.jpg)
Like clockwork, every late summer and early fall, a new list shows up. Is AI still the hot category to watch? Or is it fraud, or customer experience enhancement? The lists are ever-changing, typically tied to whichever had the loudest conference buzz this year, but the format isn’t. You pick the right category and the budget decision is made.
Or so one may think. Those who’ve been around the budgeting block know that the category has never been the specific thing that determined whether last year’s spend paid off. Two banks can buy the same type of tool and have two completely different experiences. Where the difference lies is which bank used the tool to make everything else in their stack more useful, rather than buying something that would sit in isolation doing one job.
Take fraud tools for example. One might flag a transaction and stop there, creating a closed-loop that only the fraud team ever sees. Another tool might feed that same signal into the same data layer that the bank’s marketing and onboarding functions already pull from, resulting in a sharpened opportunity for outreach and onboarding risk scoring. While both may be coming out of the same budget category, one purchase compounds the returns, while the other doesn’t.
AI-enabled tools have the same type of split. One point solution that’s bolted onto a single team’s workflow may do that one thing reasonably well, and nothing else. On the other hand, a tool built on the bank’s own core data becomes more useful the more of the bank it touches, with each department feeding data to make the next department’s output better. A label of “AI” really tells you nothing about which of those you’re buying.
Digital onboarding tools also follow this same pattern. A tool to funnel an application from step one to step five and stop does its job narrowly. One that hands off clean, structured data to everything downstream (marketing follow-up, fraud screenings, servicing) does the same job while making three other jobs easier simultaneously.
None of these purchases are wrong, per se. Where the mistake occurs is choosing to spend based on a category name instead of mapping out what happens to that data and capability once the purchase and install are complete.
There’s a deeper question to consider underneath the hype of the next shiny new piece of technology, and it’s one to understand before you add any new line items.
Can the bank show exactly what last year’s budget did?
Think beyond just the general terms, like whether your fraud tool “looks to be working”, be highly specific. Calculate which spend items lowered the bank's cost per funded account. Did any shorten time to resolution, and by how much? Also note which ones nobody can point to concrete numbers for, even a year later.
If your answers aren’t clear, plugging in a new type of technology each year won’t fix any of your problems or drive any of your growth initiatives. Instead they will continue to add new lines to a stack of unmeasurable items. Clear visibility into what’s already been spent must come before the conversations about what to spend on next, or the same cycle repeats each budget season.
Skip any trending, ranked lists, and ask two things: 1) does this purchase make the rest of our stack work harder, and 2) can the bank clearly articulate and prove what last year’s spend did to drive goals. By answering both, you can walk into next year’s budget meeting with real numbers, tying to the balance sheet, instead of just a hot topics list.
It’s the same principle behind how we built Linker Finance: modular enough to work with the core a bank is already running, and connected enough that what one team learns strengthens what the next team does. Every dollar ties back to deposits per funded account, not only accounts opened, so your budget is always ready and defendable.
If your stack still isn’t tying the pieces together, that’s worth a conversation before the budget gets finalized.