Shoppers submit through several channels with different emails and sometimes a partner's phone number, so matching on one field undercounts duplicates badly. A workable dedupe rule matches on phone, then email, then name plus ZIP, and merges rather than deletes so the full contact history survives.
Why single field matching fails
Cox Automotive shopper research has long shown buyers moving between marketplaces, store sites, and third party tools before visiting. Each stop can leave a lead with different contact details: a work email on the marketplace form, a personal email on the store site, a mobile number in one and a household number in the other. Match only on phone and the two records sit side by side as separate opportunities.
The cost is not just a messy CRM. Two salespeople work the same shopper, response time reporting counts both leads, and close rates divide by an inflated denominator. Management then judges channels on lead volume that double counts the stores best shoppers, the ones who raised their hand twice.
A merge rule that keeps the history
Run matching in passes. Exact phone match first, then exact email, then a softer pass on normalized name plus ZIP or the VIN of the vehicle inquired about. Each pass merges records into the oldest one and keeps every submission as a contact event with its source and timestamp. Deleting duplicates throws away the channel evidence you need to know which source produced the visit.
Set a human review queue for the soft pass. Name plus ZIP will occasionally catch two different people in one household, which is a real case in car buying, not an error to auto merge. The queue stays small because the exact passes clear most volume before it.
Related reading
Duplicates pollute the same reports that chat bot fake leads do, and both problems get worse when nobody owns lead hygiene. The feed accuracy piece on demo and loaner units in listings covers the inventory side of the same data discipline. A store that audits what it sends out and what comes back in makes better decisions from the same CRM everyone else complains about.
After dedupe, restate your channel reports once and say so. Close rates will jump on paper because the denominator shrank. If leadership is not warned, the improvement gets credited to a campaign instead of the cleanup, and the cleanup loses its budget.
Measure duplicates as a rate
Track merged records as a percentage of total submissions each month, by source. A rising duplicate rate from one marketplace usually means that source is recycling shoppers or your forms are failing to recognize returning visitors. Either way it is a signal about the source, not just a data chore.
The goal is not zero. Some duplication is the natural shape of a considered purchase where the shopper touches three channels. The goal is a known rate, merged records, and reports that count people instead of form submissions.
Frequently asked questions
What fields should lead deduplication match on?
Phone and email as exact passes first, then a softer pass on normalized name plus ZIP or the vehicle VIN inquired about. Run the soft pass into a human review queue rather than auto merging.
Should duplicate lead records be deleted?
No. Merge them into the oldest record and keep each submission as a sourced, timestamped contact event. Deleting loses the channel history that tells you which source produced the store visit.
Why do duplicate leads distort channel decisions?
They inflate lead counts and deflate close rates unevenly across sources, so channels that attract multi touch shoppers look worse than they are while single touch sources look cleaner.
How often should deduplication run?
Nightly is enough for most stores, timed after marketplace lead imports land. The duplicate rate by source should be reviewed monthly alongside close rate reporting.