AA10 solar cancellation — quick scan and format plan
Read: 7 September 2026. ClickFlare live reports, Pacific/Auckland, completed weeks 24–30 August and 31 August–6 September. Campaign-name scope: established AA10 Solar Exit campaigns; no fresh Meta account-ID reconciliation. Dollar figures use the existing tracker USD convention (API response has no currency field); do not compare directly to the NZD Meta winners export. Conversions are tracker leads, not verified closed sales. Raw weekly rows are saved alongside this report.
Portfolio
| Week | Spend | Leads | CPA | Revenue | ROI |
|---|---|---|---|---|---|
| 24–30 Aug | $11,559.26 | 537 | $21.53 | $18,213 | 57.6% |
| 31 Aug–6 Sep | $10,935.21 | 455 | $24.03 | $15,465 | 41.4% |
Malformed/truncated campaign labels excluded: $194.98 then $55.05. No conversions on those excluded rows. Full campaign-name rows reconcile across the campaign-level and weekly ad-level pulls. Aggregate performance deteriorated; that does not imply every control deteriorated.
Current winners, kept separate by campaign
| Ad | Campaign | Spend | Leads | CPA | ROI | Prior-week CPA |
|---|---|---|---|---|---|---|
| Solar Cancel Blue Statement Card V3 | (Test) 2 Sep 2026 | $1,897.52 | 80 | $23.72 | 41% | — |
| V-07 Motion Lender Card | (Test) 20 Aug 2026 | $1,028.31 | 50 | $20.57 | 67% | $20.74 |
| Savings Never Arrived Woman | (Test) 20 Aug 2026 | $628.40 | 42 | $14.96 | 125% | $21.01 |
| C-25__bold_text__v1 | (Scale) 14 Aug 2026 | $575.12 | 27 | $21.30 | 63% | $18.86 |
| V-08 Motion Red Notice | (Test) 20 Aug 2026 | $418.32 | 18 | $23.24 | 48% | $22.62 |
| C-12__bold_text__v1 | (Scale) 14 Aug 2026 | $366.18 | 19 | $19.27 | 82% | $14.39 |
| V-07 Motion Lender Card | (Scale) 14 Aug 2026 | $349.39 | 25 | $13.98 | 143% | $25.99 |
| C-01__bold_text__v1 | (Scale) 14 Aug 2026 | $248.39 | 15 | $16.56 | 107% | $20.43 |
What to preserve
- Savings Never Arrived Woman: strongest recent UGC efficiency with rising spend and volume (20 → 42 leads). Local source G4 confirms GoodLeap first, promised savings versus actual bill, then a sixty-second eligibility quiz. Specific recognition and a short next step are the plausible mechanism; performance does not isolate the actor from the script. Source: projects/goodleap-ugc-15s/generate.py.
- V-07 Motion Lender Card: proven format transfer already exists. Test campaign produced 51 then 50 leads at essentially unchanged CPA; scale improved from $25.99 to $13.98, on smaller volume. Preserve GoodLeap qualifier, cancellation possibility, free check and minimal typography. Source: projects/solar-exit-round5-video/artifacts/scripts.json, V-07, lineage SL-215.
- Veteran and lender cards: C-12 and C-01 still profitable in scale. Preserve the identity gate; do not blend C-12 with C-27 merely because both use veteran messaging. C-27 v2 has 63 leads but only +16% ROI this week, down from +42%.
- Blue Statement Card V3: meaningful new volume, 80 leads on $1,897.52 at +41% ROI. Actual creative was not located in this quick scan; its name alone is not enough to infer its copy or layout. Retrieve the asset before cloning.
Proposed first wave: 12 new ads, released in two batches
| Priority | Parent | New format | Executions | Keep fixed / hypothesis |
|---|---|---|---|---|
| 1 | Savings Never Arrived Woman | Plain comparison static + kinetic comparison video | 2 + 2 | GoodLeap + promised savings versus actual bill + existing quiz CTA. Test whether the strongest spoken grievance transfers to silent-scroll formats. Use generic bill labels, no invented dollar amounts. |
| 2 | V-07 Motion Lender Card | Presenter beside the lender card + overhead paperwork explainer | 2 + 2 | Same short lender/eligibility message; change presentation. Presenter points to the existing card; hands locate the lender field on illustrative paperwork. No extra legal story. |
| 3 | C-12 veteran card | Direct-address video + plain question-and-answer card | 1 + 1 | Preserve veteran identity and cancellation eligibility. No new military-benefit claim. Lineage SL-236. |
| 4 | Savings winner + historical two-person podcast | Kitchen-table two-person dialogue | 2 | Same savings discrepancy and CTA, converted into question/answer. A deliberate larger format change; historical podcast evidence supports testing, not a promise of current efficiency. |
First release: 6 ads—one execution of each of the first four treatments, plus veteran direct address and one two-person dialogue. Second release: the remaining six after first-wave delivery is understood. Each execution pair changes only one opener/visual treatment within its format. Cross-format results necessarily test a presentation package, not a perfectly isolated causal variable.
Reserve wave: retrieve Blue Statement Card V3, then make two motion adaptations retaining its verified copy/layout. Do not invent a blue-card reconstruction from its ad name.
Testing plan
- Keep current winning posts as concurrent controls. Test in separate ad sets; preserve audience, destination, offer, attribution and bidding conditions as far as possible. Do not add creatives into working scale ad sets.
- Use the same campaign-specific CPA and ROI comparisons shown above. Motion Lender Card in Test is a ~50-lead/week control; its Scale version is a separate benchmark.
- Proposed exploration allocation: 70% to the savings/lender format transfers, 20% to veteran, 10% to the two-person swing. This is a recommendation, not a budget change.
- Size spend to the live payout and target margin before launch. A zero-lead ad at roughly 2× target CPA is an early review trigger, not statistical proof. For positive ads, review lead quality and economics at meaningful spend, then seek repeat profitable performance before scale. No CTR-only winner declarations.
- Do not split scarce budget across all 12 at once. Keep low-delivery ads labelled untested.
Do-not-repeat / corrections to older plans
- C-09 neighbour note: 7 leads, $53.46 CPA, −35% ROI this week; prior week −33%. Remove from priority scale-out despite historical wins.
- SolarNews v07 in Scale: 12 leads, $86.86 CPA, −64% ROI. Recommend a spend review; no account changes made.
- Do not generalize that every news format fails: Holder Rule Breaking Report in the 2 Sep Test campaign has 11 leads at $16.40 and +113% ROI, a smaller promising sample. Its economics differ drastically across campaigns.
- The earlier claim that video was barely funded is obsolete for this window: V-07 and Savings Never Arrived now have material live spend and conversions.
Evidence and scope
Read-only analysis and a creative test plan. No assets generated, ads edited, budgets changed or messages sent. No new offer claims are required. Production follows the appropriate static/UGC/motion pipeline after selecting the batch and tool path. Historical source: Desktop/AA10 Solar Winners - Top 25/README.md, through 8 Aug; podcast scripts: projects/_analysis/aa10-winners/transcripts.json. Current findings supersede the September 1 lineage-only UGC shortlist.