Which channels produced funded accounts and loans?

Application volume is the metric credit unions often report, and it rewards whichever channel produces the most people who will not qualify. Every one of those costs staff time and produces nothing.

Ranking on funded balances rather than applications makes the underwriting funnel visible in the marketing report, and it routinely shows that the cheapest applications come from the most expensive channel.

A member who joins for an auto loan may hold a mortgage, a card and a deposit relationship for twenty years. Judged on the first loan, the channel that produced them is undervalued many times over.

Because the match is on the member, every later product credits the original channel, which is the only basis on which member acquisition spend can honestly be compared to product-level campaigns.

For many credit unions the largest single source of new members is the finance office of a car dealership. A member joins because a dealer sent the paperwork across, not because they saw an advertisement, and the relationship begins with a loan and a membership the marketing team had no hand in at all. That volume sits inside the same funded-loan totals every campaign in the report is being measured against. Every channel in the report is therefore being compared against a number it had nothing to do with producing.

Leaving it in distorts every channel figure on the page. Separating it out lets the indirect book be read on its own terms and lets the direct campaigns be judged against a total they actually influenced: either leave dealer-originated loans out of the funded loan export, or map an origination-source column as Product / service so indirect and direct show as separate lines, each with how much of it traces to a channel. Both numbers become usable at the same moment, and the comparison between the two can be the more interesting half of the exercise. Indirect volume is a business decision in its own right and deserves to be read as one, rather than as noise sitting on top of the marketing figures.

It also frames the question the credit union genuinely has about indirect members, which is whether any of them ever take a second product. A member acquired through a dealership who opens a checking account and a card two years later is worth a great deal more than one who pays off an auto loan and disappears. Because later products credit the member's original source, that answer arrives without anybody commissioning a separate piece of analysis. The answer decides whether indirect is an acquisition channel or a lending channel, which are different things to fund.

A credit union can only serve the people its charter allows, which means advertising outside that footprint is money spent against an audience that cannot join whatever it does next. Broad digital campaigns are particularly prone to this: the targeting is approximate, the impressions are cheap, and inquiries arrive from people who are ineligible and will therefore never appear in a funded loan file at all. Those inquiries still cost money to generate and still take staff time to turn away politely, and nothing in the advertising report distinguishes them from a good one.

Ranking channels on funded balances makes the waste visible without anyone auditing a targeting setting. A channel producing inquiries that never fund shows leads in the lead file and nothing in the revenue column, which is a more convincing argument for tightening geography than any debate about radius settings has ever been. A branch column mapped as Location adds the footprint: funded balances by branch, with how much of each traces to a channel, though not a ranking of channels within each branch. It turns a settings argument into a revenue observation, which tends to end the discussion rather faster.

The branch reading earns its place for a second reason. A branch where most funded balances sit in Direct / Unknown is a part of the footprint the digital channels are barely reaching, and a branch where a large share traces to a channel is one where they are. Funding a full-footprint budget on a blended result is how a marketing plan ends up working hardest in the places that needed the help least, and the attributed share by branch is a first, honest check on that. Branch-level reading is also how a credit union finds the market where it is nearly invisible.

A certificate special, a balance transfer offer or a promotional auto rate will move volume on its own. Whatever advertising happens to be running alongside it gets credited with the response, and when the promotion ends those same channels appear to collapse. This is the measurement trap specific to a rate-led institution, and it produces confident conclusions about creative that were really conclusions about pricing all along. The offer did the work, the campaign took the credit, and next year's budget gets written accordingly, which is how a pricing decision becomes a permanent line in the media plan.

Reading the period against the same period a year earlier, and giving the promotion its own campaign name in the lead file, keeps the two apart. A promotion lifts everything at once, whereas a channel that genuinely improved changes position relative to the others. The loan count beside the funded balance adds a further check, because a promotion tends to show up as more loans at a similar size rather than as better loans from the same number of members. Neither check needs anything the lead file does not already carry; both are choices about how the existing report is read.

History makes this practical rather than theoretical. Because nothing had to be running at the time, a credit union can upload several years of funded loans and past campaign lists on its first afternoon and look at exactly what happened around each previous promotion. That is a far sounder basis for pricing the next one than the collective recollection of how the last one felt while it was live. Pricing a special against what the last three actually produced is a stronger position than pricing it against instinct, and the records have been sitting in the core system the whole time.