Appendix D — Audiences, Opportunity Score, Anomalies, Auction Benchmarks

Live API pull, June 4, 2026.

Custom Audiences inventory (38 total)

Active audiences (10)

Inactive audiences (~28 broken placeholder LALs)

ALL 28 inactive lookalikes are sized at exactly 1,000 people — Meta's signature for a "broken" / orphaned lookalike (seed audience deleted or invalid). Examples:

• Lookalike (Worldwide, 1%) - IG Mil.com Account Engagers (365D) — 1,000
• Lookalike (Worldwide, 1%) - FB Mil.com Page Engagers (365D) — 1,000
• Lookalike (US, 10%) - All Site Visitors (180D) — 1,000
• (25 more similar — all 1,000 size)

These map directly to Mo's 30-day plan B4 task: "Delete ~30 dead placeholder lookalikes" — CONFIRMED. Slightly fewer than 30 (count is ~28) but the finding is exact.

Critical broken state — the retargeting layer doesn't exist

All Site Visitors (180D) = ~20 people. Visited Auto LP (180D) = ~20 people.

These are PLATFORM-subtype audiences (Pixel-seeded website audiences) that should contain everyone who has visited military.com or auto.military.com in the last 180 days. They are essentially empty (~20 people each).

Two interpretations:

1. The Pixel isn't properly feeding these audiences. Likely cause: cross-domain tracking issue between military.com (the content site, Dataset 587241028099131) and the lead-gen funnel (Dataset 332555186604235). The Pixel that fires on auto.military.com is the lead-gen one, and its audience aggregation isn't writing back to these account-level audiences.

2. SAC (Special Ad Category) is filtering them out. Financial Products SAC restricts Pixel-derived audiences for retargeting. This is the more likely explanation — under SAC, these audiences simply don't accumulate users because Meta filters them as not eligible.

Either way: this account effectively has NO retargeting layer. Mo's audit framework should be updated: under SAC, retargeting Pixel-seeded audiences may not be available at all. The page-engagement audiences (FB/IG, healthy at 30K–1.2M) ARE allowed under SAC, but website-Pixel-seeded ones are not.

This is critical context for the W4 Audiences workstream: under SAC, the legitimate audience seeds are: - ✅ Customer-list (advertiser-provided): the Recruiting-Leads seed and its LALs - ✅ FB/IG engagement (on-Meta surfaces): page likers, engagers, IG followers - ❌ Pixel-seeded website retargeting: blocked or broken under SAC

The "B1 — Define Engaged audience from FB/IG page engagers" task in Mo's 30-day plan is the right move because the engagement audiences (#7–#10 above) are healthy and usable.

The "Mil FB Customer Match 4.12.21" question

Mo's audit referenced "Mil FB Customer Match 4.12.21 (Full List) — last refreshed May 2025 (13+ months ago)". That specific name is NOT in the current audience inventory.

What's there instead: - Custom_Recruiting-Leads_08-05-2025 — created Aug 5, 2025 (~10 months old)

Possibilities: - The "4.12.21" list was archived/deleted and replaced by the "Recruiting-Leads_08-05-2025" list (this would be the "May 2025 refresh" Mo referenced, but dated Aug not May). - The "4.12.21" list is in a different ad account or business. - The audit conducted before today already led to its removal.

Recommend: confirm with Lauren which is the "live" customer-match list — the Recruiting-Leads_08-05-2025 (39K-46K seed) is the only Custom-subtype audience currently active in this account, and its 1%/3%/5% LALs are powering the ALC and Max Val campaigns.

Opportunity Score: 99/100 — CONFIRMED

Account-level Opportunity Score: 99 out of 100. EXACT MATCH to Mo's audit finding.

The single live recommendation

Decode: Meta is suggesting a campaign-objective optimization tweak that would yield a 17% lift. The body text suggests it's pre-create guidance (not a recommendation against existing campaigns). Likely flagged because the account has multiple optimization goals across BAU/ALC/Max Val and Meta thinks consolidating around offsite_conversion goal would be cleaner.

