Building a mail list through a VA without checking the work in progress, and what a third garbage list costs

The 100 row sample is the fix, and it works for basically every VA deliverable. Accepting a finished anything without reviewing a first slice is the mistake to avoid. Sample first, review, then the rest.

Verbal criteria on a call is where this died. I'm still getting licensed and even I've learned that a spoken instruction to someone in a different time zone with an accent gap in both directions is going to lose about 15% of itself. Written, then read back.

Did the agency offer any kind of QC layer, or was it just a body and a login? I've been trying to understand what the markup actually buys and this feels like the test case.

Beacon, at $8.50/hr the markup buys recruiting and payroll. QC is a different price point.

The dedupe issue isn't a VA problem at all, it's a tooling problem, and that pattern shows up constantly in service work. Nobody should be deduping 1,900 rows by eye. That's a formula, it takes four minutes, and it should be a standing step in the SOP so it happens whether the person doing it is careful or not that day. If one document comes out of this whole mess, make it the checklist for what the file has to pass before it goes to print.

The three letters to one owner problem is a common one, and it shows up just as often on the property management side, say three renewal notices landing on one tenant. Same root cause. The list is keyed off the parcel when the human is the actual unit.

One thing worth checking before the next drop, and this varies by state, is whether the counties a list was pulled from actually publish what people assume they publish. Roughly a dozen states don't make sale prices public at all, and assessor mailing address fields go stale at different rates by county. A high undeliverable rate is often the source data rather than whoever built the list.

The point about splitting the bounce rate is the right one to make. Undeliverable rate on absentee mail often runs 8 to 12% even with a clean pull, so a high bounce count is high but the whole excess isn't process error either. Split the number and roughly half of it was probably always going to bounce, with the rest attributable to duplicates and wrong-county records. That remainder is the figure worth carrying forward as avoidable spend.

$3,500 to learn to inspect work in progress. Cheaper than most tuition, and dearer than reading one thread on list building first.

Sextant's split is the useful reframe. I'd add that the two weekends of cleanup are a real cost even though they never hit the card. If girder values his own night hours at anything, the total is closer to $2,000 avoidable, and that changes whether a $200 QC add-on from the agency was worth buying. It was.

Land lists run messier than most by nature, so this step earns its place. The one to add to your checklist, girder: pull a random 25 rows from the finished file and look them up yourself before print, every single time, even after the VA has been perfect for six months. Twenty minutes. That is the check that catches a bad county filter in month nine, on a VA anyone would have sworn was flawless.