A regression test re-checks a feature that already worked, so a new change does not break it unnoticed. A regression test template is the list of those checks, kept from release to release, with a priority and a last-run date on each row. The table above is that template with six worked examples for an example online shop, and you can download it as an Excel workbook, a CSV file or Markdown.
How to use it#
- Read the examples above. Each row has a feature, preconditions, numbered steps, an expected result, a priority and the date it last ran.
- Select Download .xlsx for a workbook with a How to use tab, the examples and 25 blank rows, or Download .csv to import the rows into another tool.
- Write one row for each behaviour your users rely on, and give it a priority: High for what would stop people, Medium for the rest.
- On each release, run the High rows at least, update Last run, and set Status.
What each column means#
| Column | What goes in it |
|---|---|
| ID | A short, unique reference, so a result or a bug can point at this row. |
| Feature | The feature the row protects, so you can run all rows for one feature after it changes. |
| Title | A short name for the check, readable in a list. |
| Preconditions | What must be true before the first step: who is signed in, what data exists, which page is open. |
| Steps | What to do, one numbered action per line, written so someone new to the app can follow them. |
| Expected result | What should happen if the app works. Write something you can see or check, not 'works correctly'. |
| Priority | How much it matters if this fails: High, Medium or Low. Run High first when time is short. |
| Last run | The date of the last run, so stale rows stand out. |
| Status | The result of the last run: Not run, Passed, Failed or Blocked. |
Reading the examples#
Preconditions let each row start without reading the one before it. Priority decides what runs when there is no time for everything. Last run shows which rows have gone unchecked: in the examples, the order history and the password reset email were last run a week before the rest, and the reset email has not been run in this round yet. A failed row, such as the card payment in the examples, is a bug to report with the steps already written.
Keeping the list healthy#
A regression list grows with every feature, and a list that is too long does not get run. Remove rows for features that are gone, merge rows that check the same thing, and add a row for every bug that reached your users. Rows that run on every release with the same steps are the first to automate, because a script repeats them exactly and never skips one. Regression testing for AI-generated code covers the automated version.
Regression testing vs sanity testing#
A sanity test looks at the one thing that changed in a build. A regression test looks at everything that was not meant to change. After a fix, run the rows in the sanity test template first, then this list. Afterwards, move the sanity row for the reported bug into this list. Smoke vs sanity vs regression testing follows one release through all three.
Questions#
What is a regression test case template?#
A sheet with one row per behaviour that must keep working, with its feature, preconditions, steps, expected result, priority, last-run date and status. The template on this page has nine columns and six filled rows.
How many regression tests do I need?#
Enough to cover what your users rely on most. Start with the High priority paths and add rows as bugs are found.
How often should regression tests run?#
Before every release, at least the High priority rows. If you release often, automate them.
Is the template free?#
Yes. Download it without an account and use it in your own plans and tools.
Will these examples fit my app?#
No. They are for an example online shop. Replace them with your own app's features and keep the columns.