Agency Pre-Mortem For New Client Onboarding
A pre-mortem workshop for the first 90 days of a new DTC retainer — surfacing GA4 access gaps, CRO velocity mismatch, unclear KPI ownership, and the scope-creep patterns that quietly kill agency margin.
Quick answer
Before the kickoff call, gather your pod for a 60-minute pre-mortem: imagine it's day 90 and the retainer is being cancelled, then list every reason why. The four recurring killers on DTC accounts are missing historical GA4 access, a client who thinks CRO ships weekly, no named KPI owner on the client side, and scope creep from paid-media requests. Fix each one in the SOW before you countersign.
Agency Pre-Mortem for New Client Onboarding
A structured risk workshop run before a new DTC retainer starts, mapping the most likely reasons the engagement will fail by month four.
An agency pre-mortem for new client onboarding is a 60-90 minute workshop the pod runs the week before kickoff. The team assumes the engagement has already failed at day 90 and works backwards to name every plausible cause — data access, expectation mismatch, ownership gaps, scope drift. Each risk gets an owner, a mitigation, and an SOW clause.
On DTC retainers the failure modes cluster tightly. Historical GA4 access lands late, the client expected two tests a week, the marketing manager who signed the deal isn't the person approving experiments, and paid-media firefighting eats the CRO hours you priced. Naming these before month one is cheaper than renegotiating in month four.
The pattern the pre-mortem is designed to break is predictable. New retainer signs in month zero. Months one and two look busy — audits, tag reviews, hypothesis backlogs. Month three the client asks why revenue hasn't moved. Month four the notice email arrives.
The technique itself is borrowed from Gary Klein's pre-mortem analysis, adapted for agency retainers. Instead of forecasting success, you assume failure and reverse-engineer the causes. On a DTC pod, that surfaces onboarding risks that a standard kickoff deck consistently misses.
Risk 1: Historical GA4 access lands too late
The single most common day-90 post-mortem finding: the pod couldn't run a proper audit in month one because GA4 property access sat with an ex-employee, an outsourced dev shop, or a Google Workspace admin who was on holiday. Two weeks disappear before the first hypothesis even lands.
Mitigation in the SOW: written confirmation of GA4 admin access, Shopify collaborator access, and CMS access before the countersignature. If the client can't produce it in five business days, the start date shifts — you don't burn month one on procurement chasing.
The historical-data trap
A retainer that starts without 12+ months of GA4 history means you're building hypotheses on vibes for the first quarter. Either import the history on day one (Metricuno's GA4 backfill is one option — Fivetran or a manual BigQuery pull are others) or price a discovery sprint before the CRO work begins. Do not price a full retainer against zero baseline.
Risk 2: The client priced weekly tests; you priced monthly
CRO velocity mismatch is the second most common cause of month-four churn. The client saw a case study promising "one test per week" and assumed that's the baseline. Your pod, working on a Shopify store doing 180k monthly sessions, needs three-week cycles to reach significance on anything below hero-banner traffic.
Fix it in the kickoff, not the QBR. Walk the client through their own traffic and conversion numbers, compute the minimum detectable effect at their sample size, and put realistic cycle time in the SOW. "6-8 tests per quarter, 2-3 concurrent" reads more honestly than a weekly cadence you'll miss.
For low-traffic stores under 60k monthly sessions, the honest answer is often "we run fewer, bigger tests, and lean on UX audits and heuristic wins." Say it out loud in week one — clients handle the truth in kickoff; they don't handle it in month three.
