How to handle scope creep when the project is already over

The advice on preventing scope creep assumes you are at the start of a project. This page is for week nine, when the hours are 30% over, the client has a list of six more things, and your lead developer has stopped making eye contact in standups.

You are not going to recover everything. The aim of a reset is to stop the loss from growing, recover what can fairly be recovered, and finish the project with the client still willing to take your calls. For the full picture of causes and costs, start with the scope creep pillar. For what to do this week, read on.

Step 1: stop the bleeding today

Before any conversation with the client, change one thing internally. Tell your team that from this morning, no request that is absent from the task list gets worked on until you have seen it. Requests go into a single list. The team’s answer to any new ask is the same sentence:

Good idea. I will add it to the list and we will come back to you with what it involves.

That sentence commits to nothing and offends nobody. It buys you the two days you need for the next step.

Do not announce a new policy to the client yet. You do not have your numbers.

Step 2: audit what has been added

Go back to the signed scope and compare it with what has been built and what has been asked for. Use the chat history, email, meeting notes, and your time tracker. Build one table.

In this example the four completed extras total 47 hours, and another 12 hours have gone into item 5. That is 59 hours. At $150 an hour it is $8,850 of work already delivered and never priced. The open work is the remaining 18 hours on item 5 plus 56 hours for items 6 and 7: 74 hours, or $11,100.

Two things come out of this exercise. The first is the number. The second is the pattern. Look at the “who asked” column. If four different people on the client side are making requests, your problem is the absence of a single approver, and no amount of pricing will fix it until that changes.

Be honest in the audit about items that were your fault. If something took longer because you underestimated it, that is your overrun and it stays off the list. A list padded with your own estimating errors will be picked apart, and the genuine items will go down with it.

Step 3: decide what to concede

You will write off most of what is already finished. I know that stings. Here is the reasoning.

The client was never told those items cost money. They agreed to nothing. If you send an invoice now for work completed weeks ago, you are asking them to pay for a decision they did not know they were making. Some will pay. Most will feel ambushed, and the remaining weeks of the project will be miserable. The legal position is usually weak too, especially if your contract says changes need written approval, because you broke that clause first.

So the default split is:

• Finished items: write off, visibly. Put the value on paper.
• Half-built items: charge for the remaining work only.
• Requested items not yet started: full price, or they wait.

There is one exception. If a single finished item is large (say, more than 10% of the project fee) and the client clearly knew it was an addition, it is reasonable to raise it and propose splitting the cost. Raise it as a question.

The concession is worth more than the money. It is what you trade for the new rule. A client who has just been handed $8,850 of free work in writing is in a poor position to object to pricing the next request.

Step 4: have the conversation

Do this on a call with the person who controls the budget. Email is the wrong medium for the first pass because tone gets lost. Send the table an hour before, with one line: “Ahead of our call, here is a summary of where the project stands against the original scope.”

The opening

I want to take ten minutes to reset how we handle additions. Since we signed, the project has picked up seven items that were outside the original scope. That is normal. You have learned things as the work took shape, and most of the ideas are good ones. I should have flagged each of them at the time, and I did not. That part is on us.

Taking responsibility for the process failure is accurate, and it removes the client’s need to defend themselves.

The concession

Four of the seven are already done and a fifth is partly built. That comes to about $8,850 of work. I am not going to bill for it. I will send a change record showing them at zero so we both have a clean picture of what the project now includes.

The remainder

Three are still open: finishing the second language, single sign-on, and the weekly email. Together they are $11,100 and would move launch by about two weeks. You have options. We can do all three. We can finish the language work now, which is $2,700, and schedule the other two as a phase after launch. Or we can swap: if single sign-on matters more than the reporting module in the original scope, we can trade one for the other.

Always arrive with options. One number is an ultimatum. Three options is a budgeting decision, and clients are comfortable making those.

The rule

From here on, when something new comes up, I will send you a one-page change request within two business days with the cost and the effect on the date. We start when you approve. That way you are never surprised by an invoice and we are never guessing.

Then stop talking and let them respond. The form and the steps behind that rule are on the change request process page.

Responses you will hear

“I thought that was included.” Go to the document. “I understand why. Here is the line in the scope. It lists two user roles, and this is a third. I should have said so when it came up.” If the scope is ambiguous on the point, concede it. An ambiguity in a document you wrote is yours to pay for.

“It is only a small thing.” Agree, and price it anyway. “It is small. About two hours. I will put it on a change request so it is on the record, and if it is under an hour I will mark it at zero.”

“We do not have any more budget.” Take that at face value and offer the swap. “Then the fee stays where it is, and we choose what matters most. If the weekly email goes in, which of these comes out?”

“Your competitor would not charge for this.” Stay level. “Possibly. What I can promise is that you will always know the cost before we start, and the final invoice will match what you approved.”

“Fine, but I am not happy.” That is an acceptable outcome. A client who is mildly annoyed and paying is in better shape than one who is delighted and unprofitable. Annoyance fades once the process produces two or three clean, fast change requests.

Step 5: put it in writing the same day

Send a recap within a few hours. List the conceded items at zero, the client’s decision on each open item, the new launch date, and the rule going forward. Ask for a one-word reply.

Please reply “confirmed” so we both have a record.

If your contract has a change clause, the recap plus the reply is your written approval. If it does not, this email now does that job for the rest of the project. For your next project, fix the source with a proper scope of work that has an exclusions list and a change clause.

When to stop work

There are three different things people mean by “stop work”, and they have very different consequences.

Stop work on unapproved items. Do this immediately and without ceremony. You are entitled to decline work nobody has agreed to pay for. Keep delivering the approved scope at full pace so that nobody can say the project stalled.

Pause the project. This is a formal step. Use it when the client will neither approve nor decline the pending changes and the undecided items block the remaining work, or when an invoice is overdue beyond the period in your contract. Check what your agreement says about suspension. Most well-drafted ones allow you to pause for non-payment after written notice. Send the notice in plain terms:

We are waiting on a decision about change requests 5, 6, and 7, sent on the 3rd. The remaining work depends on those decisions. If we do not have them by the 17th, we will pause the project and reschedule the remaining work once we hear from you. The current launch date will no longer apply.

End the engagement. This is the last resort, for a client who refuses every change request, refuses every swap, and continues to demand the extra work. Before you get there, deliver exactly what the original scope says and nothing else. That is often enough to end the standoff. If it is not, read the guide to firing a client before you send anything, because how you exit affects whether your final invoice gets paid.

Never stop work silently. A team that quietly slows down on a project while the owner stews is the worst of every option. The client gets a late project and no explanation, and you lose the moral high ground you needed for the money conversation.

What to change for the next project

A reset mid-project is damage control. Each one should leave you with a specific fix to your documents. Look at your audit table and ask what would have caught each line.

• Additions nobody recognized as additions point to a missing exclusions list.
• Requests from many people point to no named approver.
• Extras your team started unprompted point to a team that has never read the scope.
• A stakeholder who appeared late with new requirements points to a kickoff they did not attend. A structured kickoff agenda puts the sponsor in the room for the scope walkthrough.

The Scope hub lays out how the three documents fit together. If you want to see how the same failures play out across different kinds of work, the scope creep examples page pairs each one with the clause that would have stopped it.