You can send forty proposals in a week. Four of them are the business.

When a measurement and a branded proposal take minutes rather than an afternoon, the natural instinct is to quote everything that moves.

That is good operations and it destroys lead-based reporting, because the proposal count stops carrying any information about whether the lead was real.

The number that still means something is the signature, and ranking channels on signed value rather than on proposals sent usually reorders the list substantially.

It also puts a cost on the volume: a source needing ten proposals per signature is consuming ten measurements, ten follow-ups and ten days of somebody's attention.

None of that appears in a cost-per-lead figure, and all of it is paid by the business.

A hailstorm produces a fortnight of enquiries that no advertising caused, at values set by an insurance adjuster rather than by your pricing.

Planned replacement is a slow, considered purchase where the homeowner has been thinking about the roof for a year and compares three contractors.

Blended, the storm weeks inflate whichever channel happened to be running, and the conclusion will not repeat because the weather will not.

Where the export carries a job type or an insurance flag, the report splits on it; where it does not, the date range usually identifies the storm itself.

The point is not to discard storm revenue but to stop it being read as proof that a campaign worked.

Door-knocking after a weather event produces a large share of roofing revenue in many markets, and it has a real cost in wages and time.

Because no invoice arrives from an advertising platform, it sits outside every marketing report and the paid channels are judged as though they were the whole budget.

A canvasser code on the lead record puts it into the same comparison on the same signed value, against what it actually cost to have people walking streets.

Contractors who run this frequently find the ranking between canvassing and paid search is not what either side of the argument assumed.

It also identifies which neighbourhoods and which crews produce signatures, which is a scheduling decision rather than a marketing one.

A claim-funded replacement has a value set by the adjuster, a longer timeline, and a conversion pattern driven by the claim rather than by your sales process.

Averaged with retail work it produces a figure describing neither, and a channel weighted towards claims will be read as though it produced retail jobs.

Separating them gives two honest rankings, and most roofers discover the sources feeding each are almost entirely different.

Where the export flags the claim, this is automatic. Where it does not, the amounts and the timelines usually make the split obvious within a few minutes.