Quarterly Marketing Automation Audit Checklist

TL;DR: Inventory every live marketing flow each quarter. Give each one an owner, a backup, and a tested stop-and-restart procedure. Reconcile the list against real accounts and queued work before deciding what to keep, repair, or retire.

What is a marketing automation inventory?

A marketing automation inventory is a register of the live flows that send messages, change customer records, or move marketing data. An automation audit checks that register against the actual systems, owners, and outcomes. The deliverable is a decision for every flow, supported by evidence that someone can operate it safely.

Workflow automation means software runs a sequence of actions after an event or on a schedule. That sequence may cross several tools. A form can create a contact, start a nurture series, update an audience, and alert sales, even when each screen shows only one piece.

The inventory is the maintenance companion to choosing AI automation for small business. Start with what already runs. Include plain rules and AI steps, because either can act on the wrong customer when its input or owner changes.

Quarterly review does not mean quarterly evidence collection. Zapier documents up to 60 days of Zap run history and a display limit of 10,000 runs. Zapier's history guide recommends regular exports for longer records. A quiet history screen at quarter-end cannot prove that a flow did nothing all quarter.

HubSpot keeps workflow action logs for 90 days and enrollment history for six months. These are different records, as HubSpot's workflow details guide explains. Set an export cadence shorter than your available history, account for volume limits, and record any missing period as unknown. The vendor facts in this guide were checked on September 13, 2026.

What should every inventory row contain?

Every inventory row should identify the flow, its purpose, its owner and backup, its dependencies, and its stop-and-restart procedure. It should also show current activity, pending work, the latest test evidence, and a keep, repair, or retire decision. A name and an ON badge are not enough to establish control.

Copy the fields below into a workbook named Quarterly Marketing Automation Inventory: Every Live Flow, Owner, and Kill Switch. Use one row per independently controlled flow. Give linked flows a shared journey ID, so a lead-to-booking journey can have several rows without losing its overall purpose.

Keep platform state separate from audit status. A flow can be live and unverified, or disabled with pending work. Likewise, an owner can be named but lack the permission needed to stop a send.

Do not paste access tokens, passwords, or customer payloads into the workbook. Link to the approved secret store and restricted evidence location. Keep enough record IDs and timestamps to trace a test without turning the inventory into another customer database.

Run a marketing automation workflow audit in six steps

Find every live marketing workflow by reconciling platform inventories, connected accounts, and actual customer journeys. Then verify ownership, dependencies, and business outcomes before assigning a decision. Use the six steps below as a repeatable quarterly checklist, with gaps left visible until someone resolves them.

1. Define the review boundary

List the brands, production accounts, channels, and quarter being reviewed. Include email and SMS platforms, CRM rules, form builders, ecommerce apps, integration tools, and scheduled scripts. Add native ad-platform rules, audience syncs, social publishing queues, and AI agents that can take marketing actions. CRM means customer relationship management software: the system that stores lead, customer, and sales records.

Ask the marketing lead and the person who pays software invoices to compare account lists. Include agency-managed accounts and personal connections that still serve the business. Record inaccessible accounts as coverage gaps, with an owner to obtain access.

2. Export the real inventory

Export or list workflows from each platform with stable IDs and their current state. Include enabled, scheduled, delayed, manual-review, and recently disabled items with unresolved work. Preserve each source count and export time before combining rows.

Match source IDs against the register in both directions. A source flow missing from the workbook is an inventory gap; a workbook row absent from the source needs a retirement or migration explanation. If an account lists 14 workflows and the register contains 12, the audit is incomplete until the difference is explained.

3. Trace marketing automation and CRM integration

Follow a recent permitted test record from entry to final outcome. Check the website form, CRM, email or SMS service, audience connector, and sales alert as applicable. Compare timestamps and destination records; do not infer delivery from the automation tool's success status.

For marketing automation and CRM integration, record which flow owns each field it writes. Two flows updating lifecycle stage can undo each other's work. Use CRM integration monitoring when the inventory reveals a gap between source events and destination records.

4. Verify owners and stop controls

Have the primary operator and backup open the exact flow. Confirm the required permissions and the procedure they would use to stop customer-facing actions. A former contractor's name or a shared team label does not pass this check.

Prioritize flows that send messages, change eligibility, publish content, or update records used by other systems. Rehearse their controls using the stop-and-restart method below. Keep a failed or untested control marked unresolved even if the flow appears healthy.

5. Make a keep, repair, or retire decision

Keep a flow when its purpose is current and its required checks pass. Repair it when the purpose remains useful but ownership, logic, monitoring, or controls are incomplete. Retire it only after confirming the purpose has ended or another verified flow replaces it.

