Verified against ChatGPT · 2026-08-11
Run a SWOT that grades every claim by how well-supported it is, instead of four lists of confident guesses
Builds a SWOT analysis for a real business situation where every entry is tagged as fact, informed inference, or open question — so the finished analysis doesn't quietly present speculation with the same authority as verified information.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are running a SWOT analysis for a real situation I'm facing. The thing that makes most SWOT analyses useless is that every entry reads with the same flat confidence whether it's a documented fact or a guess dressed up in analyst language — I want the confidence level visible, not hidden.
SITUATION AND CONTEXT
Whether our 12-person SaaS company should expand into the EU market next year.
KNOWN FACTS I CAN CONFIRM
We have 40 paying customers in the UK already, our product is GDPR-compliant, we have no local support staff in any EU timezone.
WHAT I'M UNCERTAIN ABOUT
Not sure how price-sensitive continental European customers are compared to UK customers, or whether our current pricing tier translates.
COMPETITIVE OR MARKET CONTEXT
Two established competitors already operate in France and Germany with local sales teams.
Build the four SWOT quadrants, but tag every single entry with one of three labels: [FACT] for something directly supported by what I told you, [INFERENCE] for a reasonable conclusion drawn from the facts but not itself stated by me, or [OPEN QUESTION] for something that would materially change the analysis if answered but that I haven't given you enough to resolve — do not present an inference with the same bare confidence as a fact, and do not fill an empty-feeling quadrant with generic industry boilerplate ("strong brand" for a strength with nothing to back it) just to make the list look fuller. For each entry, keep it specific to my actual situation — a strength or weakness that could apply to literally any company in the industry doesn't belong here. After the four quadrants, pull out the two or three open questions that would most change the strategic picture if answered, and say specifically what answering each one would let us conclude that we can't conclude now. Flag any strength and weakness that are actually two sides of the same underlying trait, since that's a common pattern worth surfacing explicitly rather than leaving as two disconnected entries.
WHAT NOT TO DO
Do not produce four evenly-sized lists just for visual symmetry — a thin quadrant with two honest entries beats a padded one with six weak ones. Do not soften a real weakness into euphemistic language.
OUTPUT FORMAT
Four labeled quadrants (Strengths, Weaknesses, Opportunities, Threats), each entry tagged [FACT]/[INFERENCE]/[OPEN QUESTION]. Then a short "most consequential open questions" section and any strength/weakness pairing worth flagging.Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
A generic SWOT prompt produces four lists that all read with identical declarative confidence, and this is a genuine failure mode rather than a stylistic quibble — GPT-5.1, like most models, tends to fill in a requested structure completely even when the underlying evidence to support every cell doesn't actually exist, because leaving a quadrant sparse reads as an incomplete answer to the model even though a thin, honest quadrant is more useful than a padded one. Explicitly requiring a fact/inference/open-question tag on every entry forces a different kind of reasoning: instead of generating a plausible-sounding claim and stopping, the model has to check that claim against what was actually supplied versus what it's extrapolating, which surfaces exactly the boundary between what you actually know and what the model is inferring on your behalf — a boundary a flat, untagged list erases entirely. The instruction against generic industry boilerplate targets a specific, very common SWOT failure: asked for strengths, a model will readily produce category-level truisms ("strong brand," "agile team") that could apply to nearly any company in the sector and therefore convey no actual information about this specific situation; forcing specificity to the given context makes these generic filler answers structurally harder to produce because there's nothing concrete to hang them on. Surfacing the two or three highest-leverage open questions at the end, rather than leaving open questions scattered flatly across four quadrants, does the genuinely useful analytical work of ranking which unknowns are actually load-bearing for the decision — most open questions don't matter equally, and a model that just lists them without ranking leaves you to do that prioritization yourself, defeating the purpose of asking for help in the first place.
What you get back
Strengths: [FACT] 40 existing paying UK customers [FACT] Product is GDPR-compliant
Weaknesses: [FACT] No local support staff in EU timezones [INFERENCE] Support response times to continental customers would likely lag UK customers given the timezone gap
Most consequential open question: how price-sensitive are continental customers relative to UK ones? Answering this would determine whether current pricing can transfer or needs a local tier.
Flagged pairing: existing UK customer base (strength) and no EU support staff (weakness) are two sides of the same "UK-only operating model" trait.
Verified against
ChatGPT GPT-5.1 · 2026-08-11
Changelog
- 2026-08-11 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — custom software, built and maintained for you — that's Scult's day job.
EXPLORE CUSTOM SOFTWARE
