Scope creep examples: seven projects and the clause that would have saved each
The common examples of scope creep are unlimited revisions, a late stakeholder, an integration that grows, uncounted content migration, an assumption that proves false, small favors that pile up, and post-launch support with no end date. Each case below shows the scope wording, the cost in hours and dollars, and the clause that would have prevented it.
Scope creep is easier to recognize in someone else’s project. The seven cases below are composites of situations I have seen repeatedly in agency and software work. The details are generic on purpose. The numbers use a blended rate of $150 an hour so you can compare them.
Each case has four parts: what the scope said, what happened, what it cost, and the clause that would have stopped it. For the definition and the full cost model, see the scope creep pillar.
1. The revisions that never ended
What the scope said. “Homepage and interior page designs, including revisions.”
What happened. The agency budgeted two rounds. The client’s marketing lead sent feedback, then the CEO sent different feedback, then the board saw it. Round six was approved. Feedback also arrived piecemeal, so each “round” was really three or four separate passes.
What it cost. Four unplanned rounds at about 12 hours each is 48 hours, or $7,200, on a $28,000 project. The launch moved five weeks.
The clause.
Each deliverable includes two rounds of revisions. A round is one consolidated list of changes, sent in writing by your project contact, within five business days of delivery. Further rounds are billed at $150 per hour.
The number matters, and so does the definition of a round. Without “consolidated” and “project contact”, two rounds turn into eight conversations.
2. The stakeholder who arrived in week seven
What the scope said. A detailed feature list for a customer portal, agreed with the client’s head of operations.
What happened. The head of sales joined a demo in week seven, having attended nothing before it. Sales needed account managers to see their own customers’ portals. That meant a new user role, a permissions layer, and changes to four finished screens.
What it cost. 60 hours, or $9,000. The team did the work because the request came from a senior person and sounded like something the portal obviously needed.
The clause.
Your project contact is [NAME]. Instructions and approvals count only when they come from your project contact in writing.
And a line in the deliverables table: “User roles: 2 (customer, administrator).”
The clause on its own would not have been enough. The sponsor and every department that would touch the portal needed to hear the role list read aloud in week one. That is what a structured project kickoff agenda is for.
3. The integration that became a sync engine
What the scope said. “Integrate the web app with the client’s CRM.”
What happened. The agency priced a one-way push: when a user signs up, create a contact. The client expected changes in the CRM to flow back to the app, with deduplication and a way to resolve conflicts. Both readings of the sentence were reasonable.
What it cost. The estimate was 24 hours. The two-way sync took 110. The 86 extra hours came to $12,900. The agency argued for a change order after the fact and settled for half.
The clause.
CRM integration: one-way push from the app to the CRM, on user signup and on profile update. Fields: name, email, company, plan. Excludes: data flowing from the CRM to the app, deduplication of existing CRM records, and historical backfill.
Integrations need four things written down: direction, trigger, fields, and what is excluded. A noun and a product name is an invitation.
4. The migration nobody counted
What the scope said. “Migrate existing content to the new site.”
What happened. The agency had looked at the main navigation and estimated about 40 pages. The old system held 1,300 blog posts, 400 PDF downloads, and a decade of press releases, with inline images hosted on a server that was being switched off.
What it cost. A scripted import plus manual cleanup came to 70 hours against 10 budgeted. 60 extra hours is $9,000.
The clause.
Content migration: up to 40 pages, moved manually. Migration of blog posts, documents, and media beyond this limit is excluded and can be quoted separately once an export of the existing content has been reviewed.
Any time the word “existing” appears in a scope, something needs counting. If you cannot count it before signing, exclude it and offer to quote once you have seen the export.
5. The assumption that was wrong
What the scope said. “Build a reporting dashboard using data from the client’s inventory system.”
What happened. Nobody asked for anything extra. The inventory system’s API returned stock levels and nothing else. Sales history, which six of the eight charts needed, was only available as a nightly file dropped on a server, in a format that had changed twice. The team built a file ingestion pipeline that was never in the plan.
What it cost. 45 hours, or $6,750. The client felt this was the agency’s problem, since the agency had quoted for a dashboard and the client still only had a dashboard.
The clause.
