ANALYSIS AND IMPROVEMENT
Let your data guide your business.
Your orders, customer transactions and operational records are already in your project. Describe your question and use your own project data to examine processes, measure results and prepare forecasts.
Use the data in your project
There is no need to prepare CSV files or upload your records again. Your data already lives in the connected project database; the workspace reads only the sources and fields you are authorized to use.
An analysis workspace must be connected to your project. You may need to sign in with your project account. If no connection is available, the system tells you; it never switches to another project’s data.
Start with a question
“At which step do orders wait longest?”
“Which campaigns bring better results?”
“Forecast sales for the coming weeks.”
The system reads an authorized database snapshot and runs data quality, process mining, marketing measurement, forecasting and classification flows. You review the draft and fields, then connect the result to an approved automation or report workflow.
Optional: file analysis tools
This separate tool is for external files. Use project analysis above to work with records already in your system.
What do you want to learn?
Your request helps choose an analysis type. This screen runs defined analyses, not arbitrary Python/SQL or LLM commands.
Prepare your data
Do not mix currencies in one analysis. No currency conversion is performed.
Up to 1,000 records / 20 columns. Include only necessary columns; remove names, emails, passwords, keys and unnecessary personal data. All CSV columns are sent; mapping does not remove unused columns. Choosing a file does not send it.
Findings and next step
Measurements, data warnings and suggested reviews will appear here after analysis. No analysis has run yet.
Prepare an improvement request
Record the finding, intended change and success measure together. The downloaded request is not execution approval or permission to apply it.
Continue the request in your project
Download the improvement request, open chat and select your project. In the left menu, use Saved requests → Read list → Open analysis request to select the file. Review the text, add it to the list and save. Later, Use in composer brings it into chat; it is not sent until you send it.
This link does not transfer the file or access your project. An imported summary is not a verified report or execution permission. Text you send in chat may be processed by your configured model; do not include secrets. Saved requests remain private to you and the selected project.
This screen does not change a connected app, update ad budgets or publish content. Applying a change later requires a fresh check of target, scope, authorization and approval. Re-analyze using the same data definition to compare results.
What changed after the change?
Keep a measurement and its analysis recipe. Measure the next dataset with the same recipe and see the results side by side. A numeric difference alone does not prove improvement or causation.
Which data does this measurement describe?
Run an analysis above first. The scope and dates below are your declarations; they do not filter the CSV or verify its source. Do not enter secrets or personal names.
For example, Europe/Istanbul or UTC. Use the data’s time zone; dates are not automatically converted.
The file contains column names, mapping, measurements, warnings and the declared scope above. It excludes raw rows, your chat request and project credentials. You keep the file; it is not automatically saved to your account.
Choose a baseline measurement
Up to 64 KiB JSON. Choose the measurement file from this section, not an analysis report or improvement request. Opening it does not upload it.
Prepare the new CSV above, apply the recipe, then give fresh upload consent and analyze it. Column names and order must match. Opening a file does not change mapping or upload consent.
No baseline selected.