Two questions you could only answer by opening records one at a time. Both got answered in a single pass, and one of the answers turned into something permanent.
/ The problem
Two questions came up constantly, and neither had a quick answer.
Who has gone quiet? Clients who stop keeping pace are usually the ones about to leave, and they leave without saying so first.
Whose finished work can we safely reuse? Strong examples of past work are the most useful thing a team can show a new client. But not every finished client's work is right to reuse, and figuring out which ones meant checking each person's record by hand.
Both meant opening records one at a time, which took long enough that it mostly didn't happen. The team ran on memory of who seemed behind.
/ What we did
We pulled both pictures out of the client records in one pass, and turned them into lists someone could work from right away.
For the clients falling behind, the list gave each person's contact details, who on the team was assigned to them, and a direct link to their file. It came in two versions: a full one for reviewing the whole picture, and a short one with just the links, for someone working through calls.
For the finished clients, we screened everyone who had completed the service inside the target window against three more conditions, with each answer already worked out in the file: is their finished work actually on file, were they refunded, and are they still looking or have they already landed a role.
89% cleared all four. The two checks that did the real filtering were the simple ones: about one in twenty had no finished work on file, and a similar share had been refunded. People who had already landed a role were held out on purpose, not counted as failures. The point of the screen was finding work that was right to reuse, not measuring how the service performed.
Data work, no interface · one pass, four checks
records
four checks
lists
library
89%
cleared all four checks
1
pass, and it never needed doing again
/ What changed
The team worked a real follow-up list instead of trying to remember who seemed behind.
The screening turned a question everyone avoided, which finished work is safe to reuse, into a settled answer in one pass.
That vetted batch became the starting set for an anonymized library of past client work, which now holds several hundred examples and takes in new ones on its own every day. A job done by hand became the seed for one that never needs doing again.
/ Decisions worth explaining
Each row already said yes or no, so anyone could act on the list without going back to the original records to work it out again.
It answers "is this work right to reuse," and nothing else. It would be easy to read the cleared count as a success rate. It isn't one, and the file says so.
Reviewing the whole picture and working through a call list are different tasks. One file trying to serve both would have been worse at each.