that people arrived. An available identity match gives you a record to review. A conversation can establish whether the person needs work you offer.
Then come the estimate, the booking, and the completed job. Keep those stages separate. Counting a visitor match as revenue skips the work the office and crew still need to do.
Read the method behind the claim.
Intent and lead capture
The intent and lead capture framework describes how website behaviour can be used to assess possible service interest. Treat those signals as inferences until a conversation confirms the need.
Published advertising research
The Q2 2026 home service Meta ads research includes its dataset scope and measurement notes. Use the definitions and period on that page when interpreting the results.
Methods belong to the specific study. A published sample does not establish the outcome your company will get, and this index does not validate every statistic quoted in a blog article.
Before you use a number to make a decision
- Find the denominator. Is a rate based on visits, people, matched records, enquiries, or jobs?
- Check the period and sample. A busy month in one market may tell you little about another trade's quiet season.
- Remove duplicates. Repeat visits and existing customers can inflate a count of new opportunities.
- Separate observations from estimates. Collected revenue and predicted customer value should have different labels.
A before-and-after change can have several causes. Seasonality, budget changes, and how quickly the office responds can all affect the result.
Apply it to your own marketing.
Start with marketing attribution, then track the source on each lead record. Use the same definitions each month so you can compare results without changing what the report counts.
If a published figure is unclear, send us the page and the claim so we can check the context.