Skip to content

How LeadWinner Measures Response Times and Statistics

Every speed claim and statistic we publish follows the rules on this page.

How we measure reply speed

When we say the AI replies "in about 10 seconds," the clock is defined the same way every time:

  • Start (T0): the moment the platform's webhook delivers the new lead to LeadWinner.
  • Stop (T1): the moment our reply is accepted by the platform's API.

On that clock, typical reply times are about 10 seconds on Yelp, about 20 seconds on Thumbtack, and about 15 seconds on Google Local Services Ads. These are typical figures, not guarantees — each reply depends on the platform's API and its latency at that moment. Note that the clock starts when the platform hands us the lead; how long a platform takes to deliver a webhook after the customer hits "send" is outside our measurement and our control.

The dataset behind our research

LeadWinner's product and research draw on more than 250,000 real lead conversations between US home-service businesses and their customers. Published studies state their own sample and method — for example, our lead-timing study covers 100,000+ leads analyzed in each business's local time, with the full methodology in the post itself.

Aggregate-only research

Our research never exposes customer data. Findings are published as fleet-wide aggregates — shares, medians, and distributions — from which no individual business, customer, or conversation can be identified. Timing studies use metadata such as timestamps and business categories, not message content. We publish no real names, phone numbers, or dollar amounts from customer accounts, anywhere.

External statistics

When we quote third-party research — in posts like our lead response time statistics guide — every figure is attributed to a named source with a link to the organization that published it. We check the source's own wording before citing it, we don't invent or round percentages into better stories, and when a widely repeated number has no traceable origin, we either say so explicitly or leave it out.

Corrections

Spotted a number that looks wrong, outdated, or misattributed? Tell us — a human reads every message, and we correct published figures rather than defend them.