How we work with organizations

For each type: what you hand over, what happens, and what you get back — written as steps and deliverable names. Where something does not exist yet, it says so.

Pricing depends on scale and scope, so no figure is listed here. Tell us the size of your data and what you want analyzed, and we will come back with numbers that fit.

Research and survey firms

Available now

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.

What you hand over
The questionnaire — items, scales, and which ones are reverse-coded
What happens
Each item is read for its scale (how many Likert points), its measurement level, and whether it looks reverse-coded. The reader is `backend/app/preprocessing/survey_analyzer.py`.
What you get back
Item structure table — item x scale x measurement level x reverse-coded candidate/intake
What you hand over
The raw file — Excel, CSV or SPSS (.sav)
What happens
Intake sorts the file into quantitative, qualitative or mixed, and counts type and missingness for every column. Intake is the only place where that sorting happens.
What you get back
Data diagnostic — rows and columns, per-column type, missing cells, and the lane it was routed to/intake
What you hand over
(nothing new) the diagnostic from the previous step
What happens
Reverse items are flipped, outliers and missing cells handled. Each step is a proposal, not a decision: you choose, every change is logged with its reason, and every change can be undone.
What you get back
Preprocessing record — what was changed, why, how many rows it touched, and the undo history/prep
What you hand over
Items grouped into scales (for example, six items for one construct)
What happens
Reliability and factor structure are measured: the `reliability` engine returns Cronbach alpha, item-total correlations and alpha-if-deleted; the `efa` engine returns KMO, Bartlett, explained variance and rotated loadings.
What you get back
Scale validation tables — alpha, alpha-if-deleted, KMO, Bartlett, rotated loadings, cumulative variance/analysis
What you hand over
The analysis plan — what is compared with what
What happens
The engines matching that plan are called. Crosstab, t-test, ANOVA, correlation and regression are separate engines; assumption checks, effect sizes and post-hoc tests come back in the same call.
What you get back
Result tables — test statistic, degrees of freedom, p, effect size, assumption checks and post-hoc comparisons/analysis
What you hand over
(nothing new) the results from the previous step
What happens
Tables and interpretation are assembled into a report. Sentences follow APA convention and every number comes from the table itself, so the report cannot say a different number than the table.
What you get back
Report — paper-ready tables, APA sentences, and "copy into my workbench"/report

See a finished study Contact us

Local government

Coming soon

You run a resident satisfaction or policy-demand survey every year and have to explain the result to a council and to residents. The hard part is not the arithmetic — it is answering "so where do we invest first". Here the result is split by the background variables you already collected (district, age band, years of residence), and each difference comes back with an effect size so you can tell signal from noise.

What you hand over
The questionnaire and the raw file — with background variables (district, age band, years of residence) as columns
What happens
The item structure and the data are read to separate background columns from rating items.
What you get back
Item structure table and data diagnostic/intake
What you hand over
(nothing new)
What happens
Reverse items are flipped, outliers in columns such as age are handled, missing cells are dealt with, and all of it is recorded.
What you get back
Preprocessing record — what was changed, why, and how many rows it touched/prep
What you hand over
Items grouped by domain (for example: administrative service, living environment, participation, trust)
What happens
Cronbach alpha checks whether each group really hangs together, and factor analysis checks whether the items split the way the design intended.
What you get back
Scale validation tables — alpha by domain, KMO, the rotated loading matrix and cumulative explained variance/analysis
What you hand over
How to split — district, gender, age band
What happens
The right test is called for each split: a t-test for two groups, ANOVA with post-hoc tests for three or more, crosstab with Cramer’s V for category by category.
What you get back
Group comparison tables — group means and SDs, test statistic, p, effect size, post-hoc groupings/analysis
What you hand over
The priority question — what actually raises overall satisfaction
What happens
Domain scores go in as predictors and a regression measures how much each one lifts overall satisfaction. Multicollinearity (VIF) comes back with it.
What you get back
Regression table — R-squared, standardized beta, p, VIF, and the resulting priority order/analysis
What you hand over
(nothing new)
What happens
Tables and interpretation are assembled into a report. For local government the report leads with which districts score low and which domain to invest in first.
What you get back
Report — paper-ready tables, APA sentences, and a district-by-district summary/report

The procedure above is settled. What is missing is the worked example — we cannot use a real municipality’s data, so the demonstration will be built on synthetic data and published here. Until then this slot stays empty rather than carrying a placeholder.

Small and mid-sized businesses

Concept

No one here does research full time. You run a customer or employee survey, the result comes back as a single average, and that average does not tell you what to change. This type is still being designed — what follows is a direction, not a promise.

What you hand over
The raw customer or employee survey file
What happens
Intake reads the file and works out which analyses the data can support.
What you get back
(not settled) data diagnostic/intake
What you hand over
How to split — columns you already have, such as branch, tenure or customer tier
What happens
Differences across those splits are measured, and it is checked whether they are more than noise.
What you get back
(not settled) group comparison table
What you hand over
(nothing new)
What happens
What to change first, on one page.
What you get back
(not settled) one-page summary

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.

BETAstatory public beta · feedback bonuses · Beta benefits · Pricing