Handle New Feature Requests Without Scope Creep
When a new feature request arrives mid-project, treat it as a scope decision, not an automatic addition. Capture the request, test it against the original goal, estimate the change, and let one named owner decide. Defer the rest with a written reason.
Start with the request, not the solution
Ask the requester to write one paragraph: what problem does this solve, who experiences it, and what happens if nothing changes. A request that cannot name a user or a workflow is usually a preference, not a requirement. Keep this note in a shared log so the same idea does not return three times without context.
Suppose a client asks for booking reminders in a scheduling system that is already in build. Before discussing SMS or email, the team writes down the trigger, the recipient, and the expected action. That single note turns a vague wish into something that can be estimated, tested, or deferred with a reason.
Map the affected workflows
A new feature rarely lives alone. List every existing flow it touches: data entry, notifications, permissions, reporting, and any downstream system. If reminders are added, someone must decide who can edit reminder text, what happens when a booking is cancelled, and whether reminders appear in the activity log.
This mapping is where hidden cost appears. A feature that looks like one screen often changes three. Write the affected list next to the request and share it with the decision owner before any estimate is given.
Write one acceptance example
An acceptance example is a short, concrete scenario that the finished feature must pass. It names the starting state, the action, and the observable result. For booking reminders, a usable example might be: a confirmed booking two days away triggers one reminder to the customer, and the reminder appears in the booking history.
Keep the example small enough to test in one sitting. If the team cannot write a clear example, the request is not ready for estimation. This is the practical test: can a person who did not write the request follow the example and say whether it passed or failed.
Revise the estimate and name the decision owner
Every accepted request changes the estimate. Add the new work to the current plan, show what moves, and state the tradeoff plainly: this feature now, or the original milestone on time. Do not hide the shift inside a general buffer.
One person must own the decision. That owner may be the client sponsor, the product lead, or the account manager, but it must be a single name. When two people can approve changes, scope grows quietly. The owner receives the request, the affected workflow list, the acceptance example, and the revised estimate, then chooses: accept, defer, or reject.
Keep an explicit deferred list
Deferral is not rejection. A deferred list records what was considered, why it waits, and what condition would bring it back. This prevents the same request from being re-argued every week and gives the client a visible path for later phases.
Review the deferred list at each milestone. Some items become unnecessary once the core system is in use. Others become urgent. The list is a living document, not a graveyard.
A worked example
Suppose a clinic management build is halfway through when the client asks for automated booking reminders. The team writes the request note, maps the affected flows, and drafts an acceptance example. The estimate revision shows that adding reminders now would push the billing module by a defined period. The decision owner chooses to defer reminders to the next phase and records the condition: revisit after billing is live and reminder volume is understood.
The project keeps its original scope, the client keeps a clear record of the request, and the team avoids silent rework. No outcome is promised here; the value is in the decision being visible and testable.
Continue reading
A practical next step
Explore the related Devign work, then discuss your requirements before committing to a project scope.