# How should a project change request process work?

A change request process that keeps momentum: explore before committing, size the impact, route by authority and record the decision, with a worked example.

Canonical: https://project.responseshift.academy/blog/change-request-before-commitment/

By Antonio Kirilov, Founder, Response Shift · Published 31 October 2026 · Updated 13 November 2026

## The request may be good and still need a decision

In a fictional office move, a facilities manager asks for a second entrance with card access, because it would ease congestion at nine o'clock. The idea is sensible. In a project meeting the team agrees, and the installer starts configuring the extra door. Three weeks later the sponsor asks why the access budget is €6,000 over and why go-live has slipped by a week.

Nobody did anything wrong in spirit. The problem is that the team committed before anyone with the authority to spend the money or move the date had seen the cost. A change request process exists to stop that from happening without stopping good ideas.

## What a change request process does

A working process has four steps, and each can be short:

1. **Capture.** Write down what is being asked, by whom and why. One paragraph is enough.
2. **Assess.** Estimate the effect on date, cost, scope, quality and risk. Say what you would need to find out to firm up the estimate.
3. **Decide.** Route the request to whoever has authority for that size of change. Record the decision, the conditions and the reason.
4. **Update and tell.** Change the plan, the budget, the scope statement and the schedule, and tell the people who act on them.

NASA's Systems Engineering Handbook describes the same discipline in the language of configuration management. Changes are made against an approved baseline, and a configuration control board decides on them. The handbook specifies that the board is chaired by someone with change authority for the programme or project, and that its members represent stakeholders with the authority to commit the teams they represent. The record kept includes the change requests, their disposition and the rationale for each disposition.

That last point matters most for smaller projects. The decision and its reason, written down, are what stop the same change from being argued again next month.

## Explore without committing

The biggest friction in change control is the fear that writing a request slows everything down. The answer is to separate exploring a change from committing to it.

Exploring is cheap and usually inside the team's authority: clarifying the need, asking a supplier for a quote, sketching an alternative. Committing spends money, moves dates or changes what will be delivered, and needs the right approver. Tell the requester plainly which one is happening: "We'll get a quote and an installation date by Thursday. Nothing is approved until the sponsor sees it."

This also helps the requester. They can compare adding the change now, adding it later or living with a workaround, with the consequences visible.

## Route by size, not by default

Not every change needs the sponsor. Most organisations that run IT services use change types from ITIL, the IT service management framework. Atlassian's summary describes standard changes as low-risk, commonly repeated and pre-approved, and normal changes as non-emergency changes without a defined, pre-approved process. Emergency changes, such as dealing with a server outage, sit in a third group. Projects can borrow the idea with simple thresholds agreed at the start:

| Change size (example) | Who decides | Record |
|---|---|---|
| No effect on date or budget; within agreed scope | Project lead | Note in the change log |
| Up to 5% of remaining budget, or up to one week, no effect on acceptance criteria | Project lead with the budget holder | Change log entry with impact |
| Above those limits, or any change to acceptance criteria or go-live | Sponsor or project board | Change request with options and recommendation |

The thresholds are illustrative. Agree yours with the sponsor before the first request arrives, and put them in the project's governance notes. Without them, every request either goes to the top, which slows everything, or goes nowhere, which is how the second entrance got built.

## A worked change request

Here is the fictional second-entrance request, written up for the sponsor.

```
Change request CR-07: Second card-access entrance          Raised: 3 Nov
Requested by: Facilities manager
Reason: Main entrance queue at 08:50–09:10 is about 6 minutes on
        busy days (observed 3 days last week).

Impact assessment
  Cost:      +€6,000 (installer quote, received 5 Nov)
  Schedule:  +5 working days if done before go-live
  Scope:     adds one door to the access system; no other change
  Risk:      installer capacity is already the critical path

Options
  A. Add now: go-live moves from 2 Dec to 9 Dec, +€6,000.
  B. Add after go-live: no delay; +€6,400 (separate visit); queue
     persists for about four weeks.
  C. Do not add; stagger arrival guidance for the first month.
     No cost; review queue times after four weeks.

Recommendation: B. The lease on the old office ends 5 Dec, so a
     go-live delay is costlier than four weeks of queuing.

Decision (sponsor, 7 Nov): Approved option B, capped at €6,500.
     Installer to quote a fixed date in January.
Updated: budget, scope statement v1.3, installer schedule. Told: facilities,
     installer, reception.
```

