Learn → Getting started

Getting started

Prism is not a spreadsheet editor and not a replacement for Excel. It is the step in between receiving a file and trusting it: open it, find out what is wrong with it, and decide what to do. A first pass takes a couple of minutes.

1. Open a file

Drop an .xlsx, .xlsm or .csv onto the upload screen, or browse for it. If the workbook has several sheets you choose which ones to bring in, with row counts shown so you can tell the data sheet from the cover sheet.

Reading and analysis happen in a background worker thread, which is why the interface stays responsive on large files instead of freezing while it churns. We have tested to 500,000 rows / 38 MB.

If you would rather not start with real client data, the landing page offers two sample datasets. The lead sample is a small accounts-payable register seeded with the kinds of defects this tool exists to find.

2. Check the column roles before anything else

Prism classifies every column into one of seven roles — Identifier, Date, Currency, Number, Category, Text, or Ignore — and that classification drives everything downstream. Which checks run, which charts appear, what can be grouped: all of it follows from the roles.

It gets most columns right, and it will get some wrong. A reference number stored as digits can read as a Number; a code column can read as a Category. Skim the roles first. Thirty seconds in the Configuration view here saves you puzzling over a nonsensical dashboard later, and a misread role is the single most common reason results look odd.

3. Read the quality score

The score out of 100 sits on the Data Quality view along with every issue behind it, grouped by severity. The number matters less than the list: each issue names the rows it affects, and clicking through takes you into the data with the filter already applied.

Work down from high severity. The quality score explained documents all 13 checks and exactly how the number is calculated.

4. Look around the rest of the file

• Dashboard — totals, averages, date range, distributions, and a time series when a date column pairs with a currency or number column.
• Records — the table, with search, sorting, category filters, and range filters on dates and numbers.
• Column Profile — per-column statistics: fill rate, distinct values, distribution, min/max/mean.
• Pivot Table and Browse by Category — group and summarise without leaving the review.

All of the above is free, on files of any size. The free tier is limited by capability, not by row count — a row limit would mean the free tier could never demonstrate anything on a file that actually matters.

5. Act on what you found

Seeing the problems is free. Changing things is what a licence unlocks:

Editing produces a new file — a values-only copy, with the original left untouched. That is deliberate rather than a limitation we are hiding: rewriting a workbook in the browser cannot preserve formulas, formatting, macros or pivot caches, and quietly handing a client back a file with its formulas stripped is a worse outcome than an extra file on disk. Prism warns you about this before it saves.

6. Make next month faster

The setup you just did — role corrections, checks you disabled, computed columns, report configuration — can be saved as a named recipe. When you open next month's export of the same report, Prism recognises the shape of the file and offers to apply it in one click.

That is where the time actually goes in a recurring review, and it is what turns Prism from a one-off tool into part of a monthly close.

Where your work is kept

Workspaces autosave to your browser's local storage so you can close the tab and resume later. Nothing leaves the device. Two consequences worth knowing:

• Work is per-browser, per-device. It does not follow you to another machine, because syncing would require a server.
• Browsers can evict local storage under pressure. Prism asks for durable storage and warns you in the sidebar if the browser declines. Export anything you cannot afford to redo.