Skip to content

07 · Change Management for Projects

"Change management" means two genuinely different disciplines, and conflating them causes real damage. Change control is the governance process for deciding whether a requested modification to the baseline gets approved. Organisational change management is the people-side work of getting the delivered product actually adopted. A project can execute flawless change control and still fail because nobody uses what it built. This module covers both, in that order.

Part 1 — Change control

The purpose of change control is not to prevent change. It is to ensure that every change is assessed for its full impact and approved by someone with the authority to accept that impact before it is built.

Uncontrolled change has two forms worth naming separately:

Term What it is Root cause
Scope creep Uncontrolled additions from stakeholders, unassessed No baseline, or no change process
Gold plating The team adds unrequested polish Team believes it knows better than the spec

Both bypass the process. Both consume budget that was allocated to something else. Gold plating is the more insidious because it feels like generosity.

The change request template

Field Content
CR ID CR-047
Title Add SSO via corporate identity provider
Raised by / date Head of Security, 12 March
Type Scope addition
Description Replace local username/password with corporate SSO for all users
Business justification Security policy mandates SSO for systems holding customer data by Q4
Cost impact +$62,000 (3 dev-weeks, IdP licence, security retest)
Schedule impact +9 days on the critical path (integration → UAT)
Scope impact Adds 4 stories; removes the local password reset story (−2 days)
Quality impact Requires new penetration test before go-live
Risk impact New: IdP team availability. Closes: R7 credential-storage risk
Options considered (a) Full SSO now (b) SSO after go-live (c) Do nothing — breaches policy
Recommendation Option (a): the policy deadline precedes our go-live
Decision / date / approver Approved, 19 March, Steering Committee

The five impact rows are the entire value of the form. A change request that says "add SSO" with no impact assessment is not a change request, it is a wish — and approving it commits the project to a cost nobody quantified. Note that the assessment includes what the change removes (a story, a risk), not just what it adds.

The change control workflow

Step Who Action Output
1. Submit Anyone Raise CR with description and justification Logged CR, status Submitted
2. Log PM Enter in the change log, assign an ID Unique reference
3. Assess PM + leads Full impact analysis on cost, schedule, scope, quality, risk Completed CR
4. Decide CCB or delegated authority Approve / reject / defer / request more info Decision recorded
5. Update baseline PM Re-baseline schedule, budget, scope; version the plan New baseline
6. Communicate PM Tell everyone affected what changed and why Updated plan issued
7. Verify PM Confirm the change was implemented as approved Closed CR

Step 5 is the one most often skipped, and skipping it silently destroys your performance reporting. If a change adds $62,000 and 9 days but the baseline is never updated, then from the next reporting period onward your EVM figures (module 03) compare actuals against a plan that no longer exists — CPI and SPI both degrade, and the cause is invisible. An approved change that has not been re-baselined will show up as poor performance.

Approval authority

Not every change needs a committee. Publish the thresholds:

Change size Authority Target turnaround
< $5,000 and < 2 days, no scope change Project Manager 1 working day
< $25,000 and < 5 days PM + Product Owner 3 working days
< $100,000 and < 15 days Change Control Board Weekly CCB
Above that, or any change to project objectives Sponsor / Steering Committee Monthly, or called

Turnaround times are a genuine part of the design. A CCB that meets monthly and takes three weeks to decide will be bypassed by teams who need an answer, and once it is routinely bypassed it exists on paper only. Match the cadence to the pace of delivery, or delegate more.

The change log

CR Title Raised Cost Sched Status Decision date
044 Extra approval step in workflow 02 Mar +$8,000 +2 d Approved 07 Mar
045 Custom branding on reports 05 Mar +$14,000 +3 d Rejected 07 Mar
046 Extend data retention to 7 years 09 Mar +$21,000 +4 d Deferred to phase 2 14 Mar
047 Add SSO 12 Mar +$62,000 +9 d Approved 19 Mar
048 Additional Spanish localisation 20 Mar +$38,000 +12 d Under assessment
Approved to date +$70,000 +11 d

Keep the rejected and deferred rows. When someone asks in month 8 why the Spanish version does not exist, the log answers it. And the cumulative row is the number that goes on the status report: $70,000 and 11 days of approved change is a fact about the project that must be visible, or the eventual overrun looks like poor delivery rather than accepted scope growth.

