Which channels produced homes that were actually delivered?

A new manufactured home averages well over a hundred thousand dollars, and a multi-section home considerably more. Between the agreement and the revenue sit land, a site inspection, permits, a foundation, transport and set — commonly five to eight months.

That is a different business from a vehicle sale, even though the dealership floor looks similar. Financing splits between chattel and real-property loans depending on whether the buyer owns land, and that changes both the cycle and the close rate for the same advertisement.

Matching delivered homes back to the original enquiry is the only way to compare channels on the outcome that matters, because the enquiry and the revenue are almost never in the same reporting period.

A buyer placing a home on their own land has a longer cycle, a larger home and a mortgage-like process. A buyer moving into a land-lease community closes faster on a smaller home. The same keyword produces both.

A placement-type column splits them so the report does not average two unrelated businesses into one cost per sale, which is what makes most dealership channel reporting unusable.

A signed purchase agreement on the lot is a long way from money. The buyer has to be approved, and approval depends on whether the home is going onto land they own, onto leased land in a community, or onto a family member's plot, because each of those runs through a different kind of lender with different terms. Applications get declined, co-signers get added, deposits get returned and homes come back into inventory weeks after everybody shook hands. The deal is signed, celebrated, logged and then quietly unwound, and no marketing report anywhere reflects that it happened.

If the export counts agreements rather than funded, delivered homes, the channel ranking is measuring optimism. Sources differ sharply on this: one produces buyers who are already pre-approved and clear underwriting easily, another produces walk-ins with enthusiasm and no credit file. On cost per sale the two look similar, and the salesperson on the lot already knows they are not. One of those channels can be scaled and the other cannot, and the difference shows up in the funding rate rather than anywhere on the media account. Nothing in the platform can see an underwriting decision.

Exporting funded or delivered deals rather than written ones is the single most important choice a dealer makes before the first upload. It also makes the report reconcile to what the accounts recognise, which matters when the general manager is deciding whether to believe it. A status column lets written and funded be reported separately if the dealer wants to see the gap by source. The gap is often large enough to be the most useful single figure on the first report, because it separates demand the dealership can serve from demand it merely attracted.

The sticker price of the home is the number everybody quotes and the least useful one for judging a channel. Real margin on a deal is assembled from several places: the home itself, the set and the foundation, skirting and steps, air conditioning and appliances, the delivery, and commissions on insurance and service agreements. Two deals at the same price can differ by tens of thousands in what the dealership actually earned. A general manager knows this instinctively and has no way to hold a campaign to it, because the media report only ever sees one of those numbers.

Trade-ins and used homes make it wider still. A reconditioned home sold off the back of the lot can carry better margin than a new multi-section unit, while looking like a small sale on any report built on price. A source that reliably produces used-home buyers may therefore be the most profitable thing the dealership advertises and the worst-performing line on its media report. It is also the deal the dealership can actually source, which matters in a market where new inventory arrives on the manufacturer's timetable rather than the buyer's.

The fix is to export gross rather than price where the accounting system produces it, and to carry a home-type column alongside. The report then ranks channels on what was earned, split by new, used and multi-section, which is the breakdown that decides whether a campaign is bought again. Nothing has to be modelled; the column simply travels with the row. What the report ranks is whatever column the accounts already trust, and that is usually the one the dealer principal reads on a Monday morning anyway.

Most home centres capture demand on paper or in a simple log: the up sheet, a visitor book, a spreadsheet the sales manager keeps. It records a name, a telephone number, a date and possibly which model the family walked through. It is filled in by a salesperson between appointments and nobody audits it, because no money depends on it. Six months later, when the home is delivered, nobody revisits it at all. It is, in practice, the only record that the family ever came to the lot before the paperwork started.

That log is nevertheless a perfectly usable enquiry file. It has the two things the match needs, a contact detail and a date, and it was written down at the moment demand arrived rather than reconstructed afterwards. Uploaded alongside the ad platform export and any call records, it turns lot traffic into a channel that can be ranked next to paid search rather than disappearing into a walk-in total. A hundred handwritten lines become a hundred rows, and the walk-in bucket stops being a place where all the unanswered questions are kept.

This is why a dealership that has never tracked anything can still produce a defensible report for a completed year. The delivered-deal export exists because the business had to invoice and title the homes. The up sheet exists because somebody wanted to follow up. Neither was kept for marketing, which is precisely what makes them reliable. That is the whole argument for working from exports rather than from tags: the files that were kept for money are the files that were kept properly, and they were kept whether anybody was measuring or not.