statory · for organizations
A stack of questionnaires. A year of service-request records. A production log. Your organization already has the data. Statory takes it the rest of the way — to sentences you can quote. The engine does both the computing and the reading, and the same data always gives the same result.
Every analysis shows its evidence.
Three kinds
Each kind starts from different data, so the route differs. Where we cannot do something yet, we say so.
Two things. The annual resident satisfaction survey, read by district and age band; and the service-request records that pile up every day, turned into daily, weekly and monthly reports in a fixed shape. No new survey is run.
Every business keeps its records differently, so we shape the report around the data you already have.
Hand over the questionnaire and the raw file, and you get back one set: the preprocessing log, the scale checks, the analysis results and the report. Anything changed by hand stays in the log.
A resident satisfaction survey happens once a year. Service requests, intake and processing records, on the other hand, pile up every day. Regular reporting reads the data you are already collecting, at a fixed time, in a fixed shape. No new survey is run.
The three cycles ask different questions
How are things right now
Is something moving
Why is it that way
What goes out automatically today is the monthly report. The control charts and multiple-comparison correction the daily and weekly reports rely on are still being built into the statistics engine; when they land we publish the comparison table and remove this mark.
The +310.0% jump over yesterday is a weekday artefact — yesterday was a Sunday. Against the same weekday a week earlier it is −9.9%.
One request type jumped +130.0%, but 8 types are being watched at once, so the threshold is not 0.05 but 0.00625. Not a signal.
One department has the lowest on-time rate (65.4%). Split instead by deadline length, the 3-day bucket is handled fastest (median 2 days) and yet misses its deadline most often (75.2%). The on-time rate is measuring how the deadlines were set, not how fast the work is done.
No one here does research full time. But the data is already piling up every day — production logs, sales slips, inventory books. At month end it turns into a single total, and that one line does not tell you what to change. Here that data is read as management indicators, and on top of it the question "what should we do" is backed by calculation. This type is still being designed — what follows is a direction, not a promise.
This is a concept. The steps are written but the deliverable names are deliberately not fixed — fixing them would make them a promise, and we have not decided yet. The "(not settled)" markers disappear once we do. Because the shape of the data differs so much by industry, fitting one real data set is faster than building a generic demonstration.
If you have a customer or employee survey, the research-agency flowworks for it as it stands.
You already design the study and collect the data. What is left — preprocessing, analysis, reporting — rides on whoever happens to do it, so quality moves study to study. What changes here is that those three run on the same engines in the same order. Every manual edit is recorded, and results come back as sentences you can quote as they are.
Measured from a run we actually did
Four domains together explain 59.2% of overall satisfaction, and the strongest of them carries β = .369. That figure is what the four explain together — put one in on its own and it changes.
The monthly report in the local-government example carries five of them — comparison with the same month last year, the net effect of the new app channel, department performance, the level of resident satisfaction, and the cause of the gap between districts. None can be answered from the data at hand, and we write down what more would be needed to answer each.
All demonstration data is synthetic, and we say so in the data and in every report — it is not a real survey, and it is not any real local authority’s data.