MemoLogs Book a walkthrough
Case studies Fintech
Fintech

A unicorn fintech cut CPI 37% and increased click-to-acquisition 30% in one quarter.

−37%
Cost per install
+30%
Click-to-acquisition
95%
Install spend audited
< 5wk
Time to first result
Challenge

What wasn't working.

The client — a late-stage consumer fintech running $1.2M/mo in paid acquisition across Meta, Google, and TikTok — had hit a wall. CPI had climbed 22% over two quarters despite no change in creative or targeting strategy. Platform dashboards were inconsistent: Meta Ads Manager claimed a $6 CPI, GA4 said $11, and their CRM suggested something in between.

Worse, their VP of Growth had just left. Institutional memory of why previous scale-ups failed was walking out the door.

Approach

What we did.

We started with a full attribution audit on the 95% of install spend concentrated in Meta and Google.

In week two, Creative IQ identified four fatiguing ad sets that accounted for 38% of spend — their 9-signal health scores had all dropped below 30. We queued refresh variants and began A/B testing immediately.

Weeks 3–5, we ran the first geo-lift study to establish true incremental CPI versus platform-reported CPI. Tri-model consensus (GeoTwin · CausalCore · DiffLens) returned an Attribution Inflation Factor of 2.8× on Meta and 4.1× on branded Google search.

“Meta claimed a $6 CPI. Geo-lift said $17. Once we stopped optimizing against the lie, everything got better.”

The reallocation plan: cap branded search at 60% of its prior budget, move the surplus into Meta prospecting creative that had tested well in the fatigue-replacement cohort, and add frequency caps at the audience level.

Results

What changed.

By end of Q2, CPI on the recovered-signal basis had dropped from $17 (true incremental) to $10.70 — a 37% improvement. Click-to-acquisition improved 30% as the refreshed creatives and tightened frequency caps lifted relevance.

MetricBeforeAfter 90 daysChange
Reported CPI (Meta)$6.10$4.90−20%
Incremental CPI (geo-lift)$17.10$10.70−37%
Click → acquisition1.9%2.5%+30%
Branded search spend$340K/mo$200K/mo−41% (waste)
Creative refresh cadence30 days (calendar)~14 days (fatigue-triggered)Reactive → proactive
“We analyzed campaigns driving 95% of installs. The frequency capping alone brought immediate efficiency — then the creative refresh queue and the branded search reallocation stacked on top. Every month since, the number has gotten cleaner, not noisier.”
— VP Growth · Consumer fintech app

See what your account is missing.

90-day pilot · flat fee · pilot credited to contract.

Start a pilot →