The most interesting people in a selection are the ones you didn’t pick

The most interesting people in a selection are often the ones you didn’t select.

I’ve been building selections for the best part of twenty-five years. Sometimes everything falls neatly into place. Other times it feels like cutting the grass at Wembley with nail scissors.

But regardless of how smooth or painful the process is, one thing has always been true:

Selections are one of the best ways to surface issues in your data.

A quick explanation of how I build them. I start with a base query that includes everyone we could contact. It removes people with exclusion codes, no mailing address, no email address, or whatever other criteria makes sense for the activity.

From that large pool, I then create segments. Each segment is hierarchical and removed from the base one by one. So if John Smith qualifies for segments 1 and 5, he will only appear once because the data used in segment 1 has already been removed by the time we reach segment 5. The criteria for each segment stays consistent. It doesn’t get rewritten every time a new selection is needed. Changes can be made, but they should be driven by evidence – either performance or a coding issue.

All of that is fairly standard. The interesting part comes at the end.

Once all the segments have been accounted for, I wrap the base query around to the bottom of the selection and look at who is left. And it’s those leftovers that have the most to tell us.

They often highlight things like:

• People who clearly belong in one of the existing segments who are coded differently
• New patterns of behaviour that suggest an entirely new segment
• Records that should have been excluded but aren’t currently captured in the base exclusions

In other words, the people you didn’t select tell you just as much about your data as the ones you did.

Whenever I build a new selection, the first place I go after running it is the leftovers. They almost always reveal something interesting.

Selections aren’t just a way of building lists. They’re one of the simplest diagnostic tools you have for understanding how your data actually behaves.

I’m always curious – do people check the leftovers in their selections, or just the segments? This sort of thing is exactly what I help organisations untangle – small structural issues that quietly affect reporting and selections.

I’ve got some space from mid-May if it would be helpful to sense-check how your selections are working.
If someone comes to mind who’s a bit stuck with their data, I’d really appreciate an introduction.

#DataQuality
#CRM
#CharitySector
#DataGovernance
#Nonprofits