Which channels opened accounts that fund?

A meaningful share of commercial banking relationships still begin with a call to a branch, a conversation at a chamber event, or an introduction from an existing client. Digital attribution treats all of that as untracked, which means it treats most of the book as invisible.

Matching on the business contact rather than a browser session pulls those relationships back into the picture where a lead record exists for them, and leaves the genuinely untraceable ones in Direct / Unknown rather than crediting them to whichever campaign happened to be running.

A funded commercial relationship generates fee and deposit income for years. Judging acquisition spend on the first year systematically underprices whichever channel brings relationships that stay.

Because the match is on the client and there is no attribution window, running the report across several years credits later revenue back to the channel that produced the original inquiry. That is usually the analysis that justifies a materially larger acquisition budget.

A treasury campaign aimed at operating balances and a lending campaign aimed at owners looking for a term facility compete for the same marketing money and are judged against the same blended figure. They behave nothing alike. Deposit relationships open quickly, earn through balances and fees, and are sensitive to rate. Credit relationships take months, involve underwriting, produce one large number when they fund and produce nothing at all when they are declined halfway through. Averaging two funnels with those shapes produces a cost per relationship that belongs to neither of them.

A product column mapped as Product / service separates them in the revenue breakdown, with how much of each product traces to a channel. A channel producing a steady flow of small operating accounts can be the best source the bank has while looking mediocre next to a channel that happened to produce two funded facilities in the same quarter. Reading the two together is how a bank concludes that a campaign failed when it in fact succeeded precisely at the thing it was aimed at doing. The column costs one field at export and settles an argument the bank has been having every quarter.

It also puts the decline rate where it belongs. Applications that never fund appear as unmatched leads and cost nothing against your allowance, so a channel producing volume that underwriting rejects shows up as inquiry count with no revenue behind it. That is a genuine cost in credit analysis, staff time and an applicant's experience of the bank, and it is entirely invisible in a report that treats the application itself as the outcome being bought. A channel can therefore be expensive in a way that never appears on any marketing invoice at all.

The operating account is the beginning rather than the end. Merchant services, payroll, positive pay, lockbox, commercial cards, foreign exchange, an equipment facility two years later: most of the fee income a commercial relationship produces is sold by a relationship manager to a client the bank already has. None of it is attributed to anything, because by the time it happens the marketing that opened the door is several reporting cycles into the past. The result is that the most profitable part of the relationship is credited to nobody in particular.

Matching on the business contact means each of those later revenue lines credits the channel that produced the relationship in the first place. The acquisition question changes from what a funded account is worth in its first year into what the relationship has produced since it opened, and the answer usually justifies materially more spend on whichever channel brings clients who go on to take a second and a third product from the bank. It is a different budget conversation, and it starts from what the bank actually collected.

It exposes the reverse case as well, which banks rarely quantify. A channel can produce accounts that open, sit at a low balance and never take another product. On a funded-account count that channel performs well; on fee income across three years it performs badly. Both facts come out of the same two files, in the same report, and only one of them is currently on anybody's dashboard or in anybody's quarterly marketing review. Knowing which channels produce dormant relationships is worth as much as knowing which produce active ones.

Deposit gathering is rate-sensitive in a way that makes marketing measurement genuinely difficult. Move a rate and volume arrives, and whatever campaign was in market at the time is credited with a response the rate produced. Cut the rate and volume disappears while the campaign takes the blame. This is the specific measurement trap in commercial banking, and no attribution method resolves it on its own, including this one. What it does do is stop the report from quietly converting a treasury decision into a marketing conclusion, which is where most of the damage is done.

What helps is comparing the same period a year apart rather than consecutive months, ranking channels on revenue rather than on applications, and reading the relationship count beside the money. A rate-driven surge lifts every channel at once and appears as a level shift across the whole report. A channel that genuinely improved moves relative to the others, and a month-on-month comparison cannot tell those two situations apart at all. Reading the report the wrong way round here is the most expensive mistake available to a bank marketing team, because it usually ends with the wrong channel being cut.

Because there is no attribution window and nothing had to be running at the time, historical comparisons are available immediately rather than after a year of waiting. A bank can upload three years of funded relationships and lead exports on its first afternoon and look at how each channel behaved through a full rate cycle, which is a considerably better basis for setting a budget than the single quarter that has just finished. It is also the fastest way to find out whether the exercise tells the bank anything it did not already suspect.