NovaPay Financial — 30 / 60 / 90 Day Recovery Plan

Turn −$401K into +$50K

— and a contested renewal into expansion, in one quarter.

NovaPay is catching the fraud — recall is ~95%. The cost is the problem: precision is ~51%, so half of every flag is false — one root cause behind the 3,400-alert backlog, 67% auto-decisioning, and −$401K of wasted ROI every two weeks.

Health: Yellow, trending red — and every red is fixable

What this plan delivers

Three commitments, one quarter

Every promise below is anchored to a number in NovaPay's own data — not a forecast we hope lands.

🛡️

Renewal

Stop the churn

Visible wins land before the competitor eval and the OCC exam. We move renewal risk from Medium-High to Low by proving recovery, not promising it.

High → Lowrenewal risk, by day 90

📈

Turnaround

Prove it with numbers

Auto-decisioning climbs toward 80%, the backlog clears, and net ROI crosses back to positive — each step traced to a specific tuned rule.

−$401K → +$50Knet ROI in 90 days

🚀

Growth

Power what's next

Once trust is rebuilt, RTP / Push-Fraud is the fraud layer their FedNow rollout and Q3 BNPL launch already require — sequenced after the proof point.

$480K → $600KARR with RTP expansion (+25%)

The plan, phase by phase

The 90-day recovery

Select a phase to see the projected KPIs, the trajectory, and exactly who does what, by when.

Stabilize → Optimize → Prove

Critical

Recovery trajectorynormalized 0–100 · dashed = target · ▾ selected phase

This phase — what we do

Why we're confident

One problem, three costumes

100% data-driven: every recovery number above traces back to these 10 rules — a 2-week sample of 34,261 transactions.

Four high-volume rules fire half the volume at under 25% precision. Fix four rules — not the platform.

95%

Recall (fraud caught)

51%

Precision (half false)

−$401K

Net ROI / yr

Net ROI by rule

Red bleeds left · green earns right · annualized $K

🔴 The 4 bleedersTUNE

~$496K of the negative ROI sits in these four. Even halving their false fires is a six-figure swing — and #4776 already proved tuning works.

⭐ The 4 winnersAUTO-DECLINE

+$109K combined, 84–95% precision. Auto-declining them lifts auto-decisioning and hands analyst time straight back — week one.

What I'd confirm in discovery Confident in the direction — these three I'd validate before committing numbers to paper.

NovaPay's exact false-positive methodology. Reconcile the 22% portfolio-level vs ~49% rule-level figures — different denominators, not a contradiction. Confirm which one NovaPay reports against.

The +$50K assumptions. False-decline CLV ($250) and the achievable post-tune precision (~45%) are modeled — validate both against NovaPay's real numbers and the A/B outcomes.

Competitor-eval & renewal timeline. Confirm the evaluation window so the 30/60/90 proof points are landing before the decision, not after.

Appendix

The model behind every number

Reproduces NovaPay's workbook to the dollar — change any blue assumption or pull a recovery lever and every number recomputes. Net ROI is per 2-week period (×26 = annualized). Explore it after the call.

Baseline reproduces the Excel to the dollar

Open the live model

Assumptionseditable inputs

The blue cells from the workbook — tweak to match the client's economics.

Recovery leverstied to the 30/60/90 plan

Jump to a phase, or tune by hand. Each lever maps to a committed action in the plan above.

Auto-decline the 4 winners #3 #4 #7 #10 — 84–95% precision, off the manual queue

Tune the 4 bleeders — target precision (recall held)

✓ True declines (fraud caught) are held constant — tuning cuts false positives, never recall.

#	Rule	Fires	False+	Precision	Review $ /2wk	Net ROI /2wk	Quadrant