How to run agency retrospectives: 4 formats

An agency retrospective is a structured review after a project, a client relationship or a quarter, where the team looks at what happened, why, and what it will change next time. Picture a hypothetical creative agency that closes a brand identity project 40% over budget, after two extensions and a near-loss of the client. Everyone is relieved it is over, so nobody schedules a retro. Six months later the same pattern repeats on a new client: an optimistic timeline, too few revision rounds scoped, a burned-out design lead. That is the agency-specific cost of skipping retros: not just lost learning, but the same expensive mistake repeated across the client roster. This guide gives you four agency-specific retro formats (project retros, client retros, profitability post-mortems and quarterly business retros) with templates and facilitation patterns that produce real change.

The Bottom Line:

• Agencies need four distinct retro types: project retro, client retro, profitability post-mortem, and quarterly business retro, each with a different agenda
• The 4Ls and Sailboat formats outperform Start/Stop/Continue for agency project retros; profitability retros need numbers-first formats
• Win/loss client retros surface the patterns behind renewals and churn while they can still be fixed
• Profitability post-mortems on larger projects are how estimation errors get caught and corrected
• A retro is worth nothing without owned action items, and action tracking is where most retros fail

The difference between agencies that compound their experience and those that repeat their mistakes is whether the retro is a meeting or a discipline.

Four retros, four jobs

Most agency advice treats "retrospective" as one meeting. In reality, agencies need four different retros with different agendas, attendees, and outputs.

Running these as one undifferentiated "retro" is why most retros produce vague action items that nobody implements.

Retro type 1: The project retro

This is the workhorse: a retro after every project of meaningful scale, focused on team process and execution.

When to run it

Within one week of project end. Memory fades fast; details and emotional context get lost past day 10.

The 4Ls Format (recommended for project retros)

Most agencies default to Start/Stop/Continue, which surfaces tactical issues but misses emotional and aspirational content that often holds the real learning. The 4Ls (Liked, Learned, Lacked, Longed For) works better for agency project retros.

• Liked. What did you appreciate about how this project ran?
• Learned. What new knowledge or skills emerged?
• Lacked. What was missing or insufficient?
• Longed For. What do you wish we had?

The Lacked and Longed For prompts surface the structural issues (capacity, tooling, briefs) that produce action items worth implementing.

The 60-minute project retro agenda

Minutes 0 to 5: Review previous action items. Did last project's actions get implemented? Celebrate completions, retire stalled items, understand why.

Minutes 5 to 10: Set the stage. One-word check-in. Brief context on the project being reviewed.

Minutes 10 to 25: Silent brainstorming. Each person writes one sticky/card per item across the 4Ls. Quiet, parallel, individual.

Minutes 25 to 35: Group and vote. Cluster similar items. Each person gets 3 votes to allocate to items they want to discuss.

Minutes 35 to 55: Discuss top items. Work through highest-voted items. For each: name the issue, explore root cause, generate action item with owner and deadline.

Minutes 55 to 60: Close. One sentence from each person: what is your single takeaway?

Action item discipline

A project retro that produces 7 action items typically gets 1 implemented. A retro that produces 2 to 3 typically gets all of them done. Limit ruthlessly.

Bad action item: "Communicate better with clients."

Good action item: "By next Friday, Jamie will draft a one-page client communication brief template and share for team review."

Specific, owned, dated, achievable.

Track action items in the project management tool alongside regular work, with the source retro date tagged. If you cannot track them there, you have already lost.

Retro type 2: The client retro

A joint retro with the client. Done well, these strengthen relationships and surface honest feedback that almost never surfaces in business-as-usual conversations.

When to run client retros

• After major projects, before scope renewal conversations
• Annually for retainer clients
• After challenging engagements where there is unspoken tension to clear

When NOT to run them

• When the client relationship is fundamentally adversarial: a retro will not fix that
• When internal team dynamics are unresolved: handle internally first
• For sprint-level work: clients do not need that granularity

The structured client retro agenda (75 minutes)

Pre-meeting (sent 5 days ahead): Survey with 6 questions on agency communication, deliverable quality, process, partnership feel. Both sides complete.

Opening (10 min): Acknowledge wins. What worked, why it mattered to the client's business.

Process feedback (15 min): Client speaks first: how did agency processes serve them? What got in the way? Agency listens without defending.

Deliverable feedback (15 min): What landed? What missed? Where did the work meet or fall short of expectations?

