Verified against ChatGPT · 2026-08-09
Write a troubleshooting guide that branches instead of forcing every reader down the same path
Builds a branching troubleshooting guide where each step's outcome determines the next one, instead of a single linear checklist that wastes time on readers whose situation splits off early.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are writing a troubleshooting guide for a specific recurring issue. It needs to branch based on what the reader observes at each step, not force every reader through every step regardless of relevance. ISSUE User can't log in — repeatedly told 'incorrect password' despite entering the correct one. POSSIBLE ROOT CAUSES, RANKED BY LIKELIHOOD 1) Caps Lock or autocomplete inserting a stale saved password (most common), 2) account locked after failed attempts, 3) SSO provider outage, 4) actual password was changed and forgotten. READER TECHNICAL LEVEL General consumer app users, not technical — cannot be expected to know browser dev tools or clear cache manually without exact click-by-click instructions. WHAT SUCCESS LOOKS LIKE AT EACH STAGE If retyping the password manually (not autofilled) logs them in, cause was #1. If they see a 'too many attempts, try again in 15 minutes' message, cause was #2. STRUCTURE THIS AS A DECISION TREE Start with the single fastest step that would confirm or rule out the most likely cause — not the safest generic first step every guide defaults to (like "restart the device") unless that genuinely is the most likely cause here. After each step, branch explicitly: "if you see X, go to step 2a; if you see Y, go to step 2b" — never write a flat numbered list that silently assumes everyone's situation matches step 1's outcome. State what success looks like at each branch point using If retyping the password manually (not autofilled) logs them in, cause was #1. If they see a 'too many attempts, try again in 15 minutes' message, cause was #2. so the reader isn't left guessing whether the step worked before moving on. For a reader at the stated technical level, calibrate language accordingly — do not use terms like "clear the DNS cache" without a plain-language explanation if General consumer app users, not technical — cannot be expected to know browser dev tools or clear cache manually without exact click-by-click instructions. indicates a non-technical audience, but don't over-explain basic actions if the audience is technical, since padding a technical reader's guide with unnecessary hand-holding makes them distrust the rest of the content. WHAT NOT TO DO Do not include a step that requires a permission level, tool, or access the stated reader wouldn't have. Do not end a branch without either a resolution or an explicit "contact support and mention you've confirmed X" — never let a branch just trail off without telling the reader what to do next. OUTPUT FORMAT A numbered/lettered decision tree (1, 2a, 2b, 3a, 3b, etc.), each step with: the action, what success/failure looks like, and where to go next.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Left to its own defaults, GPT-5.1 tends to write troubleshooting content as a flat linear list because that's the more common shape in general how-to writing, but a linear list silently assumes step 1 fails for everyone before step 2 becomes relevant — when in reality a meaningful fraction of readers' situations branch off after the very first step, and a flat list wastes their time walking through irrelevant steps built for a different root cause. Explicitly requiring branch points ("if X go to 2a, if Y go to 2b") forces the model to represent the actual diagnostic structure of the problem instead of collapsing it into false linearity, which is the single biggest difference between a troubleshooting guide that resolves an issue in two steps for most readers and one that makes everyone read all six regardless of relevance. Ordering the first step by actual likelihood rather than defaulting to the conventional "restart it" first step matters because that generic default, while harmless, wastes the reader's time when the ranked data shows a different cause is actually far more common — GPT-5.1 has no way to know the real distribution of causes unless it's given the ranking, and will otherwise reach for the most stereotypical first troubleshooting step regardless of fit. Calibrating language to the stated technical level addresses a specific asymmetric risk: an under-explained step to a non-technical reader creates a dead end where they don't know if they succeeded, while an over-explained step to a technical reader signals the whole guide might be beneath them and gets abandoned — GPT-5.1 defaults toward a middle, moderately-explained register that under-serves both ends unless explicitly told which end this audience sits on.
What you get back
1. Manually retype your password (don't let it autofill) with Caps Lock off. Success: you're logged in — this was cause #1, a stale saved password. Failure: still says incorrect password, go to step 2. 2. Try logging in again. If you now see a message saying to wait 15 minutes, that's cause #2 (too many attempts) — wait and retry, no further steps needed. If you get the same 'incorrect password' message with no lockout notice, go to step 3. 3. Check whether you log in with a company email through single sign-on. If yes, go to 3a. If you use a regular email/password, go to 3b...
Verified against
ChatGPT GPT-5.1 · 2026-08-09
Changelog
- 2026-08-09 — 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
