A client sends a message: “Could we put a comparison slider on the homepage?”
You add a task so you won’t forget. At the next planning session, it sits beside the approved work. A few days later, you build it.
No one made a deliberate decision about the addition. The task list made it for you.
For a fixed-price website project, one useful safeguard is to separate requests from scheduled work. You can do this in a spreadsheet, a document, or the project tool you already use. The important part is the distinction.
Give incoming requests somewhere to wait
Create a small request list outside the production queue. Record the request, who raised it, the result they want, and when they need an answer.
Give each entry a visible status: “needs clarification,” “decision pending,” “approved,” or “deferred.”
An acknowledgment should describe what happens next: “I’ve recorded the slider request. I’ll check the work and timing and come back with options before it goes into production.”
That lets you respond promptly without accidentally promising delivery.
Define the result before estimating the task
Consider the slider example. Does the client already have two matching images? Does the interaction need a usable alternative for visitors who cannot drag the control? What happens on small screens? Who will replace the images later?
These questions do not automatically make the feature expensive. They reveal what the request actually contains.
Write one short acceptance description. For example: “Visitors can compare two supplied product images on the homepage, using a control that works with keyboard and touch input.”
Then identify the design, implementation, content preparation, and testing needed. If a dependency is unknown, investigate it before promising a fixed price or date.
Keep corrections on a different path
A request list should not become a tollbooth for work you already owe.
If the approved homepage includes a working image comparison and your implementation fails to match that specification, record the issue as a correction. If the client requests a new interaction that was never agreed, assess it as an addition.
The reference point is the agreed deliverable. The size of the message, or how politely it is phrased, does not settle the question.
Put the decision beside the work
Before an addition enters production, record what is included, its price effect, its schedule effect, and who approved it.
If the client chooses a smaller version, schedule that version. If they choose to replace another deliverable, remove the displaced work from the active plan. If they defer it, keep it out of launch checklists.
Link the production task to that decision so neither person has to reconstruct the agreement from scattered messages.
Check the boundary once a week
During your weekly review, look for production tasks with no corresponding approval or original scope item. Resolve those gaps before more work accumulates.
Clients can keep bringing ideas. You can keep listening. A separate request list gives both of you room to decide which ideas belong in this project.
