How we choose what to work on
We start with the customer outcome, inspect the evidence, and choose one main bottleneck. A bottleneck is the step preventing a customer from making useful progress.
Conceptual illustration. It does not show the Pineapple interface or a measured result.
For Pineapple, the path is straightforward: a suitable business arrives, builds a useful site, pays when the product fits, gets the site working on its domain, and continues using it.
Four questions guide the work
Question | What it helps us decide |
|---|---|
Are suitable businesses arriving? | Whether we have a distribution problem |
Do they become paying customers? | Whether the offer, experience, or measurement needs attention |
Can paying customers get their own domain live? | Whether we are delivering the core job |
Do they stay? | Whether the value lasts |
Our scorecard defines these measures. Overall revenue tells us whether the business is progressing; the customer journey helps explain why.
Choose the smallest useful action
First handle severe customer harm, such as a payment error or a broken core path. For ordinary improvements, look for repeated evidence and a clear connection to the customer we serve. Then find the smallest change that could improve the outcome or resolve an important uncertainty.
“Write more content” is too broad. “Explain the first confusing domain-setup step and check whether a customer can follow it” is a task with a reader, a finish line, and a way to inspect the result.
Doing nothing can also be the right decision. If an existing guide already answers the question, link to it. If the evidence is weak, gather the missing evidence. If a test was deliberately stopped, preserve its result and require a new decision before reopening it.
Record the decision before the result
A useful decision note states the problem, evidence, chosen action, alternative considered, expected result, owner, review date, and condition for stopping. Once the result is known, append it. Keep the original expectation visible.
This prevents a successful-looking story from replacing the actual sequence of events. It also gives the next AI run something better than an old conversation to work from.
Current direction, reviewed 30 October
The latest dated operating summary in our maintained plan is 21 October. It prioritises bringing in suitable businesses and preparing safe help for existing paying customers with payment problems. Research and preparation do not authorise contacting customers or changing their billing.
Search campaigns, ads, proactive outreach, spending and wider experiment exposure still require Peter’s separate decision. The website-chat growth test is closed inconclusive; the old cohort must not be restarted.
Peter owns the business decision. AI can inspect the approved evidence and prepare a recommendation. Use the scorecard for newer billing figures; the plan’s older financial snapshot is not today’s total.
A decision still unresolved
The business decision date was 15 October 2026. As of this review on 30 October, the maintained plan does not record a decision to continue, sell, enter maintenance, or stop. The old target is expired context. This guide does not extend it. Peter needs to record that decision before AI treats it as authority for a new growth phase. Preparing the handbook does not make that decision.
No replacement decision date is recorded in the reviewed plan. Ask Peter to set it when he records the verdict; do not invent one or silently restart the expired growth test.
Useful preparation while that decision is open
Within an assigned task, an assistant can summarise the dated evidence, list the options and identify what would change the recommendation. Use the existing decision record and reports. A useful handoff is one short brief with the proposed next step, unresolved questions and the exact owner decision needed.
If only this handbook is available, keep that brief limited to its evidence. Current product eligibility needs product sources; a loss diagnosis needs billing and customer evidence; profitability needs costs. Name the missing source without inventing its contents or requesting that private records be made public. Continue other authorised work that does not depend on those answers.
Some older Pineapple documents still describe earlier acquisition tests. The newest dated decision in the current plan takes precedence. Unresolved contradictions should be raised, not silently combined.
Try the decision template on one task you are considering this week.
Updated 30 October 2026.
