The Mockaroo alternative that already knows the shape of your data

Mockaroo starts with an empty schema. You add a field, name it, choose its type from the list, set any options, and repeat until the schema matches what you need. It's a good builder — but it's a description of data you have already described somewhere else.

Because the thing you're generating data for almost always exists already. It's the signup form with fourteen inputs. It's the CSV your colleague sent with one example customer in it. Both of those already say what the columns are and what kind of value belongs in each one. Retyping that into a schema builder is duplicated work, and it's work you redo every time the form changes.

Fillr's whole approach is to read the shape instead of asking for it.

Point Fillr at any web form and save it as a preset — every field, its label, its type. That preset is a schema you didn't have to write.

From there, Generate Sheet produces a dataset shaped exactly like that form: one column per field, each one filled with the right kind of value, as many rows as you want. Download it as CSV, Excel or JSON, or keep it in your account. No schema building, no guessing which of your columns maps to which input — they came from the same place.

When the form changes, you re-capture it and generate again. The schema maintains itself, because it was never a separate artifact.

The other common starting point is a single real example: one row from a spreadsheet, one record from an export, pasted in. Fillr's From Sample Row path reads that row and works out, per column, what kind of value it is — a name, an email, a date, a postcode — using both the column heading and the value itself. You review and adjust its guesses, then generate as many synthetic rows as you need in that shape.

It's the fastest path there is from "here's what our data looks like" to "here are 500 rows of it," and there's no schema step in the middle at all.

Worth being precise about the privacy of this, because it involves a real record: the pasted sample never leaves your browser. Only the column names and the chosen generator keys are stored — never the example values you pasted.

With Mockaroo, the output is a download. If you want the same data next week you regenerate it, and it's different data — the schema persists, the rows don't.

Fillr treats the rows as the artifact. A dataset lives in your account with a name and tags, and you edit it like a sheet: change a cell, regenerate a single column with a different generator, fill only the empty cells, add rows. That matters when most of a dataset is right and a few values need to be specific — the customer ID that has to match staging, the email you actually own — which is the normal state of test data.

Export any time to CSV, Excel or JSON. The dataset stays put.

This is the second half rather than the pitch, but it's what closes the loop. The dataset you generated from a form can be linked back to that form's preset, and then one click fills the form with the next row — sequentially, with the extension telling you "row 3 of 20" so a bug is traceable to an exact record, or at random for variety.

Every field can also have its own rule: generate a fresh value, use a fixed value, leave it blank, skip it, or pull it from a dataset column. Per-field test-data modes cover the values Mockaroo doesn't produce on purpose — boundary values (empty, one character, maximum length, overflow), malformed input, special characters and emoji, and valid-structured values like Luhn-valid cards and MOD-97 IBANs.

So the round trip — form to dataset to form — never touches a spreadsheet or a copy-paste.

Volume and destination. As of August 2026 Mockaroo's free tier gives 1,000 rows per file and 200 API requests a day, Silver is $60/year for 100,000 rows per file, Gold $500/year for 10 million, and Enterprise $7,500/year self-hosted. If you're seeding a database with a million records or running a load test, that's the tool.

It also does things Fillr doesn't attempt: SQL insert scripts, mock API endpoints to develop against, and formula fields for complex derived values. Fillr's datasets cap at 50 rows on Free and 10,000 per dataset on Pro, and export as CSV, Excel or JSON — sized for filling forms and running test cases, not for seeding a warehouse.

Plenty of teams end up using both, and that's a reasonable answer. The question worth asking is which one your day actually consists of: building large data files, or working through the same forms with data you need to control.