Nick QuirkThe first time a senior SEO strategist resigned on me, I lost four days to panic and about $18,000 in revenue over the following quarter. Not because she was irreplaceable — because everything she knew lived in her head, her inbox, and a Google Doc nobody else could find.
Agency employee turnover is not a crisis you solve. It is an operating condition you design around. Marketing agencies churn people at roughly 25-35% a year, and the roles that churn fastest — analysts, coordinators, junior AMs — are exactly the roles holding day-to-day client context. If your agency only survives when everyone stays, you do not have an agency. You have a group of freelancers sharing a logo.
Why agency turnover hurts more than turnover elsewhere
In most businesses, a departure costs you a skillset. In an agency, it costs you a relationship plus a skillset plus undocumented history.
Three compounding costs show up every time:
- Replacement cost. 6-10 weeks to hire, 8-12 weeks to ramp. On a $75k account manager, that is roughly $30-40k in recruiting, salary overlap, and lost billable capacity.
- Client risk. When a long-tenured AM leaves, the client's renewal conversation moves up. I have seen retainers that were six months from renewal get renegotiated within three weeks of a departure announcement.
- Team contagion. The remaining team absorbs the workload. If two people cover a departed colleague's five accounts for ten weeks, you have just created two more flight risks.
That third cost is the one agency owners underestimate. One resignation handled badly produces the next one.
The 72-hour handoff: what to do the week someone resigns
You usually get two weeks' notice. Most agencies spend the first week hoping the person changes their mind. Don't. Spend the first 72 hours extracting knowledge while the departing employee is still motivated to look good on the way out.
Day 1: freeze and inventory
Sit down with the person and list every client, project, and recurring commitment they own. Not from memory — from the actual project system, calendar, and inbox. For each item you need four things: current status, next deadline, who the client contact is, and what the client believes is happening right now.
That last one matters most. The gap between what is actually happening and what the client thinks is happening is where churn lives.
Day 2: record, don't write
Nobody writes good documentation in their notice period. Instead, run 30-minute screen-recorded walkthroughs per client: open the project, talk through the last six months, name the landmines. "The CMO hates being cc'd on anything technical." "Their dev team only ships on Thursdays." "We promised a content audit in Q1 and quietly dropped it."
Five clients, five recordings, two and a half hours. That is more transferable context than any handoff doc I have ever received.
Day 3: reassign with names attached
Every open item gets a new owner before the departing employee's last day — not "the team," a person. Unassigned work after a departure is how deliverables silently slip for three weeks until a client emails asking where their report is.
Telling clients before they figure it out
Tell them within 48 hours of the resignation being final. Not a mass email — a call or personal note from you or the new owner.
The structure that works: the person is leaving, here is who is taking over by name, here is what they have already reviewed about your account, here is what is unchanged in your next deliverable, and here is a 20-minute intro call this week.
Notice what is missing: apology and drama. You are not confessing a failure. You are announcing a planned continuity process. Agencies that treat departures as embarrassing teach clients to treat them as alarming.
If the relationship was informal — texts to personal phones, ad-hoc Slack requests — the transition is also the moment to reset expectations. This is a good time to reintroduce structured communication, which is much easier when you already have clear client boundaries in place rather than inventing them under pressure.
The real fix: make roles replaceable before anyone resigns
Handoffs are damage control. The structural work happens months earlier.
Move knowledge out of heads and into the workspace
Here is the test I use: can a competent new hire deliver a client's next monthly cycle using only what exists in the project system, with no more than one question to a teammate?
For most agencies the answer is no, because context lives in DMs, personal notes, and email threads. Fixing it means a client profile that actually gets maintained — goals, stakeholders, access credentials location, historical decisions, known sensitivities — sitting inside the project, not in a folder someone bookmarked in 2022.
A decision log is the highest-leverage version of this. Every time you make a call that a future teammate would question — "we deprioritized link building because their legal team blocks outreach" — log it with a date. Twelve entries per client per year is enough to make anyone dangerous within a week.
Templatize everything repeatable
Roughly 60-70% of agency work is procedurally identical across clients. If that work exists as a template rather than a habit, a departure becomes an inconvenience instead of an emergency.
Start with the workflows you run most: a reusable local SEO setup template, a standard technical audit sequence, a monthly reporting checklist, an onboarding phase plan. This is one of the few genuinely useful things a purpose-built agency PM tool gives you over a generic board — PeakKR's phase templates exist specifically so the tenth client setup is identical to the first without anyone remembering the steps.
The same logic applies to the coordination-heavy work. Technical SEO sprints with a documented fix-and-verify loop survive a departure. A technical program that lives in one strategist's mental model does not.
Build shallow redundancy, not full backups
You cannot afford two people per account. You can afford a secondary on every account who joins one client call a quarter and has read the project. That is maybe three hours per quarter per client. It converts a catastrophic departure into a survivable one.
Rule of thumb: no client should have exactly one person who has ever spoken to them. No workflow should have exactly one person who knows how to run it.
Use freelancers as shock absorbers — deliberately
A bench of 3-5 vetted freelancers you already pay small amounts to regularly is cheaper insurance than overhiring. The trick is that they need to be able to plug in without a two-week orientation, which again comes back to templates and documented process. If you have never formalized this, managing freelance writers inside your workspace is the right starting point, because content is usually the first thing to fall behind after a departure.
Reducing turnover in the first place
Agency people rarely leave over salary alone. In exit conversations I keep hearing three things: unpredictable workload, no visible growth path, and constant context-switching across too many accounts.
The workload one is measurable, and it is the one most agencies ignore until it is too late. If you track time, look at utilization by person over the last 90 days. Anyone consistently above 85% billable is a flight risk, full stop. Anyone running more than 6-7 active accounts is probably already updating their CV.
Two structural moves that actually help:
- Cap account counts per role. Write it down. "Senior AM: max 6 retainers." Then enforce it by hiring or declining work, not by hoping.
- Separate scope creep from capacity. Uncontrolled scope is a turnover driver disguised as a client-service problem. If a $4k/month retainer quietly consumes $7k of labor, the person absorbing that gap is the one who eventually quits.
Both require visibility into where hours actually go. Whatever system you use, it has to connect time to retainers and to people — if yours doesn't, the agency PM tool roundup is a reasonable place to compare options against how you actually operate.
What good looks like after 12 months
An agency that has genuinely solved for turnover shows a few specific signs: a resignation produces a scheduled set of tasks rather than a war room; clients hear about transitions from the agency first; new hires bill meaningful work in week three instead of week eight; and no owner can name a single person whose departure would threaten a top-five account.
You will still lose people. Good ones, at bad times, for reasons that have nothing to do with you. The goal is not zero turnover — it is turnover that costs you two weeks of friction instead of a quarter of revenue.
Turnover-proofing checklist
- Every client has a maintained profile in the project system: goals, stakeholders, sensitivities, access locations
- Every client has a decision log with dated entries for non-obvious calls
- Every client has a named secondary who has joined at least one call
