Branch closures get covered as an inevitable consequence of digital banking, but the framing skips a more interesting question: what specifically are branches still doing that digital channels haven't replicated, and are institutions measuring that at all?
Cost-per-transaction, the dominant metric used to justify closures, treats every branch visit as an equivalent, generic transaction. In practice, the customers still visiting branches are disproportionately dealing with the kind of problem that doesn't fit a standard digital workflow — a dispute, a life event, a situation where the customer isn't sure what they need and needs a conversation to find out.
Branches that handle a high proportion of these genuinely complex interactions look expensive under cost-per-transaction and look essential under a decision-authority lens — one that measures which customer problems actually got resolved, not just how many transactions were logged.
I've worked with institutions that reviewed closure decisions under this second metric and found branches flagged for closure that were, in fact, doing disproportionate work the network genuinely needed — work that would have transferred, badly, to call centres with less authority to resolve it on the spot.
The honest position isn't that branches should never close. It's that closure decisions made purely on transaction cost are measuring the wrong thing, and institutions that add a second metric before finalising closure lists tend to make noticeably better decisions.