Three things make this request useful. The options include doing nothing differently (option C), which keeps the comparison honest. The decision records a condition, the €6,500 cap, which the team must respect if the quote moves. And the final line records who was told, because a change recorded in a log but missing from the installer's schedule has not really happened.

## What the requester should hear back

The person who raised the request is often forgotten once the decision is made. Close the loop within a day: what was decided, why, and what happens next. If the answer is no or not yet, say what would change it, such as a smaller version, a later phase or new evidence. A requester who hears a clear reason will usually raise the next idea through the process. One who hears nothing will go straight to the installer next time.

## Signs your change process is not working

- **Work starts before the decision.** Check whether any change in progress lacks a recorded approval.
- **Requests sit unanswered.** A request without a decision date tends to become an implicit yes. Give each one a date by which it will be decided.
- **The baseline keeps moving quietly.** If the approved plan changes without a change log entry, variance disappears from reports while the commitment shifts.
- **Every request goes to the sponsor.** Agree thresholds so small changes stay with the team.

## When the change is urgent

Some changes cannot wait for the next board. A safety issue, a regulatory deadline or a supplier failure may need action today. The process should bend for these without disappearing.

Agree in advance who can approve an urgent change and up to what limit. Let them act, then record the request, the decision and the reason within a day or two, as if it had gone through the normal route. The record matters even more here, because urgent decisions are the ones people later remember differently.

Watch how often the urgent route gets used. If one change in three is urgent, either the project is in trouble or the normal route is too slow, and both deserve a conversation with the sponsor.

## Keep the change log useful

A change log earns its keep when someone can answer three questions from it in a minute: what has changed since the baseline, what is still waiting for a decision, and what did the approved changes cost in total. Keep one row per request with its status, the decision date, the approver and the effect on date and cost. Add a running total of approved cost and schedule changes at the top. When the total grows past what the sponsor would expect, that is a conversation to have before the next request, not after it.

## Try it with a small request

Take a change that came up on a project you know, or invent one, and spend fifteen minutes writing it up:

1. One paragraph: what, who asked and why.
2. The impact on date, cost and scope, with what you are unsure about.
3. Three options, including one that changes nothing.
4. Your recommendation and who has the authority to decide.

Ask someone to read it and tell you which option you recommend and what waiting would cost. If either is hard to find, simplify. The [change request template](/project-change-request-template/) has these fields ready, and the [scope statement template](/scope-statement-template/) shows where approved changes should be reflected.

To practise writing a change request with options, try the open exercise [Hospital laboratory upgrade: change request](/project-management-exercises/hospital-laboratory-upgrade-change-request/).

## Sources

- [NASA Systems Engineering Handbook, 6.5 Configuration Management](https://www.nasa.gov/reference/6-5-configuration-management/)
- [Atlassian: IT change management and ITIL change types](https://www.atlassian.com/itsm/change-management)

## About the author

Antonio Kirilov is the founder of Response Shift and a Senior Product Manager at a B2B SaaS company, with more than five years in product management. He built Product Mastery and Project Mastery to train product sense and project leadership like a muscle, using AI and immersive simulations of real work to help people become the professionals AI cannot replace.

[Antonio Kirilov on LinkedIn](https://www.linkedin.com/in/antonio-kirilov)

## Continue

[Project practice week](https://project.responseshift.academy/practice-week/)
