Professional experience brief 03 of 04 · Data reconciliation
This example comes from Michael’s prior professional work. It is not a Next Concepts client engagement.
90-second storyReferral Matcher
Chapter 1 of 6: The situation
Use ← → keys
The work happened. The record did not.
I reconnected completed work the reporting process had lost.
Property teams referred interested residents, residents later purchased service, and the sales were real. But unless someone manually connected each sale back to its referral, the leasing partner and field team could lose the credit.
Open referralsWireless salesInternet sales
CompareName · unit · date
Possible matchReady for human verification✓
90-day windowA referral could remain valid long after it left someone’s immediate memory.3 data streamsOpen referrals, wireless sales, and home-internet sales had to be compared.
The process expected people to remember for 90 days
Sales and referrals lived in different systems.
A referral remained valid for 90 days—long after it had disappeared from someone’s immediate memory. Employees had to compare separate exports, retrieve order details, and manually close the loop.
Names could be misspelled, a spouse could complete the purchase, or the unit might match when the resident name did not. I needed to find credible connections without pretending every suggested match was certain.
The match had to be useful—not artificially perfect
What existedImperfect recordsWhat I sawMultiple supporting signalsWhat was neededHuman-verified matches
I narrowed the evidence. A person made the decision.
I created a focused list of likely connections for verification.
I built a workflow that considered property, unit, name, timing, and sequence. It compared open referrals with wireless and home-internet sales, then surfaced the most credible candidates for a manager or employee to verify.
See how the matcher handled uncertainty
01Connect the evidence
The workflow compared property, unit, names, dates, and whether the sale occurred after the referral.
02Allow credible variation
Close names, household members, and matching units could remain candidates instead of being discarded by exact-match rules.
03Keep a person in control
The output recommended possible matches. A manager or employee still verified the relationship before closing the referral.
Open referralsWireless salesInternet sales
CompareName · unit · date
Possible matchReady for human verification✓
01Load the open referrals02Compare the sales evidence03Verify and close the match
Missing connections became visible
Managers could work a focused exception list instead of comparing raw exports.
Initial team-level reviews surfaced dozens of completed sales needing referral verification. The tool did not create new activity; it helped the organization recognize work that had already happened.
The operating outcome
90-day windowA referral could remain valid long after it left someone’s immediate memory.3 data streamsOpen referrals, wireless sales, and home-internet sales had to be compared.Team-level useDirectors and managers could review activity across an entire team.Dozens surfacedEarly reviews exposed completed sales that still needed referral verification.
What this reveals about my work now
A performance problem may be a visibility problem.
Before I tell an owner that people need to work harder, I examine how the work is recorded and measured. I trace the evidence from beginning to end to find where ownership, reporting, or disconnected systems may be losing it.