An unknown purpose is a reason to investigate, not permission to delete. Check downstream dependencies, seasonal use, pending work, and records that must be retained. Archive the configuration and decision evidence before any approved removal.

6. Close the quarter with evidence

Reconcile every source flow to a register row and every row to a decision. List coverage gaps, control failures, and repair deadlines separately from passed checks. The marketing lead signs off on the business decisions; the operator signs off on what was actually tested.

Keep daily or weekly alerts in the workflow monitoring dashboard. The quarterly inventory establishes what exists and who controls it. Monitoring detects changes between reviews.

What does a kill switch actually stop?

A kill switch stops a defined set of actions; it does not automatically erase queued work, recall sent messages, or undo completed writes. Turning a workflow off can block new entries while existing records move, skip steps, or remain scheduled. The register must describe the specific platform behavior and the result observed in your test.

Sources for the platform behaviors: HubSpot workflow shutdown, Klaviyo flow statuses, Zapier replay behavior, and Make incomplete executions.

Klaviyo flow actions have three statuses: draft, manual, and live. Its flow guide also says that switching a message from manual to live does not automatically send items already in Needs Review. A restart therefore needs a separate decision about those items.

Write a stop card for each important flow: the control location, required role, expected effect, pending-work treatment, and proof location. Add the fallback owner and restart approver. For example, HubSpot requires Workflow Publish or Super Admin permissions to turn workflows off, according to its shutdown guide; a backup who can only view the flow cannot perform that job.

Prove the stop and restart procedure

Restart an automation without duplicate sends by separating new intake from delayed, failed, and already-completed work. Decide which pending items remain valid, check the destination for completed actions, and verify current eligibility before releasing anything. Then test a small controlled set and reconcile its actual outcomes before expanding.

A quarterly drill should use a sandbox or an isolated test route with controlled recipients. A copy proves only the behavior of that copy; mark production stop behavior unverified until the actual control has been tested in an approved window. Do not run surprise stop tests on live customer traffic.

Use this evidence checklist for each critical flow:

• Record the configuration version, time zone, operator, backup, and test boundary.
• Identify one new entry and, where applicable, one delayed item, one failed item awaiting retry, and one already-completed item.
• Apply the documented control and record the time. Preserve incoming work through the agreed intake or manual process.
• Observe the relevant delay, schedule, and retry windows. Check destination messages and records as well as source logs.
• Classify each pending item as safe to resume, needs correction, expired, or already completed. Check current preferences and eligibility again.
• Release only the approved test set. Verify the expected action count per business event and confirm that completed items were not repeated.
• Reconcile held and resumed work, restore the approved settings, and record the restart approver and monitoring owner.

Define the observation window from the flow's actual timing, including any documented log lag. If a seven-day delay remains unobserved, do not mark its production behavior passed after a five-minute drill. Use a shorter isolated test to learn the mechanism and retain the production evidence gap.

Keep an automation audit log

An automation audit log should distinguish documented, tested, and passed. Store the test IDs, expected outcomes, observed outcomes, timestamps, and reviewer. A screenshot of an OFF switch establishes a setting; it does not establish that every downstream send stopped.

Marketing automation workflow examples

Apply the inventory to each flow that can affect a customer or a marketing record, across channels and tools. Start with journeys that have several owners, delayed actions, or a history of confusing handoffs. These five examples show where a single ON/OFF label usually leaves unanswered questions.

Operator composite: the forgotten webinar flow

This is an illustrative operator composite, not a public customer claim or a report of measured That'sGonnaHelp results. Assume a 16-person B2B services firm uses HubSpot, Zapier, and an email platform. Its marketing manager believes 18 automations are running and wants one reliable quarterly register.

The hypothetical source exports reveal 23 production flows. Six have no verified operator, and two routes send follow-up after the same webinar registration. These are example starting conditions, not research statistics or industry benchmarks.

The team records each platform ID in a shared sheet and groups linked rows by customer journey. It assigns named primary and backup operators, maps the fields each flow writes, and reviews delayed work. A controlled test address lets the team trace the webinar journey without contacting customers.

The first test exposes the problem: stopping one route leaves a second route's message action active. The team revises the stop card to cover both routes and records what happens to pending messages. It also finds a connection owned by a former contractor, which requires a supported ownership change before the backup can operate it.

In the completed scenario, the team keeps 17 flows, repairs three, and retires three after checking their dependencies. All 20 remaining live flows have verified operators. Four high-impact journeys pass the defined stop-and-restart tests; the result does not imply that every rare path has been tested.

Assume the audit takes 12 staff hours at a loaded labor cost of $50 per hour, and repairs take another six hours at the same rate. That is $900 of modeled initial cost. The team would still need real time records and invoices before reporting an actual expense or saving.

