Which marketing channel sells which product?

Do different marketing channels sell different products for a contractor?

Yes, and by a wide margin. At the eight-figure Toledo exteriors company where I direct marketing, Facebook runs about 90% siding by design and paid search runs roughly 40% roofing and 40% siding. One aggregator sends a third roofing and a quarter windows, another sends 81% siding, and the company mix is 38% roofing and 31% siding.

What our product mix looks like

Roofing is 38% of net revenue this year to date on the CRM’s sales-by-product report, siding 31%, windows 8%, repairs 7%, gutters 6%, doors 5%, metal roofing 3%, and commercial, flat roof and insulation the rest. That is the blend the whole company produces. For a long time I budgeted as if every channel produced that blend in miniature, which none of them does.

Why Facebook sells siding

Because the agency built it to. The creative is siding and the campaigns are built around siding, and the result is a channel that runs about nine siding jobs in ten. That was deliberate, and it is also the reason the channel worked after four years of not working (Facebook “Didn’t Work” for Four Years. Then We Built a System.): my read is that siding is a visual product, a before-and-after in a feed is a siding ad whether or not it says so, and a roof is a roof.

Paid search is the opposite kind of channel. The homeowner types “roof replacement” or “siding contractor” and the channel sells whatever was typed, so ours runs roughly 40% roofing, 40% siding and 20% everything else, close to the company mix but not the same as it.

What the aggregators send

The two aggregators we still buy from send different houses. One, on its own zip-level cost report, runs about 34% roofing, 27% windows and 17% repairs, which is a windows-heavy mix we get almost nowhere else. The other runs 81% siding with 11% metal roofing, which is a metal share three times the company’s. Neither of those mixes is something we chose. It is what the platform’s homeowners ask for. Cutting one of those lines does not just remove leads, it removes windows leads, or metal roofing leads, and the sales team feels it as a product shortage before anyone in marketing sees it as a channel change.

What this does to a budget

Two things. A channel cut is a product cut, so the question before cutting a line is not only what its cost of marketing is but what product it was feeding and whether another line feeds that product. And a product goal is a channel plan. If leadership wants more siding, the lever is Facebook. If they want more windows, the lever is the aggregator with the windows-heavy mix, and neither of those is “spend more on marketing.”

The place this bites hardest is the demo calendar. A siding demo and a roofing demo are not the same appointment. A rep sitting in a siding-heavy month from a Facebook push closes a different ticket than one sitting in a roofing month from search.

Where this is wrong

For a single-product shop, a roofer who only roofs, there is no mix to manage and this whole note collapses to “which channel sells.” And the channel-to-product splits above are partly operator numbers. The Facebook and search shares are what the agency and I set them to and see in the reports, not a report the CRM produces on its own, while the aggregator and company-mix figures come from reports. There is no month-level product data in our CRM. Within a channel the mix is assumed constant across the year, which it probably is not.

What to do

• Tonight: pull sales by product for the year from the CRM, then pull the same by source if the CRM will do it. The gap between the two is the note.
• This quarter: before cutting or growing any channel, write down which product it feeds and which other channel feeds that product.
• Ongoing: when a product goal comes down from leadership, translate it into a channel plan before it becomes a budget number.

Sources: Lead Perfection “Sales Efficiency by Product,” Toledo, 2026 year to date, first party; aggregator zip-level cost-of-marketing files, March 2025 to March 2026; Facebook and paid-search product shares are the author’s operating figures, not a CRM report.

Related notes