Resold skip trace enrichment for six months, 41% match rate ate the whole thing

Shutting this one down, so here's what it cost.

The idea: wholesalers I already did marketing work for kept complaining that their phone numbers were bad. I had a decent pipeline of them, so I built a thin enrichment product. Upload a CSV of addresses, get back owner name, mailing address, and up to three phone numbers, scored. $99 a month for 500 lookups, $0.14 each after that. I wasn't the data source. I bought wholesale from two providers and stitched them together, which was the actual product (dedupe, phone scoring, and a clean CSV out).

Spend: $27,400 all in. $19,000 to a contract developer over four months, $3,800 in data credits during testing, the rest on the site and a lawyer for the terms.

Peak was 14 paying accounts, so $1,386 MRR. Then churn. Monthly churn ran 20 to 24% for the last three months. Every exit interview said the same thing in different words: the numbers didn't connect.

So I measured it properly, which I should have done in week two. Across 6,100 lookups my providers returned at least one number on 63% of records. Of those numbers, the ones my customers actually reached a human on came out around 41% of records. My competitors were sitting somewhere similar, but they were priced so that 41% still worked for the buyer. At my price it didn't, because the buyer was comparing me to his old provider and I hadn't made a single record better, I'd only made the CSV prettier.

What I'd do differently: buy 1,000 records from three providers first, hand-dial 200 of them myself, and only write code if the stitched match rate beat the best single provider by ten points or more. That test costs about $400 and two weekends. I spent $27,400 to learn it.

2 replies

1laurenkim53Passive

Match rate is worse than 41% out where I work. Rural counties, half the owner records are a trust or an estate and the mailing address is a PO box two states away. Any product priced on "we'll find the owner's cell" falls apart the moment the county stops being suburban. I'd have wanted your CSV cleaner for a different reason entirely, which is that nobody sells a decent parser for handwritten-era records. Probably a market of eleven people though.

1mikeoconnell7SP2APRO6

thatch, I don't think the match rate was the real cause. You were a reseller of two upstream providers, which means your ceiling was always their data and your only defensible piece was formatting. 20% monthly churn on a $99 tool usually means the buyer never had a switching cost. No workflow lived inside your product, no history, no team, no notes. They uploaded a CSV and downloaded a CSV, so leaving cost them nothing.

If the phone numbers had been 60% you'd have bought yourself maybe another six months and the same ending. The version that survives is the one where the calling actually happens in your tool and the outcomes get recorded, because then their eighteen months of dispositions are hostage. That's a much bigger build than $19,000, which is its own argument for not starting.

Your $400 test is the right test. I'd add one more: ask three of them to prepay a year at a discount before you write any code. If none of them will, you've learned the same thing cheaper.

Resold skip trace enrichment for six months, 41% match rate ate the whole thing

2 replies