Most marketing reviews measure the campaigns. Almost nobody measures the operation itself — the stack it runs on, the data it depends on, the attribution it reports through, the governance that holds it together. So teams optimise output while the machine underneath quietly degrades: duplicate tools, rotting data, dashboards nobody can explain.
This free checklist is a quarterly marketing ops health check: 40 yes/no checks across four areas — stack, data, attribution and governance. It takes one focused half-day per quarter. You don't need consultants to run it (though we're happy to). You need 40 honest answers and one named owner per section.
Grab the downloadable version below, or work straight from the page. Either way: score it, date it, and compare it against last quarter. The trend matters more than the number.
Download: Free Quarterly Marketing Ops Health Review Checklist (PDF)
Get the one-page PDF version of all 40 checks — print it, run it with your team each quarter, and keep the scores side by side.
One email, no list spamming.
How to Use This Checklist
Each check is pass / fail / partial. Be strict — "partial" means someone knows about it but it isn't documented, and undocumented might as well be failed when the person leaves.
Three rules:
- One named owner per section. Stack, data, attribution, governance — four people, four names. Unowned sections score zero by default.
- Attach it to a rhythm. Run it inside your existing governance rhythms that keep systems aligned — quarterly, same meeting, same agenda slot. Standalone reviews get skipped; embedded ones don't.
- Any check failing two quarters in a row is an ownership problem, not a checklist problem. Escalate it — see the governance section.
If you want a deeper, more diagnostic pass over the automation platform itself, pair this with our marketing automation audit checklist — that one goes tool-deep where this one goes system-wide.
Section 1 — Stack (Checks 1–10)
The infrastructure layer: what you pay for, what's connected to what, and who can change it.
- Every tool in the marketing stack is inventoried — owner, monthly cost, renewal date, business purpose.
- No two tools do the same job without a documented reason (two form tools, two email tools, two "temporary" spreadsheets).
- Every integration is documented: what syncs, in which direction, which fields map where.
- Zombie tools identified — anything paid for but unused in 90+ days is cancelled or justified.
- Licence counts vs actual active users reviewed this quarter.
- API keys and service accounts inventoried, with owners and rotation dates.
- Workflow changes go through a change log — no silent edits to live automations.
- Every automation has a named owner and a rollback path (what to switch off if it misfires).
- Staging or sandbox exists for testing automation changes before they hit production.
- Total stack cost per marketing-sourced lead is known, tracked, and reviewed quarterly.
Section 2 — Data (Checks 11–20)
Agents, automation and attribution all amplify whatever the data contains — including its errors.
- A CRM field dictionary exists: every field has a definition, a purpose, and an owner.
- Critical fields are enforced at entry (dropdowns, validation) — not cleaned up later by hand.
- Duplicate rate is measured monthly and sits under an agreed threshold.
- Hard bounces, invalid records and unsubscribes are suppressed and synced across every connected tool.
- Lead source values are standardised — no free-text source field, no "Google / google / GOOGLE, organic?".
- Lifecycle stages are defined with explicit entry and exit criteria, and applied consistently.
- Data retention rules exist and are applied — stale records are archived per policy, not hoarded forever.
- No personal data lives in ungoverned spreadsheets, form exports or AI prompts outside approved systems.
- Every active contact's provenance and consent basis can be evidenced — not assumed.
- Someone owns data quality: a monthly report or dashboard exists, with a name attached.
Section 3 — Attribution (Checks 21–30)
If nobody can explain how a number is calculated, it isn't reporting — it's decoration.
- Tracking is implemented to current spec on every active channel (site changes break it silently).
- A UTM convention is documented and enforced — no improvised tags from whoever remembered.
- First-touch and last-touch definitions are agreed and have been stable for 12 months.
- CRM-to-platform lead matching rate is measured (target: above 90%) and known.
- The loop closes: revenue and stage data flow back from CRM into marketing reporting.
- The attribution model is documented — including a written list of its known blind spots.
- Baseline metrics (CPL, MQL→SQL rate, cycle length) are tracked quarter over quarter.
- No report is in active use whose calculation nobody in the room can explain.
- A dashboard of record is defined: one source of truth per metric, agreed by name.
- Every metric that moved more than 20% quarter-over-quarter has a written explanation.
Section 4 — Governance (Checks 31–40)
The section that decides whether the other three improve or decay.
- A RACI or ownership map exists for marketing ops — start with who owns what in your marketing stack if you don't have one.
- The quarterly ops review is on the calendar with a named chair — not "when things calm down".
- An access review was completed this quarter: no departed users, no orphan admin accounts, no shared logins.
- Documentation is current: runbooks exist for critical automations, and match reality.
- An incident process exists: who to call, what to switch off, who tells the client — before something breaks.
- Vendor agreements and DPAs reviewed in the last 12 months — including what your tools' AI features now do with your data.
- AI and agent usage in marketing follows the documented governance model — tools listed, purposes approved.
- One person owns the stack budget line and sees every renewal before it auto-charges.
- This checklist's score is tracked quarter over quarter — the trend is the KPI.
- Every review ends with at least one committed improvement: owner, deadline, next-review follow-up.
Scoring: What Your Number Means
- 34–40 — Healthy. The operation is governed. Keep the rhythm; protect it during busy quarters — that's when it dies.
- 25–33 — Degrading. Workable, but drifting. Pick the two lowest-scoring sections and fix ownership first, tools second.
- Below 25 — Structural. Don't start with the checklist; start with the operating model. Score yourself against the Connected System Maturity Model to see where the system actually stands, then rebuild from there.
A low score isn't a verdict — it's a baseline. The teams that improve fastest are the ones who score honestly once, then never let the same check fail twice.
The First Quarter: Starting from a Bad Score
- Week 1: Name the four section owners. Inventory the stack and the admin accounts (checks 1, 6, 33). This alone closes the most common holes.
- Weeks 2–3: Fix data entry points — field validation, source standardisation, suppression sync (checks 12, 14, 15).
- Week 4: Document the attribution model you actually have, blind spots included (checks 25, 26, 28). Then re-score.
One quarter of honest work typically moves a score from the low 20s into the 30s. The second quarter is easier. The habit is the hard part — which is why check 39 exists.
Websfarm runs these health reviews with B2B tech companies as part of marketing operations engagements — stack, data, attribution and the governance to keep all three honest. If you'd rather have the first one done with you, talk to us.
Download the free checklist (PDF)
All 40 checks on one page — score it, date it, and compare it against last quarter.
One email, no list spamming.