How we approach marketing
Useful marketing helps a suitable customer make a decision or complete a job. For Pineapple, that starts with the owner of an existing service business trying to get a professional website live and keep it updated.
Conceptual illustration. It does not show the Pineapple interface or a measured result.
The audience for this handbook is broader: customers, business owners trying AI workflows, and people following the experiment. That does not change the product's customer focus. Interest from other founders is not evidence that Pineapple's core customers are getting more value.
Our actual choices, reviewed 30 October
The channel decisions below serve that customer focus. We review interest in this handbook separately from evidence that customers value the website product.
Channel or activity: Product documentation and useful feature explanations
Recorded decision: Prepare accurate answers from verified product behaviour and keep their evidence dates visible
What would justify a change: A checked customer question and permission for the exact publication
Channel or activity: Search, ads and proactive outreach
Recorded decision: Capped acquisition preparation exists; activation and spend remain off without Peter's approval
What would justify a change: A dated business decision, defined audience, budget and measurable outcome
Channel or activity: Lifecycle and recovery email as a growth channel
Recorded decision: Retired in July after the recorded tests produced no useful read, click or recovery signal
What would justify a change: New evidence and a new decision; do not reopen the old loop
Channel or activity: Website-chat experiment
Recorded decision: Closed inconclusive on 8 October: three eligible apps matured, below the ten-app decision threshold; incomplete event coverage prevented a reliable conversation result
What would justify a change: A new dated test with a reliable cohort and full observation window
These are decisions from our maintained plan and experiment register, not a complete inventory of every historic marketing asset. The 21 October plan prioritises suitable new demand and safe preparation for paid-customer recovery. The broader verdict from the passed 15 October strategy deadline is still unrecorded. Essential customer support is separate from the retired email-growth experiments. No channel gets new authority because this page was refreshed.
One useful inconclusive result: the final chat-test read found three mature eligible paid apps and no recorded conversation outcomes. Missing conversation events mean we cannot conclude that no conversations happened. The sample also missed the ten-app decision threshold. We kept the feature unchanged and stopped the expired dedicated measurement loop. Calling setup clicks a lead-generation success would have taught the wrong lesson.
For this guide, the immediate job is reader understanding: can an owner follow the complete FAQ example and explain where AI must stop? Distribution and paid-growth attribution come later. We have no measured acquisition cost, profitable-channel claim, or reader-completion result to publish yet.
Start with a real question
A support question, repeated failure, customer conversation, or verified product change can reveal something worth explaining. Inspect what already exists before writing. Choose one of three actions: update an existing page, create a page for a distinct job, or skip it.
An accurate explanation of a difficult domain step may help more than ten general articles about AI. Content volume is not our measure of progress.
Put each explanation in the right place
Reader's need | Home for the answer |
|---|---|
Complete a product task | Product documentation |
Understand how Pineapple operates | This handbook |
Follow a dated result or change of mind | A company update linked to the relevant handbook page |
Try a process in their own business | A worked example and reusable template |
A blog post can introduce an idea and tell its story. The handbook remains the maintained reference. If a rule changes, update the reference first and make the dated article's context clear.
What AI can contribute
AI can gather approved evidence, find an existing answer, draft a page, compare claims with sources, and prepare a distribution draft. A person still needs to own uncertain claims and any publication outside the authority already granted.
We should publish an honest failure when it teaches something useful. A result is not a success because the article is persuasive. Customer quotes require permission; examples we invent for teaching must say they are fictional.
How we would judge this guide
First ask whether a reader can understand who Pineapple serves and where AI's authority ends. Then ask whether an owner can complete the starter workflow without help. After publication, inspect qualified product visits and later billing-backed outcomes as separate measures.
Downloads and page views show interest. They do not establish that the guide caused paid growth. At Pineapple's scale, a few completed reader tasks and candid conversations may be more useful than a premature conversion claim.
Search titles should describe the question each page answers. The handbook keeps company decisions separate from product instructions; the blog introduces the experiment and links back to the maintained reference. We will measure search visibility after publication, without promising rankings.
Our research drew on PostHog's marketing handbook and content handbook. The transferable idea is to connect helpful material to a specific customer; Pineapple's channels and budget need their own evidence.
Next: follow one task through the operating loop.
Updated 1 November 2026.
