Guide · Project management
What is scope creep? Causes, examples and how to prevent it
Why software features grow after the sprint starts, how to tell real scope creep from work that was always there, the questions that catch most of it in planning, and what to do when scope changes mid-sprint.
In short. Scope creep is work added to a feature after it was planned and agreed, without the time, budget or people changing to match. Most of it comes from questions nobody asked in planning. Prevent it by writing down what done means, breaking the feature down, and trading work instead of adding it.
Scope creep, defined
Scope creep is work added to a feature after it was planned and agreed, without the time, budget or people changing to match. On software teams it rarely arrives as a new request. It arrives as a question nobody asked in planning, about permissions, email, billing or edge cases. Asking those questions before the sprint prevents most of it.
Each addition looks small. Together they push the date out.
A scope creep example
A team plans “let admins invite teammates by email.” Here’s how the five weeks went.
- Week 2“What if they already have an account?”
- Week 3Billing: invites have to respect seat limits.
- Week 4Support: “Why did my invite never arrive?”
- Week 5Shipped. Four weeks late, and nobody can say why.
None of that work was new. Existing accounts, seat limits and bounced emails were part of team invitations from day one. Nobody had written them down, so they arrived one at a time, mid-sprint.
What causes scope creep
- Nobody broke the feature down before estimating it, so the estimate covered the obvious parts only.
- “Done” was never written down. Everyone had a slightly different finish line in mind.
- Edge cases surface during the build: errors, limits, existing data, people who are already signed in.
- The people affected join late. Billing, support or security hear about the feature after it’s half built.
- Nobody owns the trade-off. New asks get added without anyone deciding what moves out.
Scope creep or discovered scope?
Not every late addition is scope creep. Telling the two apart tells you what to fix.
Scope creep
Someone wants more than was agreed.“Can invites also go out over Slack?”Fix: trade it. Something moves out, or the date moves.Discovered scope
The work was always there. Planning missed it.“What if the invited email already has an account?”Fix: ask these questions before the sprint, not during it.How to prevent scope creep
- Write down what done looks like, and what’s explicitly out of scope.
- Break the feature into parts across frontend, backend, data and rollout before anyone estimates it.
- Ask the edge-case questions up front. The checklist below is a start.
- Get sign-off from everyone affected: product, design, the tech lead, and billing or support when it touches them.
- Estimate in ranges, so every addition has a visible cost in days. How range estimates work.
- Trade, don’t add. When something new comes in mid-sprint, decide what moves out or which date moves, and write the decision down.
Checklist
Questions to ask before every sprint.
Each one is a part of the work that usually turns up mid-sprint if nobody asks.
Backend
- What happens if the same request arrives twice?
- Which endpoints need rate limits?
- What runs in the background, and what happens when it fails?
Frontend
- What does the screen show while loading, when empty, and on error?
- What happens on a double-click, or a slow network?
- Can it be used with a keyboard and a screen reader?
Data
- Does this need a migration on a large table?
- What happens to existing rows? Is there a backfill?
- When a user is deleted, what happens to this data?
Shipping it
- Is it behind a feature flag, and how do you roll back?
- What will you log, and what should wake someone up?
- Who needs to know: support, docs, sales?
Sprint scope change: what to do mid-sprint
In short: When scope changes mid-sprint, protect the sprint goal first. Size the new work, and if it fits the goal, swap it in by taking out work of similar size. If it doesn’t fit, put it in the backlog for the next sprint. Write the decision down. If the goal itself is obsolete, the product owner can cancel the sprint.
The Scrum Guide doesn’t say sprint scope can never change, and it doesn’t define a change request process. Here’s what it does say:
“No changes are made that would endanger the Sprint Goal;”
“Scope may be clarified and renegotiated with the Product Owner as more is learned.”
“A Sprint could be cancelled if the Sprint Goal becomes obsolete. Only the Product Owner has the authority to cancel the Sprint.”
So scope can change during a sprint, as long as the change doesn’t endanger the sprint goal. Before you act, check which kind of change it is. If it’s discovered scope, the work was always there and the goal may depend on it. If someone wants more than was agreed, it needs a trade-off, or it becomes scope creep.
5 steps
- Check it against the sprint goal. Does it help, hurt or not matter?
- Size it the way you sized the sprint, in story points or a best, likely and worst range in days. If it’s too big to size, break it down first.
- Decide: swap, defer or stop. If you swap, name what comes out, of similar size. If you defer, it goes in the backlog for a later sprint. Stop only if the goal itself no longer makes sense.
- Decide who agrees. The product owner owns the backlog order. The developers own the sprint backlog and say what fits. Stakeholders can ask and are told the outcome, but they don’t make the call.
- Write it down in the change log below, and tell the people who rely on the date.
Urgent bugs and incidents
Treat a production incident as an interrupt. Fix it, log it, and count the time so the next sprint’s capacity is realistic. The sprint planning template has a place for capacity. If the fix puts the sprint goal at risk, go back to step 3.
When to cancel a sprint
Rarely. The Scrum Guide names one reason, the sprint goal becoming obsolete, and gives the decision to the product owner alone. A new request or a slipped estimate doesn’t make a goal obsolete. A change in business direction might.
The guide doesn’t say what happens to the work next. A common approach: keep what’s finished and meets your definition of done, put unfinished items back in the product backlog, and plan a new sprint around a new goal.
Example
The team invitations sprint is under way when someone asks. The sprint goal is “let admins invite teammates by email.”
- GoalSlack doesn’t help admins invite teammates by email, so it doesn’t fit the goal.
- SizeBest 2, likely 3, worst 5 days. That’s as big as the seat-limit work the goal still needs.
- DecisionDefer. Swapping it in would push out seat limits, so it goes in the backlog for the next sprint.
- LogOne new row in the change log: Defer, nothing moved out. The person who asked hears when it could be picked up.
Sprint scope change log
Copy this table into your doc or tracker and add one row per change.
| Date | Change requested | Requested by | Fits sprint goal? (Yes/No) | Size (points or best/likely/worst days) | Decision (Swap / Defer / Cancel sprint) | What moved out | Decided by | Who was told |
|---|---|---|---|---|---|---|---|---|
On a small screen, swipe the table sideways to see every column.
Deferred work comes back as something to plan. Run it through the feature planning checklist before it goes into a sprint.
FAQ
Scope creep and sprint scope questions
What is scope creep?
Can you change the scope of a sprint?
Who can add work to a sprint?
What should you do when a stakeholder asks for something mid-sprint?
When should a sprint be cancelled?
Is a scope change the same as scope creep?
Catch it in planning, not mid-sprint.
SprintProof’s AI planner asks these questions for every feature, from a check library written by engineers. Each answer becomes a part with its own estimate, or a decision your team can see later.