Which channels produced households that actually qualified?

In income-restricted housing an application is only the start. Income has to be verified, household composition documented and eligibility certified, and a large share of applicants do not complete or do not qualify.

That makes application volume an actively misleading metric. A channel that produces a flood of ineligible applications costs the leasing team weeks of work and produces no revenue, while looking like the best performer on every dashboard.

Matching certified move-ins back to the original enquiry ranks channels on households that cleared the process. During a lease-up, when the compliance deadline is real, that is the only ranking worth having.

Housing authorities, caseworkers and community organisations send a meaningful share of qualified households, and they cost nothing in media spend. They are usually excluded from marketing reporting entirely, which makes the paid channels look more important than they are.

Uploading partner referrals as a source file puts them on the same axis as paid search, so the leasing budget can be judged against the alternative of spending that time on relationships instead.

Marketing here is not a steady programme. A waiting list opens for a defined window, sometimes by lottery, sometimes for a fortnight, and closes again once it holds enough households to fill the foreseeable turnover. Advertising runs hard for those weeks and then stops entirely, occasionally for years. Comparing month against month, or running a campaign continuously because the platform prefers it, is straightforwardly the wrong shape for the problem. The advertising calendar is set by the list, and the list is set by turnover and by whatever the property can realistically process.

That burstiness breaks most conventional reporting. A source measured across a quarter containing one open window and two closed months produces an average that describes nothing, and the platform's own optimisation has nothing to learn from because the objective disappears halfway through. Meanwhile the real question — which sources filled the list fastest with households who could qualify — goes unanswered. Platform optimisation is built for continuous objectives and this one is not continuous. By the time a campaign has learned anything useful, the window has closed and the spend has stopped.

Because the report is built from date ranges you choose rather than from a rolling window, the open period can be analysed on its own and compared against the last one, however long ago it was. That is the comparison worth having: this opening against the previous opening, by source, on households that eventually moved in rather than on applications received. Two openings two years apart are a perfectly reasonable pair to compare, and there is no window or expiry that prevents it.

A property with units at different income bands is not marketing one product. The deepest-subsidised units frequently need no advertising at all and fill from a list that is already too long. The shallower band, closer to market rent but with restrictions attached, competes directly with unrestricted apartments nearby and can be genuinely hard to lease. The same campaign is expected to serve both, and its performance is reported as a single figure. Leasing staff spend their effort where the difficulty is, and the reporting spreads the credit evenly across work that was not evenly hard.

That single figure is dominated by the easy units. A source producing plenty of households for the band that needed no help looks excellent, while the spend aimed at the band that actually required marketing is buried inside the same total. Leasing teams often know which units are the problem and cannot show it in the numbers they are given. The result is a budget defended by numbers drawn from the units that never needed it, and cut on numbers drawn from the same place. Nobody in the chain is misreading anything; the report simply cannot separate the two.

An income-band or unit-type column separates them, so each band's leasing can be judged on its own evidence. It changes what gets bought as well as how it is reported: the channels that reach households just above the deepest subsidy are usually not the same ones filling the rest of the building, and nothing in a blended report would ever reveal that. One column, present in almost every rent roll already, is the difference between a report that describes the building and one that describes an average of it.

A unit turns over and then sits. The outgoing household's paperwork closes, the unit is cleaned and inspected, the incoming household's income has to be verified and their file completed, and any of those steps can add weeks. The rent lost during that period is real and is reported as vacancy loss to the owner, and it is frequently assumed to be a marketing failure when it is nothing of the kind. It is one of the few places in leasing where the marketing team is routinely blamed for a delay it has no means of influencing.

This matters for measurement because it decouples enquiry volume from revenue entirely. More applications do not shorten the process, and a property drowning in applicants can still be losing rent on empty units. Buying more enquiries in that situation is expensive and does nothing, yet it is exactly what a lead-based report will recommend. The dashboard improves, the applicant list lengthens, the vacancy loss continues unchanged, and the next review concludes that the campaign needs more money. That sequence repeats in a great many properties.

Ranking sources on certified move-ins and rent billed, with both the enquiry date and the move-in date exported, shows where the time actually goes. If the gap is uniform across every source, the constraint is the process and the budget should be left alone. If one source consistently produces households who complete faster, that is worth knowing and worth buying more of. Either answer is actionable, which is more than a cost per application has ever offered anybody running a lease-up against a deadline.