The fact that the only outstanding recommendation is a 1-point lift means structural levers are exhausted. Mo's audit conclusion holds: "leverage is in measurement plumbing, creative, and the success metric — not structure."

Anomaly Signals: NONE FLAGGED

API returned: "No anomaly signal data available for the given criteria."

This is interesting. Mo's audit cited "2 narrow Custom ad sets flagged by Meta anomaly detector." Possibilities:

1. Those anomalies have cleared since Mo's pull (the ad sets are still active with the same ~Custom names but no longer flagged).
2. The anomaly endpoint surfaces a narrower set than the in-product detector.
3. Meta resolved the flags after seeing performance.

Recommended check: Tuesday Open Ads Manager → Campaign view → look for any in-UI anomaly badges on Custom (Auto BAU) (120240808864440498) and Custom (Home BAU) (120227142278180498). If clean, the issue is resolved. If still flagged, the API just isn't exposing them — Mo can address via the UI.

Auction Ranking Benchmarks (last 30d)

Optimization cohort: OFFSITE_CONVERSIONS + complete_registration + prospecting (new customers).

One ad surfaced with diagnostics:

Key insight: TYFYS has good conversion rate (above average) but weak engagement (bottom 35%). The ad earns the click-throughs that lead to conversion — but most users scroll past it. This is the profile of a niche/identity creative that converts well within a small interested segment but won't scale broadly.

Recommended action: keep TYFYS at BAU low spend (where it works) but do NOT scale it into Max Val or broad audiences. This is exactly what Mo's v3 audit concluded — keep it in BAU only.

Other ads not surfaced by the auction-ranking endpoint:

The endpoint only flagged TYFYS_03/09 — the other top spenders (Odometer, Lower-Your-Auto, Lower-Your-Home) aren't surfaced as having auction-ranking issues. This suggests they're all operating in the "expected" auction quality range, even Odometer despite its high CPL ($38.85 in Max Val). The reason Odometer is expensive isn't auction quality — it's the broken value signal in Max Val campaigns optimizing for ROAS against malformed Purchase value data.

Auction overlap diagnosis (the cannibalization question)

The auction-ranking endpoint did NOT surface significant cross-campaign overlap. No "merge campaigns" or "consolidate ad sets" recommendations were returned.

This validates Mo's Integrated Audit finding: the BAU/ALC/MaxVal trio is intentional and not cannibalizing. The media buyer's structural design is sound.

Mapping to Mo's 30-day plan audience workstream

What's worth flagging now

1. The retargeting layer doesn't functionally exist. All-Site-Visitors and Visited-Auto-LP audiences are essentially empty (~20 people). Under SAC, the Pixel-seeded website retargeting may not be allowed. Account effectively has only customer-list LALs + page-engagement audiences. No warm-website retargeting layer.

2. The "Custom_Recruiting-Leads" seed is 10 months old. Refreshing it with last-180-day sold leads (per B2) is high-leverage because: - All three active LALs (1%, 3%, 5%) are downstream of this seed - A 10-month-old seed underweights recent buying behavior - Lauren has the fresh data; this is a 1-day task

3. The 28 broken placeholder LALs are a cleanup-only task — no positive impact from leaving them, no risk from deleting them. Inactive state is safe to delete.

4. The page-engagement audiences are an under-used warm layer. FB Page Likers (1M), Page Engagers (341–401K), IG followers (30K), IG engagers (93–110K) — all healthy sizes, all SAC-compliant. None visible as targeted audiences in current ad sets. Building an ALC or ad set targeting these = real warm-audience layer the account currently lacks.

5. TYFYS is correctly tagged by Meta as low-engagement, high-conversion. Confirms keeping it in BAU low-spend (where it works) and NOT scaling it into Max Val. v3 audit conclusion intact.

6. Opportunity Score is 99/100. Confirms Mo's "no structural fixes; leverage is in measurement, creative, metric" thesis. The 1-point recommendation is a minor objective tweak.

7. No active anomaly flags. The "2 narrow Custom ad sets flagged" issue from Mo's earlier audit has cleared.