Part 2 — Organisational change management

Delivering the product is not the objective; the benefit is realised only if people change how they work. The standard model is ADKAR — five states each individual must pass through, in order.

Stage The person must... Fails when Intervention
Awareness Know why the change is happening "Nobody told us this was coming" Sponsor communication, town halls
Desire Want to participate "I understand it, but it's worse for me" Address WIIFM, involve in design
Knowledge Know how to work the new way "I don't know how to do X now" Training, documentation, job aids
Ability Be able to do it in practice "I did the training, but I'm slow and error-prone" Floor-walking, practice environment, coaching
Reinforcement Keep doing it Quiet reversion to the old system Metrics, recognition, retire the old system

The diagnostic power of ADKAR is that it identifies which stage is blocked — and each has a different fix. If adoption is poor because people do not want the change (Desire), more training (Knowledge) is wasted money; the fix is addressing what they lose. Most failed rollouts respond to every adoption problem with more training, because training is the easiest thing to buy.

The most-skipped stage is Reinforcement. Teams celebrate go-live, disband, and three months later half the users are back on spreadsheets. If the old system is still available, some people will keep using it — a decommissioning date is a change management intervention, not an IT task.

Stakeholder resistance analysis

Group Current Target Main concern Intervention Owner
Branch staff (200) Resistant Supportive "Slower than the old screens at first" Practice environment 4 weeks early; super-users on the floor for 2 weeks Ops Lead
Team leaders (24) Neutral Advocate Accountable for team throughput during dip Brief on expected 3-week productivity dip; adjust targets Sponsor
Finance (12) Supportive Supportive Month-end close must not slip Dry-run a full close in parallel before cutover Finance Lead
Regional managers (6) Resistant Neutral Loss of local reporting customisation Show new self-service reports; agree 3 must-have local reports PM

Being explicit that some groups only need to reach Neutral matters. Trying to convert every stakeholder into an advocate wastes effort that should go to the group whose resistance actually blocks go-live. Note also the branch staff intervention: their concern is real, not irrational — the new system will be slower for them at first, and a plan that denies this loses credibility immediately.

Never assess a change request by cost alone

A $5,000 change that adds 9 days to the critical path is far more damaging than a $40,000 change that consumes float on a parallel path. Always assess schedule impact against the network (module 02), not against the effort estimate. The question is not "how much work is this?" but "does this move the end date?" — and only the critical path can answer that.

How It Actually Works

The change control workflow's approval-authority tiers exist to solve a queueing problem, not a political one: if every change request, from a one-line wording fix to a $200,000 scope addition, had to route through the same CCB, the board becomes the bottleneck and average change-approval latency grows without bound as request volume rises (the same throughput-vs- WIP relationship from Level 1 Module 9's Little's Law). Tiering by dollar/ schedule impact — a PM can approve changes under a threshold, only larger ones go to committee — is a load-balancing strategy that keeps the high-stakes queue short precisely by routing low-stakes requests around it. On the organizational-change side, resistance isn't a fixed trait of people, it follows a predictable adoption curve (Rogers' diffusion-of-innovation shape): roughly 2.5% Innovators, 13.5% Early Adopters, 34% Early Majority, 34% Late Majority, 16% Laggards — and the mechanical implication is that a rollout strategy targeting the wrong segment fails predictably: pushing hard on Late Majority resisters before Early Majority has visibly adopted wastes effort, because Late Majority's primary decision input is social proof from peers who've already switched, not the change argument itself — they need to see the Early Majority succeed, not hear the pitch again.

Exercise

Build a complete change management approach for a project of your own.

  1. Write your approval authority table with at least three tiers, giving monetary and schedule thresholds, the deciding body, and a committed turnaround time for each.
  2. Take a realistic change request for your project and complete the full template, including all five impact rows. Assess the schedule impact against your critical path — state whether it consumes float or moves the end date, and by how many days.
  3. Give the change at least two genuine alternative options and a recommendation with reasoning.
  4. Build a change log with at least six CRs including at least one rejected and one deferred, and compute the cumulative approved cost and schedule impact. Write the one sentence about it you would put in a status report.
  5. Pick one stakeholder group that will resist. Diagnose which ADKAR stage they are blocked at, justify the diagnosis, and design an intervention that targets that specific stage — then state why the obvious response (more training) would or would not work.
  6. Build the resistance analysis table for at least four groups, with current and target positions, and state explicitly which groups you are not trying to convert to advocates and why.