From CSV to report: upload, analyze and share a file in clariBI

A lot of business data never makes it into a database. It arrives as a CSV: an export from a store platform, a payout report from a marketplace, a vendor's monthly statement, a survey download. This walkthrough takes one such file from your desktop to a dashboard, a few answered questions and a report you can send as a PDF.

The running example is a sales export with one row per order: order_date, order_id, customer, region, category, quantity and revenue.

When a file upload is the right choice

Upload a file when:

• the data comes from a system clariBI does not connect to directly;
• it is a one-off analysis, or a monthly export you are happy to upload again each time;
• you want to try clariBI on real data before connecting live systems.

Connect the source system instead when the data changes daily and you want dashboards to keep up without anyone uploading anything. An uploaded file is a snapshot: clariBI does not re-sync it. If the data is in Google Sheets, the Google Sheets connection is often the better route because it can be synced again.

Step 1: prepare the file

Ten minutes here saves an hour of confusing results later.

• One header row with clear column names. order_date and revenue_usd are easier to ask about than Col7.
• One date format per column. YYYY-MM-DD is the safest.
• Numbers without symbols. "$1,500.00" should be 1500.00. A column that mixes numbers and text ("1500", "pending") is read as text.
• No totals or notes rows at the bottom. A totals row is read as data and doubles your revenue.
• No blank rows or repeated headers from copy-pasting.
• One table per file. Orders in one file, customers in another. Uploading each as its own source keeps questions clear.

clariBI also reads Excel and JSON files, and for Excel it reads every sheet. The delimiter in a CSV (comma, semicolon, tab) is detected automatically.

Step 2: upload it

1. Click Data Sources in the sidebar. This opens Settings > Integrations.
2. Click Add Data Source, choose the Files category and then CSV File.
3. Enter a Data Source Name, such as "Orders Jan to Jun 2026".
4. Click the upload area, pick the file and click Create Data Source.

The file is processed in the background. Each uploaded file counts as one data source and as one of your plan's monthly uploads (5 on Free and the Trial, 10 on Lite, 50 on Starter, 500 on Professional). The file upload help article lists every supported format and limit.

Step 3: check what was read

Before you trust any chart, check the import.

Open the source in Settings > Integrations. The data overview shows the number of rows and columns, the date range found and each column's detected type and number of distinct values. Check three things:

1. Row count matches the file (minus the header).
2. Dates were read as dates, and the date range is what you expected.
3. Amounts were read as numbers. If revenue shows as text, a stray symbol or word is in the column.

Then open the automatic dashboard under AI Dashboards. clariBI builds one for every source on every plan, with headline numbers and charts for the columns it found, and it uses no AI credits. It is a quick visual check: a revenue trend that jumps tenfold in one month usually means a unit or totals-row problem, not a great month.

If something is wrong, fix the file and upload it again as a new source, then delete the old one.

Step 4: ask questions

On the Trial and on Starter and above, open Ask clariBI, start a session with the file as its source and ask in plain English. Useful first questions for a sales file:

• "What was total revenue by month?"
• "Show revenue by category for the last quarter."
• "Which 10 customers brought in the most revenue, and what share of the total is that?"
• "Compare revenue by region in Q1 and Q2."

Each answer comes with a chart. Behind it, the language model plans the calculation and the arithmetic runs as code on your rows, not in the model. Any sentence with a figure the chart data cannot back is removed. Click a bar or point to see the drill-down: which rows were aggregated and over what date range.

The top-customers question is worth asking for any business: if three customers make up 40% of revenue, that is a risk worth knowing about.

Step 5: build the dashboard you will keep

The automatic dashboard shows everything the file contains. A dashboard you build shows what you care about. Two ways:

• AI Dashboards > Create Dashboard (Trial, Starter and above): pick the source, choose or describe the metrics, and clariBI builds up to 15 widgets.
• Manual mode (Trial, Lite and above): start from a dashboard template and choose which column fills each slot. It uses no AI credits.

A layout that works for sales files: headline numbers on top (total revenue, orders, average order value), the monthly trend in the middle, breakdowns by category and region below. The dashboard creation guide covers both routes.

Step 6: generate a report

A dashboard is for looking around. A report is for telling someone what happened.

1. Open Reports and click New Report.
2. Pick a template, for example a revenue trend or monthly performance template. Search by keyword; templates that fit your file are marked Recommended.
3. Select the file as the data source, give the report a title and click Generate Report.

The report has an executive summary, key metrics with charts, findings, risks and next steps. As with answers, every number in the text is checked against the chart data. If the file lacks a field the template needs, you see Data Mismatch Detected with the fields that were found.

Click Download to save it as a PDF (the docs on downloading reports show what the file contains). On Professional and Enterprise you can also share it to a team workspace or through a public link. Reports and AI answers are part of the Trial and of Starter and above; Free and Lite have the automatic dashboard, and Lite adds manual dashboards.

Step 7: next month's file

Because an upload is a snapshot, a monthly export needs a routine:

1. Upload the new file as a new source. Keep the column names exactly the same.
2. Check its data overview.
3. Regenerate the report or rebuild the dashboard from the new source.
4. Delete the old source.

If the export always contains the full history, this is all you need. If each file holds only one month, combine the months into one file before uploading, so trends cover the whole period.

When this routine becomes a chore, that is the signal to connect the source system itself. The guide to moving from Excel to a BI tool covers that step.

A concrete next step: take the export you rebuild in a spreadsheet every month, clean it with the checklist in step 1, upload it in a trial, and ask "What was total revenue by month?" If the numbers match your spreadsheet, the rest of the steps are worth ten minutes.