How we collect and publish data

What our numbers mean, how we got them, and what we refuse to claim.

Last updated Aug 5, 2026

The rule

Every figure on this site is either from our own survey, with the method stated, or attributed to a named public source. We don't republish the unsourced statistics that circulate in this industry — the ones about what percentage of calls get missed that trace back to nothing you can check.

We sell call answering. Quoting numbers we can't defend would be the fastest way to lose the argument we're trying to win.

The Ontario HVAC survey

• Sampling. HVAC-related queries against the Google Places API on 2026-07-31, across 8 Ontario cities, deduplicated by phone number and business name. 428 businesses.
• Enrichment. Each business's own public website was fetched and parsed for contact details and platform fingerprints.
• Platform detection. Fingerprint-based, on the public website only. A positive result means we found evidence. A negative means we didn't — not that the shop runs no software.
• Review counts. As reported by Google Places at fetch time. A proxy for how long a shop has been operating and roughly how big it is. Nothing more.
• City pages. Only cities with at least 20 businesses in the sample. Below that the sample can't support a claim.

Known limitations

• Places API coverage isn't a census. A shop with no Google presence isn't in our sample. That biases toward businesses that market online.
• It's a snapshot. One day, 2026-07-31. Businesses open, close and rebrand.
• Eight cities isn't Ontario. No Toronto, no Ottawa. We describe it as an eight-city sample because that's what it is.
• Website fingerprinting is a floor, not a ceiling. It systematically undercounts platform adoption. Treat the platform figure as a minimum.

The after-hours study, not yet run

The survey holds a field for after-hours call status. It reads not_tested on every record because we haven't run it. When we do, it will be observational — dial, record whether a human answers, voicemail picks up, or it rings out, then hang up. No message and no sales pitch, so nobody's time gets wasted and it stays clear of telemarketing rules. Method and results will be published here together.

Corrections

If a number here is wrong, tell us and we'll fix it and say that we did.

Put in your number and try to stump it. No sales call, no card.