A project can finish on time and still leave your team exhausted. Another may miss its launch date but produce a strong result and a healthier client relationship than expected. Looking only at the final outcome hides what happened along the way—and makes it harder to improve the next engagement.
An agency project postmortem is a structured review of how work was sold, planned, delivered, and handed over. Its purpose is not to decide who was at fault. It is to identify which conditions helped or hindered the work, then change a process, decision, or working agreement where that change is likely to matter.
Done well, a postmortem gives a small agency a practical answer to a recurring question: what should we keep, change, or stop doing on the next project?
Choose the right project and the right moment
Do not reserve postmortems only for disasters. If you review only troubled projects, the process can feel like an investigation and you will miss useful examples of work that went smoothly. Review a mix: a project that exceeded its budget, a successful launch, a difficult client relationship, or a project that involved a new service or team structure.
For a short engagement, meet soon after delivery, while the details are fresh. For a longer project, hold a brief review at a meaningful milestone and a final one after handoff. Avoid waiting until every invoice, support request, and memory has faded. If the client relationship is sensitive, separate an internal review from any client-facing debrief.
Prepare evidence before the meeting
Ask participants to bring observations, not verdicts. A project manager might bring the original estimate, timeline changes, meeting notes, and actual hours by phase. A creative lead could bring the feedback history and approval dates. The account lead might note changes in client stakeholders or decision-making.
Use a small, consistent evidence set. Useful items include:
- The proposal, scope, assumptions, and agreed definition of done.
- Planned versus actual timeline, including approval and dependency delays.
- Budget or hours by phase, where the team tracks them reliably.
- Change requests, revision rounds, and significant decisions.
- Client feedback, team feedback, and outstanding handover or support issues.
Check what the records can and cannot tell you. For example, a timesheet may show that design took longer than estimated, but not whether the estimate was unrealistic, the brief changed, or the team had to redo work after late feedback. Treat the data as a prompt for questions, not a verdict.
For a useful reference point, the Scrum Guide describes a retrospective as an opportunity to plan ways to increase quality and effectiveness. An agency does not need to use Scrum to apply the underlying idea: review the work, then agree on a practical improvement.
Run a focused, blameless discussion
For a small project, 45 to 60 minutes is often enough. Invite the people who can explain important decisions and influence future work. Keep the group small enough for honest conversation; a large audience can make people cautious. A neutral facilitator helps, particularly when the project lead is also involved in the problems being discussed.
Start with the intended outcome
Restate what the agency and client expected: deliverables, timing, budget, and any important assumptions. If those expectations changed, note when and why. This prevents a review from judging the team against a goal that was never actually agreed.
Map the work, not the personalities
Walk through the project in sequence: qualification and sale, kickoff, production, review, launch, and handoff. Ask what the team expected at each stage, what happened, and what helped explain the difference. Questions such as “When did we first know the deadline was at risk?” are more useful than “Who caused the delay?”
Make it safe to surface inconvenient facts. A designer who raised a concern about unclear feedback but was overruled is describing a process signal, not making an excuse. Likewise, “the client was difficult” is too broad to guide a change. Ask what specifically happened: Was there no single approver? Did feedback arrive in separate, conflicting rounds? Was the decision owner unavailable?
Separate causes from symptoms
Suppose a website launch ran two weeks late. “The team underestimated development” may be part of the explanation, but check the sequence. Did content arrive late? Were integrations discovered after the estimate? Did the client add pages without a scope discussion? Several causes can combine. A postmortem should describe the chain of events without reducing it to one convenient explanation.
When a problem involves scope or review, compare the project’s working agreements with your actual process. Your team may benefit from clearer revision rounds and feedback rules, or from defining how to respond when the work starts to move beyond its original scope.
Turn observations into changes the team can test
A postmortem that ends with “communicate earlier” has not yet produced an operational improvement. Translate each important finding into a specific action, an owner, and a point at which the team will check whether it worked.
For example, the team finds that a campaign launch was delayed because the client supplied approved copy after design had begun. A vague action is “get content sooner.” A usable change might be: “For projects with client-supplied copy, the producer will confirm a named content owner and a delivery date at kickoff. If copy is late, the project lead will show the timeline impact before the next production phase begins.”
This is also a chance to improve the start of future engagements. If project teams repeatedly discover missing access, unclear approvers, or unconfirmed dependencies after work begins, review the kickoff and onboarding steps. A structured agency project kickoff agenda can make those questions visible before production.
Prioritize a few actions, rather than creating a long list that nobody maintains. Choose changes that address a repeated or costly problem, are within the agency’s control, and can be checked on another project. Record the action where the relevant team will see it: a checklist, estimate template, project-management workflow, or sales handoff—not only in meeting notes.
Close the loop on the next project
Assign one person to maintain the action list. At the next project kickoff or internal planning meeting, confirm that the change is in place. After the next comparable project, ask whether it helped. The goal is not to prove that every change worked; it is to learn whether the process became clearer or the problem became less likely.
If the change did not help, revise it or remove it. A new approval step, for instance, might reduce rework but add so much delay that it is unsuitable for small, fast-turnaround jobs. Keep the improvement proportionate to the risk and the type of work.
A simple agency project postmortem template
- Project: What was delivered, for whom, and by which team?
- Intended outcome: What scope, timing, budget, and assumptions were agreed?
- What happened: What went well, what was difficult, and when did key changes occur?
- Evidence: Which dates, hours, approvals, feedback, or decisions support the observations?
- Causes: What conditions or process gaps contributed to the outcome?
- Actions: What will change, who owns it, and where will the change live?
- Follow-up: When will the team check whether the change helped?
Keep the record brief enough to reuse. A one-page summary with a few concrete actions is more valuable than a polished report that sits unread. Over time, patterns across reviews can point to changes in estimating, client qualification, staffing, or delivery—not just fixes for one project.
FAQ
Should the client attend an agency project postmortem?
Usually, hold an internal review first. A client debrief can be useful for shared issues such as approvals or handoffs, but it is not the right place to discuss internal staffing, pricing assumptions, or team performance. Decide what you want to learn and share accordingly.
How long should a postmortem take?
For a straightforward project, 45 to 60 minutes is a reasonable starting point. A complex engagement may need a longer session or separate reviews by phase. Keep the discussion focused on evidence and decisions rather than recounting every task.
What if the project went well?
Review it anyway. Identify what made delivery smooth—such as a clear approver, timely content, or a realistic schedule—and decide which parts are repeatable. Successful projects can reveal useful practices just as troubled ones reveal risks.
How do we keep the discussion from becoming a blame session?
Use a facilitator, discuss decisions and conditions rather than character, and ask what information or process would have helped at the time. Be candid about individual choices, but connect them to the context and system in which they occurred.
How many actions should we take away?
Start with one to three meaningful changes. Give each an owner and a follow-up date. If the list is too large to maintain or apply to the next project, prioritize before closing the meeting.