Agency observations (15 min): Agency raises what made the work harder or easier from their side. Framed constructively: "we observed that approval rounds extended timeline by 8 days; here is what we'd do differently."

Joint improvements (15 min): What changes (on both sides) would make the next quarter better?

Close and next steps (5 min): Document commitments from both sides, schedule check-in.

The win/loss client retro framework

For agencies running BD, every won and every lost proposal deserves a 30-minute retro within 7 days of the decision.

Win retro questions:

• What from our proposal resonated most?
• Who was our champion on the client side?
• What was our biggest competitive differentiator (in their words)?
• What did we underprice or under-scope?
• What expectations should we manage now?

Loss retro questions (call the prospect; most are willing if asked once):

• Who won? Why?
• Where did we lose them: proposal, pricing, fit, references?
• What in our process was clunky?
• What feedback would help us in the future?

A 20-person agency that runs win/loss retros for 12 months typically improves close rate by 6 to 12 points. The pattern recognition is where the gold lives.

Retro type 3: The profitability post-mortem

This is the retro most agencies skip and most need. After every project over a $25K threshold, run a 45-minute post-mortem on the project economics.

Required pre-work

• Project budget at quote
• Actual hours by role
• Actual COGS (contractors, software, third-party costs)
• Gross margin at quote vs. actual
• Revision rounds count
• Days over/under timeline

The 45-minute profitability post-mortem agenda

Minutes 0 to 5: Numbers review. Quote vs. actual, line by line. No discussion, just the facts.

Minutes 5 to 15: Variance analysis. Where did hours go over? Which line items got blown? Specific, evidence-based.

Minutes 15 to 30: Root cause. For each variance, what was the cause? Estimation error, scope creep, internal inefficiency, client behavior, team-side overruns?

Minutes 30 to 40: Estimation calibration. What should we have priced differently? Update the estimation model for the next similar project.

Minutes 40 to 45: Action items. Specific changes to scoping, pricing, or process for the next engagement of this type.

What you learn

Run 8 to 12 of these in a year and patterns emerge: a service line that is structurally underpriced, a project type that consistently overruns, a client size where margins do not work, a role you are systematically under-allocating.

The payoff comes mostly through better estimation, not by squeezing the team: each post-mortem shows where your quotes were wrong, so the next quote is closer.

For deeper estimation frameworks, see the agency project estimation guide.

Retro type 4: The quarterly business retro

A 2 to 3 hour leadership session at the end of each quarter, looking across all projects and accounts for systemic patterns.

The agenda

Hour 1: Numbers review. Revenue, gross margin, net margin, utilization, pipeline. By client, by service line, by team.

Hour 2: Patterns and themes. Looking across the quarter's project retros, client retros, profitability post-mortems: what patterns appear? What did we learn three or more times?

Hour 3: Strategic priorities. Three or four priorities for the next quarter, each with an owner and measurable outcome.

This is not a board meeting; it is an operating meeting. The output is two or three concrete changes that will be enacted over the next 90 days.

Facilitation patterns that actually work

The facilitator makes or breaks every retro. Here is what works at agencies specifically.

Psychological safety

Google's Project Aristotle research found that psychological safety was the most important of the five dynamics that set effective teams apart (Google re:Work). In agency contexts, this matters most when discussing project failure modes, and almost no agency naturally has it.

How to build it:

• "Blameless" framing: name systems, not people
• Senior leaders go first with their own mistakes
• Anonymous input options for sensitive topics (e.g., a pre-meeting Typeform)
• Explicit "what's said here stays here" rule for internal retros

Time discipline

Retros that run long produce diminishing returns. Past 90 minutes, people stop being honest. Timebox every section and enforce.

Driving to action

The most common failure mode: a 90-minute discussion that ends with "we should communicate better." Force specificity.

• Each action item gets one owner, one deadline, one success criterion
• Limit to 2 to 3 per retro
• Track in PM tool, not a Google Doc

Rotating facilitators

The same facilitator every time creates dependency and predictable patterns. Rotate among senior team members. The rotation itself improves the retro because each facilitator brings different patterns of attention.

Common retro failure modes at agencies

The Action Item Graveyard. Action items from previous retros consistently unfinished. Solution: open every retro by reviewing prior actions. If they are not getting done, the retro process is broken or the action items are wrong.

The Complaint Session. Retro becomes venting without action. Solution: facilitator redirects every complaint to "what specifically would improve this?"

