G4: Retargeting Leak — Acquisition Spend Hitting Past Buyers
Data as of 17 seconds ago
Recoveries shown on this page are alternative paths, not additive to other lever pages. Observable March 2026 gap to break-even ≈ 90k KES/mo. Stack levers carefully — claimed totals across G1+G2+G3+G4 exceed the observable gap. Pick highest-confidence move first.
Active Acquisition Adsets
Without Purchaser Exclusion
Unsuppressed Spend Share (90d, w11)
Customer Files Wired as Exclusion
ZERO of 90d Meta spend hits customer-list audiences. 100% is cold acquisition — 51.6% with no exclusion at all. The
retargeting_repeatbucket (adsets using customer-file custom audiences) has 0 adsets and 0 KES spend.cold_broadalone accounts for 51.6% of all 90d adset spend (978,072 KES).MaxValue_Local_Allalone = 903k KES (47.7% of all adset spend), zero exclusions, Advantage+ Audience enabled. At industry-typical 15–25% past-buyer reach within unsuppressed acquisition: ~50–80k KES/mo modeled bleed to past buyers — alternative not additive to other levers.
Sources: w11 adset_insights × w1 adset_targeting cross-join. See glossary.
Spend by audience bucket (90d)
All 90d Meta spend falls into three cold-acquisition buckets. The retargeting_repeat bucket — adsets wired to customer-file custom audiences — has zero adsets and zero spend. Two customer-file audiences exist in the account but are wired to nothing.
Retargeting Bucket Breakdown — 90d Adset Spend
Which adsets are leaking?
Of 19 active adsets, 9 have zero purchaser audience in their excluded_audiences_ids. Three are retargeting adsets (RT_*) intentionally targeting past engagers — they are excluded from this list. The 9 below are all prospecting / acquisition campaigns.
Acquisition Adsets With No Purchaser Exclusion
Website Traffic ad set has one exclusion (Website View Content Last 14 Days) but that audience is site visitors, not purchasers. It is included here because purchasers within the view-content cohort are still reachable — no purchaser signal suppresses them.
All 19 active adsets have Advantage+ Audience enabled. This means Meta's automation can override manual targeting parameters — including exclusions — if it judges past buyers more likely to convert. Advantage+ does not guarantee that manual exclusions are honoured.
For comparison, the 10 adsets that DO exclude purchasers, and their exclusion windows:
Adsets With Purchaser Exclusions (reference)
Even the "protected" adsets have narrow windows — mostly 7–14 days. A buyer from 30 days ago is reachable again. Only LAL_AllEvents_5% adset uses a 180-day window. No adset uses a 180-day purchaser exclusion on the local/broad acquisition campaigns. The suppression perimeter is thin even where it exists.
Claims: G4-01 (9 of 19 acquisition adsets have zero purchaser exclusion), G4-02 (all 19 adsets have Advantage+ Audience enabled — exclusion enforcement uncertain)
Per-adset spend detail
Live spend from meta_metadata__adset_insights joined to meta_metadata__adset_targeting on adset_id. Sorted by 90d spend descending to surface highest-leaking adsets first.
Per-Adset 90d Spend — Targeting × Insights Join
Customer file upload status
Two CUSTOM (customer-file) audiences exist in the account. Neither is wired as an exclusion on any active adset. Both exist but are idle — they cannot suppress email-identified past buyers on any current campaign.
Customer-File Audiences — Status
Neither customer file is wired as an exclusion on any adset. Cross-referencing all 34 adsets' excluded_audiences_ids against both customer-file audience IDs returns zero matches. Past buyers identifiable by email are invisible to Meta's exclusion engine on acquisition campaigns.
Additionally, both uploads are stale — the newer one is 10 months old. Buyers who made their first purchase in the last 10 months are not represented in the customer list at all, even if the list were wired correctly.
Claims: G4-03 (0 of 2 customer files wired as exclusion), G4-04 (customer files are 10–18 months stale)
What to fix and how
G4 Fix Actions — Ordered by Priority
Why this matters
At 1.10x ROAS in April 2026, every impression served to a past buyer on an acquisition campaign is a misfired dollar. Past-buyer impressions in an acquisition context do one of two things: upsell (rare without a dedicated upsell creative and bid strategy) or waste reach that should have gone to a net-new prospect.
This range is modeled — not measured. The 15–25% figure is an industry benchmark for unexcluded acquisition campaigns. Actual bleed depends on: (a) how many past buyers are in Meta's matched pool for each adset's targeting, (b) whether Advantage+ Audience is actively re-serving past buyers, and (c) creative relevance to past buyers vs new audiences. The only way to measure it precisely is with the w11 insights worker + Meta Audience Overlap API. The range is plausible and directionally correct; the exact figure requires data.
The ~570k KES April spend figure is an extrapolation from available spend signals across campaigns — it is an approximation pending the w11 pull.
Comparison to break-even gap: The observable March 2026 shortfall to break-even is approximately 90k KES/mo. At the mid-range bleed estimate (20%), the retargeting leak alone (~114k KES/mo) would theoretically close the gap — but this is additive to G1+G2+G3 recoveries. Do not stack all lever recoveries; the 90k gap is a single ceiling. Fixing the retargeting leak does not create 114k of new profit — it redirects budget to higher-probability new-customer impressions, improving ROAS on the acquisition campaigns where the fix is applied.