By Dave LaRoche, Managing Partner, U.S. Operations, Predictive Analytics Group
Part 1 of two. [Part 2: Why Most Analytics Projects Quietly Languish]
You do not need mature data to work with us. You need a reason that will not wait.
A chief risk officer at a mid-market lender sat in a board meeting and got a question she could not answer. Why was one auto vintage charging off faster than the book said it should?
She had the signals. They were buried three systems deep, and a warning light buried behind the dash does not slow the car. Her team reconciled the numbers by hand, and by the time they agreed on a figure, the quarter had closed and the loss was booked.
She said something on her first call with me that I have heard, almost word for word, more times than I can count. “The average still looks fine. That’s what scares me.”
That is when we do our best work. Not for the shop with the cleanest data, but for the one that has stopped trusting its own numbers and has a decision on the calendar.
The moment matters more than the maturity
Most firms in our line of work sell to a maturity level. They want the company that already has the warehouse, the governance, and the team, and just needs another set of hands.
We look for a moment instead.
A board question nobody could answer. A merger that doubled the systems and cut confidence in half. An exam notice on the calendar. A loss creeping up in one segment while the portfolio average stays calm. Those are the weeks when analysis stops being a line item and starts costing real money. That’s when the work takes hold, because the whole room already agrees something must change. And some of our strongest engagements started before anything broke, with a client who could see it coming and decided not to learn the hard way.
The maturity question tends to sort itself out. We have built this for big banks and stood the same thing up for mid-market lenders who needed it running yesterday. What we look for has nothing to do with the tech stack. Something must be on the line.
Everyone has a week like this
One of these usually lands. Each is a problem we have helped resolve, linked where it fits.
- The average looks fine and a segment underneath it does not. Your scorecards were calibrated for a different rate environment, and a cohort is deteriorating faster than the model assumes. You find out when a board member asks for the next cut. (We covered the warning signs in Three Signs Your Underwriting Has Quietly Drifted from Reality.)
- Finance, Risk, and Operations walk into the same meeting with three different numbers for one metric. Everybody made a reasonable assumption about what “30-day delinquency” includes, nobody wrote it down, and now every steering committee opens by arguing whose data is right. That is the whole subject of The Dashboard Was Never the Problem.
- Your collections shop is still built for a customer who picks up the phone, opens the mail, and remembers who they bank with. Roll rates are creeping, another dialer did not fix it, and adding headcount has stopped paying. Your Collections Strategy Is Really an Operating Model Strategy goes deeper on that.
- The board wants an AI story, and your data is scattered across legacy systems that cannot agree on a number. We wrote about that foundation problem in Your AI Strategy Has a Data Problem.
- You grew by acquisition. Somewhere in the building sits a model nobody owns, still in production, built by people who left two years ago.
What these institutions share
Plenty of banks carry drifting scorecards and messy reporting for years and never pick up the phone. What our strongest engagements have in common is narrower than any one symptom: a cost the client can already name, and a date it comes due.
The CRO from the top of this piece had both. She could put a dollar figure on the vintage that slipped, and she had a board meeting on the calendar where that number would surface whether she was ready or not. A real cost and a hard date. That is what turns analysis from a project into a priority, and once it exists, nobody in the room has to be talked into caring.
We’re not a good fit for everyone
Some institutions just want the problem to go away, and recognizing these early saves everyone time and effort.
If you want a single report and then silence, we are the wrong call. Our work compounds, and a one-time extract does not. If price is the only variable that matters, someone will always be cheaper. If what you want is a one-off deck to wave at the board so it looks like progress, there are probably better alternatives.
We build reporting you can defend. If you are not willing to change a decision when the data contradicts something you believe about your best segment, no platform will help you, ours included.
None of that is a knock on anyone. It is a filter. The executives who read that list and bristle usually know already that this is not their moment.
If you saw your own board meeting up there
The people who call us are ready. What they lack is time, because something on the calendar is forcing the question. If you recognized your own week somewhere in here, the useful next step is knowing what a partnership like this asks of you, because it does ask something real. I wrote that up in the second part of this series [Why Most Analytics Projects Quietly Languish].
And if your teams are still walking into meetings with three numbers for the same metric, that is the conversation worth having.
Dave LaRoche is Managing Partner, U.S. Operations, for Predictive Analytics Group, which helps clients cost-effectively bridge the gap between analysis and impactful decisions. To see whether your data environment is ready for what’s coming, schedule a discovery call.
