The Funnel Is Leaking. Find Out Where.
There is an obvious optimisation available. It is not the constraint. Find the one that is. It takes 75 minutes, the full rubric is published below before you start, and AI tools are allowed.
The brief
The situation. A B2B SaaS product with a self-serve funnel. Growth has stalled for two quarters. You have been given the funnel data and asked where to focus next quarter.
The funnel, last 90 days. Landing page visits: 84,000. Signup started: 12,600 (15%). Signup completed: 9,700 (77%). Workspace created: 8,900 (92%). First data source connected: 2,100 (24%). Invited a teammate: 1,850 (88% of connectors). Converted to paid: 640 (35% of inviters).
Segment data. Users who connect a data source within 24 hours convert at 41%. Users who connect later convert at 9%. Users who never connect convert at 0.4%. Among users who abandoned at the connect step, 68% opened the connection screen and left within 90 seconds.
Additional context. The connection screen requires an admin-level API key from the user's own system. Support logs show 40+ tickets that read some version of "where do I get this key." Marketing has proposed a landing page redesign, projecting a 15% lift on the 15% signup-start rate, and has mockups ready.
Your task. Say where the constraint is, size the opportunity before proposing anything, and design the test that would prove your intervention works.
What you hand in
A written analysis: where the constraint is with the arithmetic that shows it, your proposed intervention, the expected size of the win, and a test design including what result would make you abandon the idea.
Any tools including AI assistance and a spreadsheet. The arithmetic is deliberately simple; what is scored is which arithmetic you chose to do.
How is this challenge scored?
Every criterion, with its weight and the skill it produces evidence for. Nothing else is measured — not writing quality, not grammar, not length. You see this before you start, and there are no hidden criteria.
Locates the 24% connect step as the constraint. Choosing the landing page redesign — the visible, well-championed, pre-mocked option — caps the overall level at 2.
Shows the arithmetic comparing the two opportunities end to end, rather than comparing step conversion rates in isolation.
Connects the 90-second abandonment and the 40+ 'where do I get this key' tickets into a specific hypothesis about why the step fails.
Notices that '24h connectors convert at 41%' is confounded by intent and does not propose simply nagging people to connect faster as if that would transfer the effect.
A design with a primary metric, a stated decision rule, and an explicit statement of what result would cause abandonment of the idea.
Which skills does this prove?
Identifies confounds, quantifies uncertainty, and states clearly what the data cannot support.
Identifies the binding constraint, sizes the opportunity before building, and designs tests that isolate the change.
Isolates the variable, controls for confounds, and pre-commits to a decision rule before seeing results.
Questions people ask
- Is the landing page redesign wrong?
- It is not wrong so much as small. The rubric rewards showing the arithmetic that compares the two end to end. If your arithmetic genuinely supports the redesign, show it.
- Do I need to run statistics?
- No formal statistics required. Arithmetic and clear reasoning about confounds are what is scored.
- What is the most common capped submission?
- Picking the pre-mocked landing page work because it was ready to go. That is precisely the pattern this challenge is built to detect.