Context

A small B2B SaaS team had a shapeless funnel and a crowded roadmap. They traded the quarter for a week and asked a humbler question: What can we finish by Friday that changes user behavior?

Baseline (Before Week 1)

• Time‑to‑route: 6m 40s median; many mislabels.

• Correct intent label: 42%.

• First‑aha in 24h: 38%.

• Click‑to‑trial from offer page: 22%.

The numbers weren’t catastrophic; they were blurry. Blurry is expensive.

What We Did (Mon → Fri)

Monday — Choose 3 NBAs and write the camera test

• Router rule: normalize inputs; map Learn • Trial • Buy; compute a simple 0–10 score.

• First‑aha checklist: embed a 3‑minute setup after signup.

• One‑promise offer: reduce the offer page to one promise and one CTA.

Tuesday–Wednesday — Build the smallest systems

• Ship router normalization (ISO country, E.164 phone, disposable‑email filter).

• Add hidden field intent_label + events (route_decided, aha_reached).

• Draft one‑promise offer; wire UTMs.

Thursday — Observe and adjust

• Kill a fragile regex; moved scoring to config.

• Enlarged the “Start Setup (3 min)” button; added a 15‑sec GIF on mobile.

Friday — Demo and decide

• Show logs: inputs → normalized → score → route.

• Show scorecard: North Star + levers + health.

• Write one‑sentence learning for Monday.

Results (After Week 1)

• Time‑to‑route: ↓ to 1m 52s median (p50).

• Correct intent label: ↑ to 78%.

• First‑aha in 24h: ↑ to 50% (+12 pp).

• Click‑to‑trial: ↑ to 26% (+4 pp).

• Health metrics: refund rate steady (< 3%); support SLA unchanged.

The week didn’t change the world. It changed the slope.

Evidence (What We Showed on Friday)

• Router logs: sample JSON with inputs, normalized fields, score breakdown, route.

• Scorecard screenshot: North Star, levers, health.

• Offer page before/after: annotated captures.

• Checklist screen recording: 60‑sec clip to first‑aha.

What We’d Change Next Time

• Instrument earlier. One mis‑fired event cost half a day of confidence.

• Cut Variant B from the offer test; the lever was the promise, not the palette.

• Pre‑approve the Friday demo definition with stakeholders on Monday.

Playbook You Can Copy (5 steps)

1. Router first. Normalize inputs; label Learn • Trial • Buy; compute a simple score.

2. Aha checklist. Embed a 3‑minute checklist post‑signup.

3. One promise. Reduce the offer page to a single job and a single CTA.

4. One scorecard. North Star • 3–5 levers • 2–3 health metrics.

5. Friday decision. Winner rolls forward; loser is archived with one learning.

Friday demo prompt: “Here are the logs, the one number that decided, and what we’ll change on Monday.”

Method Notes & Data Fingerprint

• Window: Monday–Friday, same week.

• Traffic: ~2,400 sessions; 610 trial signups; 190 mobile completions to first‑aha.

• Attribution: UTMs set at source; intent captured via hidden field.

• Stability: Directional reads; re‑test if variant samples < 300 events.

• Privacy: PII masked in logs; consent respected for nurture.

What to Do Next (Pick Your Path)

• Open the Picker → We’ll map you to Book, $97 Challenge, or Toolkits.

• Try the 7‑Day Challenge → Your $97 becomes credit toward a Toolkit.

• Get the AI Business Growth Toolkit → Routers, scorecards, test matrices, and blueprints.