Bug priority vs severity: the matrix, with examples
Short answer
Severity is how badly a bug breaks the product. Priority is how soon it must be fixed. QA usually sets severity from technical impact, and the product owner sets priority from business need. The two are independent: a crash can be low priority and a typo can be urgent. A severity by priority matrix sets the order of work.
Every bug needs both values, because one answer cannot say both how bad it is and how soon it matters. ISTQB defines severity as the degree of impact a defect has on a component or system, and priority as the level of business importance given to it. Below is the matrix teams use to turn the two into a fix order, with examples for each cell that causes arguments.
Priority and severity side by side
| Severity | Priority | |
|---|---|---|
| Question it answers | How bad is it? | How soon do we fix it? |
| Driven by | Technical impact: crash, data loss, broken function, workaround | Business need: deadline, who sees it, revenue, contract, release date |
| Usually set by | QA engineer or the person who found it | Product owner, project manager or stakeholder |
| Changes over time | Rarely, only if the bug behaviour changes | Often, as releases and business needs shift |
| Typical scale | Critical, High, Medium, Low (some teams add Trivial) | P1 to P4, or Urgent, High, Medium, Low |
Severity by priority matrix
Read down for severity and across for priority. Each cell gives the usual handling. The corners are easy; the interesting cells are the ones that cut across.
| Severity \ Priority | P1 Urgent | P2 High | P3 Medium | P4 Low |
|---|---|---|---|---|
| Critical Crash, data loss, security hole, no workaround | Stop and fix now. Block the release | Fix before release. Consider a hotfix | Unusual. Check the priority is not understated | Crash only on input no real user sends. Log it and revisit |
| High Major function broken, workaround exists | Fix this sprint. Usually on the release path | Fix before release | Schedule for the next sprint | Backlog |
| Medium Partly broken, limited impact | Seen by a buyer or executive. Fix fast | Fix before release | Next sprint | Backlog |
| Low Cosmetic, typo, spacing | Wrong wording on a launch page. Fix today | Fix with the release | Batch with other work | Fix if time allows, or close |
How to use it. Severity goes in the bug report, written by QA. Priority is decided in triage with the product owner in the room. The cell is a starting point for the conversation, not an automatic rule.
When severity and priority disagree
High severity, low priority. The application crashes when a user pastes more than 50 phone numbers into a registration field. It is a hard crash, so severity is critical, but no real user does it, so the fix waits in the backlog.
Low severity, high priority. The company name on the homepage has a typo and a launch event is tomorrow. Nothing is broken, so severity is low, but every visitor will see it, so it is fixed first.
High severity, high priority. Checkout fails for every card payment. Nothing works and money is being lost: the top-left cell, fixed immediately.
Low severity, low priority. A hover colour is a few shades off on an admin page few people use. It waits for the next design pass.
Four bugs from our own website, rated twice
When we tested betterqa.co in October 2026 we rated each bug for severity and priority separately. The two answers disagreed more often than they agreed.
- Severity low, priority high. FAQ questions turned bright pink on hover, on 37 pages. Nothing broke, but buyers saw it on the pages meant to answer their doubts. Fixed the same day.
- Severity high, priority high, after a check. Firewall logs with visitor IP addresses could be downloaded. Before treating it as an emergency we measured the reach: file names carried a random suffix and the folder could not be listed. It still went to the top, and a server rule now refuses the folder.
- Severity medium, priority high. Service pages showed their headings in Times, because 41 pages asked for a font the site never loaded. On paper cosmetic; the priority changed when a prospect pointed at it on a call.
- Severity low, priority low. The rotating line under the homepage headline was blank most of the time. It waited a few days and shipped with other homepage work.
Across the bugs our engineers reported in BugBoard on client projects from November 2025 to October 2026, 5% were critical, 24% high, 55% medium and 16% low. About 3 in 10 are high or critical, which is why severity alone cannot set the order of work.
How to set both values in four steps
- QA rates severity first. Does it crash, corrupt data, block a core function, or only look wrong? Is there a workaround?
- The product owner sets priority. Consider the release date, how many users it reaches, who will notice and what it costs not to fix.
- Record both in the ticket. Developers need the urgency and the technical scope. Do not merge them into one field.
- Review in a regular triage slot. Priority drifts as releases approach. Severity should not change unless the bug does.
Tools differ. Azure DevOps has Severity and Priority fields on bugs, each with four levels. Jira has a built-in Priority field and usually needs Severity added as a custom field. GitHub Issues relies on labels or project fields. Pick names once and write them down, because "Critical" means different things in different teams.
Classification as part of defect management
We report every bug with a severity, the steps to reproduce it and the evidence, and leave priority to the people who own the release. If your backlog has become an argument, our defect management service sets the classification rules, runs triage with your team and keeps the numbers honest.
Frequently asked questions
Where this comes from
ISTQB, Standard glossary of terms used in software testing, entries for severity and priority. IEEE 1044-2009, Standard classification for software anomalies. Microsoft, Azure DevOps documentation for the bug work item type. Atlassian, Jira documentation on priority. The own-website examples and the severity split are BetterQA data.
Need help with triage and reporting?
Tell us how bugs are handled today and where it hurts. We will suggest a classification scheme that fits your team.
Book a call Defect management