The Blame Game. Personal criticism instead of process critique. Solution: facilitator pivots: "rather than who, let's talk about what in our process allowed this."

The Echo Chamber. Same two people dominate every retro. Solution: silent brainstorming, structured round-robin, anonymous tools.

Skipping When Busy. The first thing cut when work intensifies. This is exactly when retros are most needed: busy stretches accumulate process debt quickly. Even a shortened 30-minute retro is better than skipping.

Repetitive Issues. Same issues come up retro after retro. The problem is not awareness; it is execution. Escalate recurring issues to leadership and assign real resources (time, budget, process change) to solve them.

Tools and templates

Templates that should be standard

The Project Retro Template (one page):

• Project name, dates, team
• Liked / Learned / Lacked / Longed For (one column each)
• Top 3 themes discussed
• Action items: description, owner, deadline, success criterion
• Source retro date

The Profitability Post-Mortem Template:

• Budget at quote (hours, COGS, gross margin)
• Actuals (same)
• Variance % and root cause for each line
• Estimation calibration notes
• Action items

The Client Retro Template:

• Wins (client perspective, agency perspective)
• Process feedback (both sides)
• Deliverable feedback (both sides)
• Joint improvements committed (with owners)
• Next check-in date

Digital tools

• Miro or FigJam: Virtual whiteboards with retro templates
• EasyRetro / Parabol: Purpose-built retro facilitation
• Your project management tool: track retro action items as tasks on the project or client they came from

The retro library

Build a single repository (for example a Notion database or a shared wiki) of every retro the agency has run, searchable by client, service line, or theme. This is where pattern recognition becomes possible.

Example: a six-month retro rollout

For example, a hypothetical 18-person digital agency adopting all four formats over six months might run:

• Project retros within a week of every project close
• Client retros after every larger project
• Profitability post-mortems on every project above its size threshold
• A quarterly business retro at the end of each quarter

Decide up front what you will measure: median revision rounds per deliverable, gross margin by service line, client renewals and the number of retro action items actually completed. Compare them after six months to see which retro types earn their time.

Building a retro culture

Process matters. Culture matters more. Three habits that build retro culture:

Non-negotiable scheduling. Block retro time at project kickoff, not after delivery. If leadership cancels retros for client work, the message is clear: improvement does not matter.

Visible improvements. When an action item produces a real result, call it out. For example: "Last quarter's retro action on handoff checklists cut design-to-dev rework by 80%. Here is the data."

Connect retros to outcomes. Show how retro improvements connect to agency-level metrics: margin, retention, on-time delivery. The connection makes the practice feel like investment, not overhead.

For deeper context on continuous improvement at agencies, see the agency operations and SOPs guide and the agency workflow optimization framework.

Frequently asked questions

How often should agencies run retrospectives?

Project retros within one week of every project close. Profitability post-mortems on every project over a $25K threshold. Client retros after major engagements or annually for retainers. Quarterly business retros at leadership level. Skip retros without a triggering event: they tend to produce repetitive feedback that nobody acts on.

How long should a retrospective take?

Project retros run 60 to 90 minutes. Sprint retros 30 to 60. Profitability post-mortems 45 to 60. Client retros 60 to 90. Quarterly business retros 2 to 3 hours. Past those times, honesty and energy drop fast, so strict timeboxing is what keeps retros productive.

Who should attend agency retrospectives?

Everyone who did the work on the project being reviewed, the project lead, and the account lead. For profitability post-mortems, add finance and a leadership-level reviewer. Exclude clients from internal retros; candor drops materially when clients are present. Run separate client retros for joint reflection.

What is the most common retrospective mistake?

Failing to track and follow up on action items. Retros that produce action items nobody implements quickly become venting sessions, and the team loses faith in the process. Assign one owner and one due date to each action item, and review prior actions at the start of every retro. If action items are not getting done, fix that before adding more.

How do you make retrospectives psychologically safe?

Establish norms upfront: blameless framing, focus on systems not individuals, anonymous input options where needed, no-retaliation rule. Senior leaders going first with their own mistakes signals the culture more than any rule. Without safety, retros surface only safe topics, and the real issues stay buried.

Want the data your retros depend on in one place? AgencyPro keeps each client's requests, messages and files together on every plan; time tracking is on Pro and profitability reports are on Plus, so post-mortems start from real numbers. Book a demo.