A bank tells me they can't measure the campaign. I ask three questions and we almost never make it as far as the analytics.
That's not a measurement problem. It's a definition problem wearing a measurement problem's clothes.
Measurement is the last step in a chain. It reports back on decisions somebody made months ago: what counts as a conversion, where the customer record lives, whether anything got tagged before it went out the door. By the time you're staring at a dashboard wondering why nothing reconciles, the answer is upstream. Usually a long way upstream.
So before anybody builds you a report, four things have to be true.
1. A Conversion Has Exactly One Definition, and It's Written Down
Ask five people at a bank what counts as a conversion on a checking acquisition campaign. You'll get five answers. Application started. Application completed. Account opened. Account funded. Account funded and still open at ninety days.
All five are legitimate. Only one of them gets to be the number, and you pick it before the campaign runs, not after. Pick the one closest to revenue that you can actually observe. Funded and open at ninety days tells you something about the business. Applications started tells you your form works.
2. The Customer Exists Once
This is the one that quietly ruins everything else.
Somebody opens a checking account in 2019 as Robert. Refinances in 2022 as Bob. Comes through a merger and shows up a third time with a brand new customer number. Now marketing credits the response to one record, the core system holds the balance on another, and your campaign analysis is measuring three people who are one person.
You can't fix this in the report. Nobody can. The dedupe has to happen where the records live, and it has to happen before you mail anything. If you've been through a core system conversion in the last few years, assume this is broken until somebody proves otherwise.
3. It Gets Tagged Before It Ships
Direct mail gets a unique code. Digital gets consistent UTMs off a written convention, not whatever the person building the ad felt like typing that morning. The landing page passes both through to the CRM.
None of this is hard. All of it is impossible to add later. Once the mail is on the truck, whatever you didn't tag is gone for good.
4. One Person Owns the Number
A person, not a team.
The second two departments report the same metric two different ways, leadership stops trusting both versions, and now your measurement problem is a political problem. That's a much worse problem. I've watched it happen more than once, and it takes a lot longer to unwind than the data underneath it.
And Then There's Your Stack
Here's the number one thing on my end, and it's the one nobody wants to do. Look at your tools.
If you have more than three, you're not going to move the needle. Not because the tools are bad. Because every additional system is one more place for the customer record to fork, and because you're spending your week reconciling instead of marketing. One central database. One primary channel you actually invest in. One more tool for the thing your business genuinely requires.
Most institutions I meet are carrying eleven, and the eleven don't talk to each other. Each subscription looks like a hundred dollars a month on its own, which is why nobody questions any of them.
Where to Start
Agree on the definitions. Dedupe the file. Tag before you send. Put one name against the number. Then look at the stack.
If you only do one of those this quarter, dedupe the customer file. Everything else is easier once the same person stops appearing three times.
Do that and the measurement question mostly answers itself. And if you've done all of it and you still can't measure the campaign, then you've got a real measurement problem, and I'd like to hear about it, because those are rarer than people think.
Kelly Evans works with banks on the data and reporting infrastructure behind their marketing programs, including performance measurement and attribution.
If your campaign reports never quite reconcile, the report is rarely the problem. BKM works with banks on the data underneath the marketing, before anybody builds another dashboard.
Marketing Measurement for Banks FAQs
Usually because something upstream was never decided. Measurement is the last step in the chain and it reports back on choices somebody made months earlier: what counts as a conversion, where the customer record lives, whether anything got tagged before it shipped. If those were never settled, no report will reconcile.
Whichever definition you agree on before the campaign runs. Ask five people at a bank and you'll get five answers: application started, application completed, account opened, account funded, funded and still open at ninety days. All of them are legitimate. Pick the one closest to revenue that you can actually observe, and write it down.
Because records get created at different moments under different names. Somebody opens an account as Robert in 2019, refinances as Bob in 2022, then arrives a third time through a merger with a new customer number. The dedupe has to happen where the records live, before anything goes out the door. You can't fix it in the report.
Three, in most cases. One central database, one primary channel you genuinely invest in, and one more for whatever your business specifically requires. Every extra system is another place for the customer record to fork, and another hour spent reconciling instead of marketing.





SUBMIT YOUR COMMENT