Good thematic analysis reconciles every count back to N — most don't. Here is the prompt that themes open-text, counts each theme, holds single mentions, and grades the result.
In short: Most thematic analysis quietly loses track of how many people said what. Done well, it names each theme, defines it, counts it, gives two quotes, parks the leftovers in a defined 'other', and reconciles every count back to N — holding single mentions out of the headline until a second response confirms them. One prompt does all of this and grades the result green, amber, or red.
Point the Sopact Sense Assistant at your dataset so it works from clean records with persistent contact IDs, then have it load your Decision Brief first. Clean text and a known N are what let the assistant reconcile theme counts back to the total rather than inventing prevalence.
You are the Sopact Sense Assistant working over the DEMO-04 · Open-Text Barriers dataset (clean data + persistent contact IDs). Load my Decision Brief (decision, audience, outcomes, indicators, evidence standard) first, then wait for my task.
Analyse all responses to the barriers question: themes, definitions, counts, 2 quotes each, 'other', prevalence reconciled to N. Hold single mentions. Grade green/amber/red.
Five elements make this prompt work: it runs over your dataset; it produces themes + counts with definitions and quotes; it reconciles to N so percentages add up; it holds single mentions out of the main pattern; and it ends with a grade of green, amber, or red.
Run on the Open-Text Barriers dataset (DEMO-04) already loaded in Sopact Sense.
GRADE: green | Recurring theme | counted to N; amber | Ambiguous other | undefined; red | Single mention | not a pattern
A green theme recurs, is defined, and reconciles to N. An amber 'other' bucket holds responses that were never defined — usable but unfinished. A red single-mention theme is one response dressed up as a pattern, and should be held back until a second response confirms it.
Take the lowest-graded element above and fix it using only what the program could realistically measure. Show the before → after grade and the single indicator/edit that moves it to green.
Create a 'missing & incomplete' report from this analysis in Sopact branding [or paste your website URL / brand guideline to apply your own]. List every element graded amber or red, what is missing, and the one input that fixes each. Lead with the decision this report informs.
Create a shareable link for this report and open it in a new tab.
Reconcile every count to N. If your theme counts add up to more than the number of responses without a clear note about multi-coding, the prevalence is wrong. Always state the denominator.
Define the 'other' bucket. A catch-all 'other' is where rigor goes to hide. Either define what belongs in it or split it into named themes until it is small and explainable.
Hold single mentions back. One person raising an issue is worth recording, but it is not yet a theme. Keep single mentions in a watch-list, not the headline counts.
Give two quotes per theme. One quote can be a fluke of phrasing; two show the theme is a shared idea, not a single voice.
Show me the 'other' bucket verbatim and propose the smallest set of named themes that would empty it while still reconciling to N.
Read every response, group them into named and defined themes, count how many people fall in each, give two quotes per theme, and reconcile the totals back to N. Park genuinely one-off responses as single mentions rather than inflating them into themes.
It means the theme counts can be traced back to the total number of responses, so prevalence figures are honest. If counts exceed N, you must say so and explain that some responses were coded to more than one theme.
Because one response is not evidence of a pattern. Recording it keeps it visible for next cycle, but counting it as a theme overstates how common the issue is. Promote it only when a second response confirms it.
Open Sopact Sense, paste your program description, and put it to work.
Try in Sopact
