Answer Si's questions
The clarify and review step — Si explains how it understood your data, then asks a few multiple-choice questions about what you want to track, unclear columns and rows with missing data; how to answer, add context and change the data type.
Si does not guess where it is unsure: it asks you. The Clarify and review step has two parts: How Si reads your data and Si needs you to choose.
How Si reads your data
Si reads any kind of data: sales, HR, finance, operations, marketing, education, healthcare and more. This section states the domain, what the data is about and a few questions it can answer. The collapsed Notes about the data lists anything worth knowing, for example a schema name that does not match its contents or a column that always holds the same value; these notes never block building the dashboard.
Data without a date column still gets a dashboard: the metrics are current values of the loaded data (for example headcount or attrition rate), not compared across periods.
Multiple-choice questions
Each question has 2–4 options; one is labelled Si's suggestion and is preselected. Questions that accept several answers say "You can pick several." If no option fits, pick Other and explain; Si keeps your note as context for its next analysis. Examples:
- What do you want to track first? (attrition rate, payroll cost, satisfaction, workforce mix): the metrics you choose go to the top of the Focus page and onto the home page;
- What are monthlyrate, hourlyrate and dailyrate? (an amount of money, a plain number, ignore): Si picks the unit and calculation from your answer;
- About 29 rows are missing a department and position. Keep or drop them?: keep them and they are labelled "Unknown"; drop them and they are left out of every metric of that table;
- Si is not sure: … Is this reading correct?: something Si could not verify; confirm it or write down the correct reading.
With a ready-made analytics template (for example multichannel retail), Si also asks about mappings, such as which order statuses mean a cancelled order.
Answer, then click Save and continue. Your answers and the assumptions are saved with the approved version; the next time Si re-analyses, your earlier answers become the suggested options.
Uncertain mappings
With a ready-made template, the section Si isn't sure about N mapping items shows the confidence of each item (for example 62%). Si's detailed description of the data is the full write-up. Fine-tune details in the advanced review screen, see Review the proposal and build.
Add context
The box Data context for Si (optional) takes notes (Save notes) and description documents (Upload description document). Click Extract knowledge and Si turns the terms and rules in the document into Knowledge entries awaiting approval. After adding context, click Let Si analyse again.
Change the data type
When none of the ready-made templates fits, Si builds a custom pack from the columns of your own data. To use a different template, open the Data type box, choose Change data type and then Change and let Si analyze again.
Related
Mask personal data
The data and personal-data step — choose which columns to mask, decide whether raw values may be stored, pick the processing region and consent to sending a masked sample to Si.
Review the proposal and build
See which metrics and dashboard pages Si will build, your answers and the assumptions, fix mappings in the proposal review screen (yellow cells), approve, build and follow progress until the dashboard is ready.