Research

Verified against ChatGPT · 2026-08-13

Separate a real trend from a noisy spike before it goes in a strategy deck

Runs a trend analysis that explicitly checks whether a pattern is sustained and structural or a short-term spike, before presenting it as a trend worth acting on.

ChatGPT (GPT-5.1)4 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

I've noticed what looks like a trend and want to know whether it's real before I put it in a strategy document. Help me stress-test it, not just describe it.

WHAT LOOKS LIKE A TREND
Sign-ups from a specific industry vertical (dental clinics) have tripled over the past 6 weeks.

DATA OR SOURCES I'M BASING THIS ON
Weekly sign-up counts by industry tag for the past 6 weeks, no data further back than that available yet.

TIME WINDOW OBSERVED
6 weeks.

WHAT DECISION THIS WOULD JUSTIFY
Building a dedicated onboarding flow and marketing push specifically targeted at dental clinics.

First, check the time window — a pattern visible over 6 weeks. needs to be checked against a longer baseline before calling it a trend rather than a seasonal blip or a one-off spike; if you don't have longer historical context in Weekly sign-up counts by industry tag for the past 6 weeks, no data further back than that available yet., say explicitly that the trend can't yet be distinguished from a short-term fluctuation. Second, name at least one plausible cause for the pattern that is NOT the interesting narrative explanation, and check whether the mundane explanation is at least as consistent with the data as the interesting one — a specific one-off event, a seasonal cycle, a change in how the data was measured, or a small sample size inflating an ordinary fluctuation into something that looks dramatic. Third, only call it a genuine trend if it's structural (driven by a lasting underlying cause) rather than episodic, and state which one your evidence actually supports.

WHAT NOT TO DO
Do not describe a pattern as an emerging trend just because the framing I gave it sounds compelling — the fact that I noticed it and found it interesting is not evidence that it's real or lasting. Do not recommend Building a dedicated onboarding flow and marketing push specifically targeted at dental clinics. on the strength of a pattern you've just flagged as possibly noise.

OUTPUT FORMAT
1. A verdict: genuine structural trend / likely noise or short-term spike / can't tell yet given available data.
2. The mundane alternative explanation you checked, and why it does or doesn't hold up.
3. What additional data or time window would resolve the uncertainty if the verdict is "can't tell yet."
4. Whether Building a dedicated onboarding flow and marketing push specifically targeted at dental clinics. is currently justified by this evidence, stated plainly.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

A model asked to analyze "what looks like a trend" is primed by the user's own framing to find and elaborate on that trend, since agreeing with the premise of a request is the path of least resistance — which means without a specific instruction to stress-test the pattern, the output will tend to describe why the trend makes sense rather than checking whether it's actually there. Requiring the mundane alternative explanation first forces a genuinely adversarial pass: a specific one-off cause (a single referral partner, a seasonal event, a measurement artifact) is concrete enough to actually check against the data, unlike a vague instruction to "be careful" which produces no real scrutiny. The time-window check matters mechanistically because six weeks of data has no way to distinguish a structural shift from a seasonal or coincidental blip — the honest answer when longer historical context isn't available is genuine uncertainty, and a model that fills that gap with a confident trend narrative anyway is doing exactly the kind of unwarranted extrapolation that turns short-term noise into a strategy-deck slide. Explicitly forbidding a recommendation to act on a pattern the analysis itself just flagged as possibly noise closes the most consequential failure mode: a hedge buried in paragraph three of an analysis gets ignored by a reader skimming for the recommendation in the executive summary, so the instruction has to prevent the recommendation from being made at all rather than trusting the hedge to survive downstream summarization.

What you get back

Verdict: can't tell yet given available data. Mundane alternative checked: a single dental-industry conference happened in week 3 of the window and one attendee posted about the product in a niche Slack community — this alone could explain a temporary spike without any lasting shift in the dental-clinic segment's interest. Additional data needed: sign-up trend for this vertical over the prior 6 months, and whether the spike sustains for another 4-6 weeks after the conference effect fades. Current recommendation: do not yet build a dedicated onboarding flow based on this data alone.

Verified against

ChatGPT GPT-5.1 · 2026-08-13

Changelog

  • 2026-08-13 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
All Research prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY