Twenty days to close the year's last 15%.

Monday, March 9. Close the gap in twenty days, or the account changes hands.

On the regular weekly call, the client set the number: the remaining gap on the year's revenue commitment, worth over 15% of the annual target. The month's pipeline was already approved, five tests, sequencing ours. One test had just concluded; a second was mid-flight, its sample indifferent to any deadline.

Not more ideas. Fewer, in the right order.

First move: restructure the approved pipeline to front-load revenue into the first ten days. Three filters made every call:

The physics were fixed: a test needs four to five working days for a significant sample, and weekends are dark. Twenty days is not twenty chances. It is three, maybe four, clean cycles.

The projection came back short: the designed tests closed about 75% of the gap, and the quota of five tests would need to become seven. The client had offered a fail-safe, rerun past winners that had already earned once. We logged it and left it there: a number built from reruns would not survive scrutiny. The rest would need something new. I started hunting.

Visitors were stepping backward.

In GA4 user flows, the funnel's sharpest behavior shock: visitors at the guest-details step abandoning rooms they had already chosen, stepping back to re-browse.

The client killed the first idea. The better one shipped.

The data backed a social-proof badge on the top-booked rooms; the first framing, Most Preferred, died in client review for ranking a luxury brand's own suites. Instead of arguing, we shipped an ABC test: same slot, same styling, only the words changed.

Reassurance lifted all three metrics. Urgency downlifted all three. In luxury, being told a room is loved converts; being told to hurry repels.

The safety net returned a tenth of its old yield.

One rerun did make the board: hotel highlights on the personal-details page, our strongest past winner, placed to justify the full price where it lands. Live day thirteen, paused day sixteen, at a tenth of its historic yield. The refusal to build the plan on reruns stopped being a principle and became a measurement.

The board also showed where the money had come from. Four tests on the booking landing page had earned 89% of everything banked; the off-path pages returned the rest across three tests. The third filter was no longer a preference, it was the map, and the diagnosis pointed at the same page: the surprise was born on the room cards. The last test goes where the choice begins.

Defuse the surprise where the choice begins.

What shipped instead: one line under every room card's price, the real total visible from the first moment of choice, the explanation one tap away. The tax-inclusive display was locked off by an upstream SEO decision; nothing said it could not be annotated.

The finding outlasted the number.

What I would do differently: the shortfall was visible in January to anyone with a live view of the number. Given sixty days, the first five would go to a weekly, client-visible revenue view against commitment. The sprint should never have needed to be one. Built with a team lead, a project manager, a UI designer, and a developer; the pipeline calls, the hypotheses, and the two shipped framings were mine to get right.