Our fee assumes that the inventory system provides all data listed in Appendix A through a documented API. If an assumption turns out to be wrong, we will tell you promptly and the difference is handled as a change.
This kind of creep has no requester, which makes it the hardest to raise without a written assumption. With one, the conversation is short: the assumption is on the page, here is what we found, here are the options.
6. The favors that added up
What the scope said. A clear, well-limited scope for a mobile app. Nothing wrong with the document.
What happened. The client’s product manager was friendly and responsive, and sent small requests directly to the developers in chat. Change a label. Add a field. Make the list sortable. Each was done the same day. Over fourteen weeks there were 34 of them.
What it cost. At an average of 1.5 hours each, 51 hours, or $7,650. Nobody decided to give that away. It also set an expectation: when the agency raised a change request for a larger item in week twelve, the client was surprised to be charged for anything.
The clause.
We do not start work on a change until your project contact has approved the change request in writing.
The clause was in the contract. What was missing was the routine. A change request process that logs every request, including the free ones at a 100% discount, would have shown both sides a running total by week four.
7. The launch that never finished
What the scope said. “Includes post-launch support.”
What happened. The site launched. For five months the client sent a steady flow of requests: a bug here, a new landing page there, a question about analytics, a change to the checkout. The agency could not tell where warranty ended and new work began, and neither could the client.
What it cost. About 6 hours a month for five months is 30 hours, or $4,500, plus the constant interruption to other projects. The larger cost was the lost retainer. This client clearly needed ongoing help and was getting it for nothing.
The clause.
For 30 days after acceptance we will fix, at no charge, any defect where a deliverable does not work as described in this scope. Requests for something to work differently from how it was described are changes. Support after the warranty period is available under a separate monthly agreement.
A defined warranty period gives you a natural moment to propose ongoing work. Setting the fee and allocation for that is covered under retainer pricing.
What the seven have in common
Add up the cost column.
| Case | Extra hours | Cost at $150 per hour |
|---|---|---|
| 1. Revisions | 48 | $7,200 |
| 2. Late stakeholder | 60 | $9,000 |
| 3. Integration | 86 | $12,900 |
| 4. Migration | 60 | $9,000 |
| 5. False assumption | 45 | $6,750 |
| 6. Small favors | 51 | $7,650 |
| 7. Open-ended support | 30 | $4,500 |
| Total | 380 | $57,000 |
Seven ordinary projects, 380 unpaid hours. No client in these cases acted in bad faith. Each asked for something that seemed reasonable from where they sat, and the document either agreed with them or said nothing.
The fixes fall into five groups, and a good scope has all of them:
- A number on everything that can multiply (cases 1, 2, 4)
- Specifics on anything named by a noun (case 3)
- A list of assumptions (case 5)
- A named approver and a written-approval rule that the team follows (cases 2, 6)
- A defined end (case 7)
The short-form scope of work template has a place for each one. Cases 3 and 5 also raise a pricing question. Work with that much unknown in it may be a poor fit for a fixed fee in the first place, and the comparison of fixed price and time and materials explains how to decide.
If one of these cases describes a project you are in right now, the documents come second. Start with how to handle scope creep mid-project, then come back to the Scope hub to fix the paperwork for the next one.
Common questions
- What is an example of scope creep?
- A client signs for a website with 'design revisions included'. They request six rounds of changes where the agency had budgeted for two. The fee stays the same and the agency absorbs roughly 50 extra hours. The scope allowed it because it never stated a number of rounds.
- What is the most common type of scope creep in software projects?
- An integration or feature described by its name alone. 'Integrate with the CRM' can mean a one-way push of a contact record or a two-way sync with conflict handling, and the second is several times the work. Without a stated direction, list of fields, and number of endpoints, the client's version wins.
- Is it scope creep if the client did not ask for anything new?
- It can be. When an assumption behind the price proves false, such as an API being undocumented or content arriving in the wrong format, the work grows without any new request. A scope that lists its assumptions turns that growth into a priced change.
- How do I use these examples to check my own scope?
- Take your current scope document and ask of each example whether your wording would have stopped it. Look for a number on every deliverable that can multiply, an exclusions list, a list of assumptions, a named approver, and a change clause that requires written approval before work starts.