Three agents called the same person. The report blamed all three!
I once watched three agents leave voicemail for the same person on the same afternoon. Same name. Same number. Three campaign records. I pulled all three call logs afterward, laid them side by side, and brought them to the team lead running the report that week.
“These three are all the same person,” I said.
She looked at the screen. “They’re not the same lead ID, though.”
“No, but it’s the same phone number. Same name. Same everything except which campaign it came in under.”
“So it’s a data thing, not a performance thing?”
“Right now, on paper, it’s three separate missed contacts for three separate agents.”
She was quiet for a second. “That doesn’t seem fair to them.”
“It isn’t. But nobody built anything that would’ve caught it before the calls went out.”
That is the Duplicate Parade: old records in new costumes marching through the pipeline while everyone applauds the volume.
I am retiring the fake percentage
The earlier draft of this piece cited broad duplicate-rate estimates as though they could describe a Medicare lead database. A colleague caught it before I did.
“Where’d the 30 percent come from?” she asked, reading over a draft.
“General B2B data quality research.”
“For Medicare leads specifically?”
“No. It’s just… a number that sounded right for the shape of the problem.”
“That’s not evidence. That’s a guess wearing a citation.”
She was right. Duplicate prevalence depends on sources, matching rules, time windows, campaign overlap, household structure, CRM behavior, and the definition of duplicate itself. I don’t need a universal percentage to prove a local problem. I need the local database.
Duplicates are not only identical rows
I explained this to a new analyst once, expecting it to be obvious. It wasn’t.
“I already deduplicated the list,” he told me. “Same name, same number — pulled all the exact matches.”
“What about the ones where she’s listed as ‘Bob’ in one record and ‘Robert’ in another? Same address, same phone, different spelling?”
He checked. “Those are still in there.”
“What about her husband, same household, different phone, calling about the same policy?”
“…I didn’t think about that one.”
Exact matching catches the easy ghosts. Normalized and probabilistic matching reveal the relatives.
Why agents inherit the blame
A contact-rate report rarely knows that another agent called the same person two hours earlier. I sat in on a coaching session once where this played out in real time.
“Your contact rate’s down 15 percent this week,” the manager said.
“I know. I don’t understand it. I’m working the list the same way I always do.”
“Maybe you need to warm up the intro a little more before you ask for the appointment.”
Nobody in that room had pulled the duplicate report. Nobody knew that four of her “no contacts” that week had already been dialed once or twice by someone else first. Without shared identity and call history, an input defect becomes a coaching conversation.
Consent does not improve through copying
A duplicate record may carry conflicting source, timestamp, seller, or permission information. I raised this with a vendor once after finding the same lead with two different consent timestamps three months apart.
“Which one’s correct?” I asked.
“Probably just take the most recent one,” they said.
“Why would the most recent one automatically be the accurate one? What if the original consent expired and this is just a re-scrape?”
Silence, then: “We’d have to look into that.”
That is not clerical clutter. It can change whether and how outreach should occur. The FTC’s guidance describes covered telemarketing duties involving Do Not Call procedures, calling practices, and written permission in specified circumstances. The right response is to preserve and resolve provenance — not simply keep whichever record has the newest date. [1]
The pre-routing gate
Before assignment, I normalize names, phones, emails, and addresses; match against active and historical records; establish a household rule; preserve the strongest provenance; suppress conflicts; and document why records were merged, retained, or blocked.
The first time I actually built this into a workflow, an agency owner asked me why it mattered.
“You’re basically making the list smaller before anyone even sees it.”
“I’m making it accurate before anyone sees it. Smaller is a side effect.”
“But we paid for those leads.”
“You paid for names on a spreadsheet. Whether they were ever actually reachable, valid, and un-duplicated is a separate question.”
Then I compare contact rate, complaints, agent time, qualification, and outcomes before and after the gate. That is how a duplicate problem becomes measurable instead of folkloric.
The FLS audit
Count repeated phones, emails, normalized names, and household addresses across active campaigns. I asked an operations director once how confident she was in her own numbers.
“Pretty confident,” she said.
“Have you actually counted the collisions? Across all your active campaigns?”
“No, but I’d know if it were a big problem.”
“How would you know, if nobody’s counted?”
She didn’t have an answer for that one. If nobody knows how many records collide, the finding is not that duplicates are rare. The finding is that the parade has no security checkpoint.
Continue the story
Next: Consent from the Crypt—when yesterday’s permission returns without its paperwork.
Sources and scope
[1] FTC, Complying with the Telemarketing Sales Rule
Published facts are cited and qualified. First-person observations, metaphors, and satire are commentary—not universal claims. This article is education, not legal, tax, compliance, financial, or insurance advice.
See what your own pipeline looks like once it’s actually scored. Start the audit at freshleadsolutions.com
Continue The Lead Pit series at blog.freshleadsolutions.com
Leave a Reply