Excel Sales Dataset Generator

Yes. No account, no credit card, no trial. You give an email address because that is where the file is sent, and you can unsubscribe from anything that follows with one click.

Because every product is guaranteed a sale in every month of every year, so a pivot table never has empty cells in it. That floor is your product count times your year count times twelve. Ask for 100 rows with 40 products over 4 years and you get 1,920, because that is what the guarantee costs.

The readout under the sliders always shows the real number before you generate anything, and tells you when it had to round up. Duplicate records are added on top of it.

You can, and for twenty rows it is quicker. Past that it falls apart in a specific way: ask for sales data without a very long brief and what comes back is generic. Round prices, sales spread evenly across every month and every product, no weekend effect, no seasonal peak, every item selling like every other one. There is nothing in it to find, which is the one thing practice data has to have.

It also does not hold up at length. Ask for 5,000 rows and you tend to get a few hundred, or a script that produces them with the same flat distribution you were trying to avoid.

This runs one generator with the patterns already built in, so they are there whether you ask for them or not: categories peak in their own season, weekends outsell midweek, expensive items get fewer but deeper discounts, prices end in .99, and the mess is exactly as much as you asked for. Every row is consistent with every other one, all the way to 10,000.

A sample file is one fixed thing. It has the rows it has, the mess it has, and everyone practicing on it works on identical data. If you need it bigger, cleaner, dirtier or spread over four years instead of one, you are out of luck.

This builds the file to your spec each time, from 100 rows to 10,000, clean or deliberately broken. That matters most if you are teaching or setting an exercise: generate one clean copy and one messy copy of the same shape, or give everyone their own file so nobody can copy an answer.

No, and that is the point. Every row is generated, so there is no customer in it, nothing confidential, and nothing you need permission to publish. Use it in a tutorial, a course, a job application or a blog post without having to anonymize anything first.

One hour. The file is deleted from the server after that, so grab it while the page is still open or use the link in the email. If it expires, just generate another one, it takes a few seconds.

No. Every run uses fresh random values, so two files with identical settings still have different transactions in them. That is useful for setting an exercise nobody can copy the answer to, and it does mean you should keep a file you liked rather than expecting to rebuild it.

Because working it out is the exercise. You have Quantity, Unit_Price and Discount_Rate on every row, which is everything you need to build a line total, and doing it yourself is closer to the real job than reading a column somebody else calculated.

Yes. It is a standard .xlsx workbook, so Excel, LibreOffice, Numbers and Google Sheets all open it, and pandas.read_excel reads it directly. Pass sheet_name=None to load all three sheets into a dictionary of DataFrames in one call.

It sends you the download link, and it goes on my free users list so I can tell you about new tools and the occasional deal. One click unsubscribes, and I do not sell or share it. There is more detail in the privacy policy.