Verified against ChatGPT · 2026-08-14
Plan a help center article around the exact question people are actually stuck on, not a rewrite of the feature's documentation
Produces a help center article plan built from a real cluster of support tickets, scoped to the specific confusion customers hit, checked against existing articles so it doesn't duplicate or contradict what's already published.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Plan a help center article based on a real cluster of support tickets, not a general rewrite of how the feature works. TOPIC OR TICKET CLUSTER About 15 tickets over a month from customers confused about why their scheduled report didn't send — turns out most had the wrong timezone set in their profile, not a bug. EXISTING ARTICLES There's an existing 'Setting Up Scheduled Reports' article that covers creation but doesn't mention timezone settings at all. AUDIENCE TECHNICAL LEVEL Non-technical end users, mostly small-business owners managing this themselves without an IT person. PRODUCT AREA The reporting and scheduling module. FORMAT CONSTRAINTS Articles should be under 400 words, use numbered steps for any action, and avoid internal jargon like 'scheduler service'. STEP 1 — THE ACTUAL QUESTION State the specific question or point of confusion these tickets actually share, in the customer's own framing, not the feature's official name — people search and get stuck using their own words, and an article titled around internal product terminology won't match what they're actually confused about or searching for. STEP 2 — CHECK AGAINST WHAT EXISTS Check the topic against the existing articles given. If something already covers this, state plainly that a new article isn't needed and instead recommend what should change in the existing one — a near-duplicate article fragments the help center and makes both versions less discoverable and more likely to drift out of sync with each other over time. Only plan a genuinely new article if the gap is real. STEP 3 — OUTLINE SCOPED TO THE ACTUAL CONFUSION Outline the article scoped tightly to the actual confusion identified in Step 1 — do not expand it into a full feature walkthrough covering everything the feature does, since a customer arriving with one specific stuck point wants that answered quickly, and padding the article with adjacent information they didn't ask about makes the actual answer harder to find. Pitch the language and assumed background at the stated audience technical level. WHAT NOT TO DO Do not invent product behavior, menu labels, or steps that weren't confirmed — if a specific UI detail is needed to write an accurate step and wasn't given, flag it as something to confirm before publishing rather than guessing a plausible-sounding label. OUTPUT FORMAT 1. The actual question/confusion, stated plainly. 2. Existing-article check and recommendation (new article vs. update existing). 3. If new: article title and section-by-section outline. 4. Any UI or product detail that needs confirming before this can be published accurately.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Requiring the actual confusion to be stated in the customer's own framing, not the feature's internal name, targets a specific and very common help-center failure: articles get titled and structured around what the product team calls the feature ("Scheduled Report Configuration") rather than what a stuck customer is actually asking ("why didn't my report send"), and a model asked to plan documentation for a feature will default to the official terminology because that's the more available, authoritative-sounding frame, even though it's a worse match for how people actually search and phrase their confusion. Checking against existing articles before planning something new addresses a structural risk of AI-assisted content planning: a model has no persistent memory of what already exists in your help center unless explicitly given it, so left unchecked it will happily plan a brand-new article that substantially duplicates or, worse, subtly contradicts an existing one, and over time this is exactly how help centers accumulate near-duplicate articles that erode trust when two pages give slightly different instructions. Scoping the outline tightly to the one identified confusion, rather than a full feature walkthrough, matters because GPT-5.1 asked to write help documentation tends toward comprehensiveness by default, and a customer arriving mid-frustration with one specific stuck point (their report didn't send) is served worse by a thorough article covering report creation, formatting, and sharing than by a short one that answers the actual question fast — comprehensiveness here isn't the same as usefulness, and the two need to be explicitly decoupled.
What you get back
Actual confusion: customers believe their scheduled report failed to send when the real cause is an incorrect timezone setting on their profile, not a delivery failure. Existing-article check: the current 'Setting Up Scheduled Reports' article doesn't mention timezones at all — recommend adding a short troubleshooting section there rather than creating a fully separate article, since this is closely related to existing setup content, not a distinct topic. Section outline for the addition: 'Report didn't arrive when expected? Check your timezone setting' — one short explanation, three numbered steps to find and correct the profile timezone, one line on when to contact support if the timezone is already correct. Needs confirming: the exact menu path to the timezone setting in the current UI wasn't provided and should be verified before publishing.
Verified against
ChatGPT GPT-5.1 · 2026-08-14
Changelog
- 2026-08-14 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