Risk 3: No named KPI owner on the client side
Typical onboarding risks on DTC retainers, ranked by frequency in month-4 churn interviews
| Risk | Frequency | Avg weeks lost | SOW mitigation |
|---|---|---|---|
| Delayed GA4 / analytics access | ~55% of retainers | 2-3 weeks | Access confirmed before countersignature |
| CRO velocity expectation mismatch | ~45% | 0 (but drives churn) | Test cadence tied to actual traffic in SOW |
| No single KPI owner on client side | ~40% | 1-2 weeks per test | Named approver + backup in kickoff doc |
| Paid-media firefighting scope creep | ~35% | 4-6 hours / week | Explicit scope boundary + change-order clause |
| Checkout / dev freeze mid-quarter | ~25% | 3-4 weeks | Dev-freeze calendar shared in month one |
| Founder overrides marketing lead | ~20% | varies | Founder attends monthly review, not weekly |
The KPI-owner problem sounds procedural but it's the one that quietly kills retainers. The marketing manager who signed sits in the middle: the founder wants revenue, the performance manager wants ROAS protected, the CX lead wants returns down. Without one named approver, every test gets committee-reviewed and cycle time doubles.
Risk 4: Scope creep from paid-media firefighting
On DTC accounts, paid media and CRO share a border and the border leaks. Meta CPMs spike, ROAS drops, the client Slacks the CRO pod asking for a "quick landing page test." Four hours a week gets absorbed into work you didn't price. The retainer margin erodes without any single conversation triggering a scope review.
The pre-mortem outcome here is a written scope boundary and a change-order threshold — anything over two hours of unplanned work generates a change order, not a Slack yes. It feels stiff in month one. It saves the retainer in month five.
Running the workshop: 60 minutes, five outputs
Get the pod (strategist, analyst, designer, PM) in one room. Ten minutes of silent writing: "it's day 90, the retainer is being cancelled — why?" Twenty minutes clustering the answers on a board. Twenty minutes assigning owners and mitigations. Ten minutes translating the top five risks into SOW clauses or kickoff-doc red flags.
The output isn't a document — it's five things: a revised SOW, a kickoff agenda with the named KPI owner confirmed, a data-access checklist with dates, a test-cadence commitment that matches the client's traffic, and a scope-creep escalation path. Do this once per new client and month-four churn drops noticeably.
Frequently asked questions
A risk register lists risks abstractly and rates them. A pre-mortem forces the team to assume the failure has already happened, which surfaces specific, concrete causes the risk-register format tends to smooth over. It's a technique from Gary Klein's pre-mortem analysis work, adapted for the agency onboarding context.
Before countersignature, ideally between verbal agreement and signed SOW. That's the window where you can still change scope, cadence commitments, and access requirements without renegotiating. Running it after signing turns findings into awkward change orders.
No. This is an internal pod exercise. Clients will filter what they say in the room, and the point is to name uncomfortable risks like "the founder will override the marketing lead by week six." Share the mitigations with the client, not the raw output.
That's your answer — the retainer will struggle. Push back in kickoff: one named approver, one named backup, everyone else consulted. If the client insists on committee approval for every test, price cycle time accordingly or decline the account.
Compute minimum detectable effect at their current traffic and conversion rate. A store with 40k monthly sessions and a 2% baseline realistically ships 3-4 tests per quarter, not per month. Show the client the math in kickoff and commit to a cadence you can actually hold.
No — it feeds into it. The pre-mortem produces the agenda, red flags, and clarifying questions for the kickoff call. Think of it as the pod's prep session for a much sharper kickoff conversation.
Either import the client's historical GA4 data on day one so you can audit against a real baseline (Metricuno, Fivetran, or a manual BigQuery export all work) or price a paid discovery sprint before the CRO retainer starts. Do not price full CRO work against zero baseline data.
Two hours of unplanned work per week is a common line. Below that, absorb it as goodwill. Above it, generate a written change order — not to bill hard, but to make scope visible before the retainer margin quietly disappears.
Once at onboarding, then again at any structural change: new stakeholder, replatforming, entering new markets, major seasonal push. A mid-retainer pre-mortem before a Q4 sprint tends to catch dev-freeze conflicts before they cost you two weeks.
The SOW clause on data access with a fixed deadline. That one clause alone — GA4, Shopify, CMS access confirmed within five business days or the start date shifts — prevents the most common cause of month-one waste on DTC retainers.
See Metricuno on your data
Bring your stack — Google Analytics, Stripe, a CRM, anything — and we'll walk through the metric tree that turns your funnel into one number.