Zone whitelists and the kill rules

Pop networks sell you a firehose sorted by zone ID, and the zone ID is the only lever that matters. GEO targeting, OS splits, and dayparting tune the firehose; the whitelist decides whether you are buying from the 63 sources that convert or the 1,337 that merely exist. This chapter is the exact procedure we run on every account, in the order we run it.

Rule 1: fund every zone to exactly 3× payout

A test that cannot conclude is a donation. Each zone in a sweep gets a budget of three times the offer payout and not a dollar more: a $10 payout means $30 of test spend per zone, a $1.20 sweeps payout means $3.60. Below 3× you kill winners on variance; much above it you fund losers out of politeness. The number is not sacred, it is a working compromise we have used since March 2025, and the discipline of applying it evenly matters more than the multiplier itself.

Sweep order follows cost of evidence. Tier-1 CPCs run two to five times Tier-2/3 levels, so when an offer spans tiers we sweep the cheap GEOs first and use surviving zone patterns to shortlist the expensive ones. The VPN rebuild ran Germany and Poland before touching US inventory for exactly this reason.

Rule 2: a zone dies on 3× spend with zero quality conversions

The kill rule is binary and automated: once a zone has spent its 3× budget without one quality conversion, it goes to the blacklist the same day. No intuition, no "it feels close." The corollary matters as much: a zone is never killed before its budget concludes, because half-run tests produce whitelists full of lucky zones and graveyards full of slandered ones.

"Quality conversion" is doing the heavy lifting in that sentence. The judging event is always the deepest advertiser-side signal we can get passed back: a confirmed lead, a trial start, a rebill. Raw form fills flatter bots; the median new account we audit shows 27% of conversions failing quality checks, and judging zones on raw conversions would have kept most of those sources alive.

Rule 3: survivors earn bids one rung at a time

Inside the whitelist, bids climb in $0.05 CPM steps from the network floor ($0.10 on PropellerAds pops, $0.50 on RichAds) while the account CPA holds under goal. One rung per zone per review cycle, never chased during a bad day. Frequency caps start at 1 impression per user per 24 hours and loosen only when a zone proves it converts on repeat exposure, which most never do.

Rule 4: re-audit the whitelist monthly

Zones rot. A source that converted cleanly in March can be reselling junk by June, so every whitelisted zone re-earns its place monthly against the same quality thresholds, and the scoring model flags time-to-conversion and device-mix drift between cycles. Expect 5–10% of a mature whitelist to churn out each month; a whitelist with zero churn usually means nobody is looking.

Rule 5: the whitelist belongs to the client

Full zone lists, whitelist and blacklist, ship inside every weekly ledger as plain CSVs. An agency that keeps the whitelist hostage has an incentive we do not want: making itself hard to fire matters more than making the account good. Ours is the opposite bet. If the lists are yours and the rules are public, the only reason to stay is that the numbers keep working.

The whole chapter as a checklist

1. Define the quality event with the advertiser before spend moves.
2. Fund every zone to exactly 3× payout; sweep cheap GEOs first.
3. Kill on 3× spend with zero quality conversions, same day, automated.
4. Never kill early; never extend a losing zone's budget.
5. Climb bids $0.05 per rung inside the whitelist while CPA holds.
6. Cap frequency at 1/24h until a zone proves otherwise.
7. Re-audit monthly; expect 5–10% churn.
8. Ship both lists to the client weekly.