Pay the branch for the account the customer keeps

I no longer believe a retail bank can make an individual product-count incentive safe by shortening the list, raising the quality threshold or adding monitoring. Each of those changes reduces harm. None of them changes what the plan tells a banker the job is. My position, after redesigning branch pay at two banks almost twenty years apart, is that branch colleagues should be paid as teams, from a pool funded by accounts customers are still using months after opening, and that the people who investigate sales misconduct should co-sign the plan before it is issued.

This is not an argument against paying branch staff incentives, nor a claim that every banker paid on products behaves badly. Most do not. It is an argument about design responsibility. When a plan creates a predictable pressure and a minority respond to it, the bank has chosen that outcome as surely as if it had written it into a procedure. The industry has had enough public lessons in the consequences of sales pressure, at banks of every size, that nobody designing a branch plan today can describe those consequences as unforeseeable.

What a points table actually instructs

In 2005, at Ravensholt Bank, I led the redesign of a consumer incentive plan that paid points for twenty-two products. A review had found secondary accounts that customers had not used and, in a small number of cases, did not recognise. We cut the table to nine products, required funding within sixty days and asked branch managers to review a sample of openings each month. Complaints about disputed accounts fell by more than half. I presented the result as a solution, and I was wrong to.

The shorter table still paid a banker for the next product, and it still paid nothing for an hour spent untangling a customer's problem. The quality reviews were carried out by managers whose own bonuses depended on the same openings. The funding rule produced a new habit: small deposits made at opening, sometimes at the banker's suggestion, to clear the sixty-day test. Very little of this was dishonest. It was adaptation. People do what a plan pays for, with whatever ingenuity the plan leaves room for, and a points table leaves a great deal of room.

When I arrived at Callowmere in 2021, I found a fourteen-product table and the same family of side effects. Our branches logged 312 complaints a year from customers who said they had not asked for a product, and 23% of new checking accounts closed within ninety days. Kelverdine, the bank we acquired in 2023, had paid its branches as teams for a decade. Its complaints about unrequested products ran at a fraction of ours, even allowing for a smaller network. That comparison did more to persuade our executive committee than anything I had said in the previous two years.

Count the account at ninety days, not at opening

The plan we adopted from July 2024 for 2,900 branch colleagues has three parts. A branch pool is funded by the branch's contribution and by the number of new accounts still active at ninety days, measured by customer-initiated transactions rather than balances the bank could move itself. The pool is shared across the team in proportion to salary, with a manager's discretion of up to fifteen per cent either way, recorded and reviewed. A smaller individual element rewards observed quality, scored from counter and call reviews by a team outside the branch's reporting line. A substantiated conduct finding removes the individual element for the year and is reported to the compensation committee in aggregate.

Ninety days is a judgement, not a scientific threshold. Thirty days is too easy to satisfy with an opening deposit, and a year is too far from the behaviour to influence it. We tested sixty, ninety and one hundred and twenty days against two years of historical openings and chose ninety because it separated used accounts from abandoned ones most cleanly in our data. Another bank's customers may behave differently, and it should run its own test rather than borrow our number.

The cost was real. In the first full year, new checking accounts fell 9%. Our head of consumer banking had predicted a larger fall and argued against the change in writing; his memo went to the committee with my proposal. Eleven of our top-decile producers left within six months, several of them to competitors whose plans still paid per product. Against that, the share of new accounts still active at ninety days rose from 61% to 78%, average balances on new accounts rose by about a third and complaints about unrequested products fell from 312 to 41. Branch incentive spending fell by US$2.4 million. I mention that because it surprised people, not because it was the purpose.

Give the second line a signature, not a review slot

Most banks already ask risk and compliance to review incentive plans. The federal banking agencies have expected for years that incentive arrangements be compatible with effective controls, and a review step is the usual answer. In my experience, a review that arrives after the plan is designed can object to features but cannot propose a different shape. The second line ends up negotiating thresholds with HR and the business, and the points table survives in a slightly safer form.

At Callowmere, any incentive plan for customer-facing staff now carries a design page signed by the head of conduct risk before it reaches the executive committee. The signature means the conduct team helped choose the measures, not merely that it found no fatal flaw. They brought something HR and the business lacked: their case files. They knew which behaviours had produced complaints, which measures people had found ways around and which teams generated a disproportionate share of findings. That knowledge is the most useful design input a bank has, and it usually sits in a different building from the people writing the plan.

There are objections I take seriously. Team pay can carry weaker performers and frustrate strong ones, which is why the manager's discretion exists and is recorded. Some customers do benefit from an additional product, and a banker who identifies that need deserves recognition; our quality reviewers score it when it happens appropriately. A competitor may hire away producers with a richer per-product plan. Some of ours went, and I would rather lose them than keep the plan that suited them.

I would not ask another bank to copy our design. I would ask its compensation committee to do three things. Read the branch plan as a teller would, line by line, and write down what it says to do on a busy afternoon. Ask the conduct team which of the bank's complaint themes the plan could make more likely. Then look at the accounts opened last year and count how many customers are still using them. If the answers are uncomfortable, a shorter points table will not fix them. I tried that once.