Dashboard Green. Funnel Red.
The dashboard was green.
Revenue trend was red.
Same day. Same traffic. Opposite story.
I saw this in a high-friction health funnel. Quiz-heavy flow. Paywall at the end. The leak was in the middle, not at the top.
I track this chain in rates only: click -> onboard_finish -> purchase_initiated -> purchase_completed -> conversion_to_purchase.
No vanity totals in the decision view. Only percentages.
The sample size is large enough to remove noise. Tens of millions of clicks.
Where the Lie Starts
Most B2C dashboards over-celebrate early funnel events.
Clicks. Quiz starts. Quiz completion. CTR.
All useful. None of them are payment.
I had the same trap in a health funnel. The top looked alive. The paywall looked tired.
One Number Changed the Room
I added a hard metric to the first row.
purchase_initiated_rate_from_onboard_finish
No weighted score. No composite KPI.
One boundary metric.
When this number dropped, the team could not hide behind green vanity cards.
The paywall is the purchase moment.
That is where psychology does the heavy lifting: FOMO, timing, social proof, price framing, offer clarity.
If this screen is weak, the funnel leaks even when everything above it looks healthy.
Onboarding still matters just as much.
Onboarding loads intent. It answers objections early. It builds trust before the ask.
The paywall closes the deal. It should not carry the full persuasion job alone.
Metric Layers That Help
I now keep four layers.
- attention: click-through, quiz start
- engagement: quiz completion, screen depth
- intent: paywall_view, purchase_initiated
- money: purchase_completed, refund_rate
One layer can rise while another collapses.
That is normal.
Confusing layers is expensive.
What We Changed
We stopped reporting one blended “funnel health” score.
We reviewed the boundary between engagement and intent first.
Every week. Same order. Same chart.
The conversation moved from “traffic is up” to “conversion_to_purchase is down, why”.
That is where real fixes start.
We also personalized the paywall using onboarding data.
We mirrored key funnel inputs on the paywall:
- primary goal from onboarding
- key constraints selected in quiz steps
- selected plan path from the last onboarding decision
The screen felt continuous, not reset.
The user no longer had to re-interpret the offer at the final step.
The paywall looked like step N+1. Not a random new page.
conversion_to_purchase moved from ~1.5% to ~1.8-2.0% across weekly windows.
That is roughly +20-33% on the same chain.
The Weekly View That Worked
I review this as two weekly windows at a time.
Week A vs Week B.
Same chain. Same metric names.
click -> onboard_finish -> purchase_initiated -> purchase_completed -> conversion_to_purchase
No blended score. No hiding behind top-funnel growth.
Trade-Off
You lose pretty dashboards.
You gain decision speed.
The first one makes meetings feel good.
The second one makes the business better.
Takeaway
Green dashboards can still hide a dying paywall.
Track one chain end to end:
- click
- onboard_finish
- purchase_initiated
- purchase_completed
If the chain breaks, fix the exact link. Do not optimize averages.
Next: Day-0 Lies. Day-7 Decides..