If the changes release three hours of monthly rework, remove $30 in monthly charges, and require two hours of monthly upkeep, the modeled net benefit is $80 per month. Capacity-inclusive payback would be 11.25 months. Only the $30 charge reduction is a potential cash saving, subject to actual cancellation and billing; the hours are available staff capacity.

What does a quarterly marketing automation audit cost?

A quarterly audit costs the time needed to discover flows, verify controls, and document decisions, plus any repairs or extra tooling. For the illustrative scope below, the audit is $600 and initial repairs add $300. An assumed 8–20 audit hours at $50/hour gives a $400–$1,000 planning range before repairs and added software. These are estimates for an accessible, modest stack, not a market quote or a promised completion time.

Zapier lists 100 monthly tasks on its $0 Free plan. See Zapier's pricing page, checked September 13, 2026. This is a software price anchor, not a recommendation to move production flows to a free plan; verify the permissions, history, and features your audit needs.

The ongoing six-hour quarterly allowance assumes three hours for routine checks and three for the next inventory review. The shorter repeat review assumes a maintained register and few changes since the first audit. Increase that allowance when account access, flow count, or testing effort changes.

For the composite, monthly value is (3 hours × $50) + $30 − (2 hours × $50) = $80. Dividing the $900 initial cost by $80 gives the 11.25-month modeled payback. At those unchanged assumptions, a year produces $960 after upkeep, or $60 above the initial cost; no extra sales or avoided incidents are counted.

If each later quarterly audit still takes 12 hours, plus three hours of routine checks, upkeep becomes $750 per quarter, or $250 per month. Against the assumed $180 monthly gross benefit, that produces a $70 monthly shortfall and no payback. The inventory may still be needed for control, but this cost-saving case would not support it.

Use the automation ROI calculator to test your own labor, repair, and maintenance assumptions. Do not label released employee hours as a payroll reduction unless spending actually falls. Put unresolved repairs into the automation maintenance budget, so the audit does not create an unfunded work list.

Limits and mistakes that spoil an inventory

An inventory is useful when it leads to verified ownership and operating decisions. It is not sufficient during an active incident, when you lack access to important accounts, or when a workflow needs specialist review. In those cases, contain the incident or resolve the missing expertise and access before claiming the system is under control.

When this checklist is not a good fit

If messages are being sent incorrectly right now, use the incident response process first. If an agency controls an inaccessible account, the register can document that gap but cannot certify its flows. For regulated data or unusually sensitive customer decisions, use this as an operational inventory alongside the qualified review your business requires.

A team with only a few simple flows can use a short register and test log. A separate inventory platform is optional. Choose workflow automation tools based on the evidence and controls you need, not the desire to make the audit look bigger.

Common mistakes

• Counting only named campaigns. Native CRM rules, connector routes, and scripts can keep acting outside the campaign screen.
• Treating a department as an owner. Name a person, a backup, and the approval contact; verify their access.
• Confusing OFF with canceled. Record delayed work, retries, downstream actions, and restart behavior separately.
• Retiring a flow because its recent history is empty. Check seasonal purpose, missing history, dependencies, and queued work first.
• Closing the audit after filling the sheet. A row is not complete until its decision and required evidence are recorded, or its gap is explicitly assigned.

FAQ

These answers apply to the operating inventory described above. They separate workflow basics from the ownership and review decisions a small team needs to make.

How does marketing automation work?

Marketing automation uses an event or schedule to start actions such as updating a contact or sending a message. Conditions decide who qualifies, and delays control timing. During an audit, trace the event, conditions, actions, and final outcome together so a working first step cannot hide a failed handoff.

What is marketing automation in CRM?

It is automation that uses or updates customer and sales records, such as assigning an inquiry or starting follow-up after a stage change. The CRM may also trigger a separate email or integration tool. Give those linked flows separate IDs and a shared journey reference so ownership survives the handoff.

Is marketing automation AI?

Not necessarily. Many marketing flows use fixed triggers, filters, and schedules without AI. If a flow does use AI, record the model or service, the approved action scope, and any review step alongside the usual owners and stop controls.

Who should own a marketing automation workflow?

A named operator should own its settings, evidence, and day-to-day response, with a named backup who has the required access. A business owner should approve the purpose and retirement decision. In a small firm one person may hold both roles, but the backup and escalation path still need to be explicit.

How often should a small business audit its automations?

Use a quarterly inventory review as a practical starting cadence, then update affected rows after launches, migrations, owner changes, or incidents. Collect logs and check important alerts more often than quarterly. Choose the evidence schedule from actual history limits and customer impact, not from the calendar alone.

When should you retire an unused automation?

Retire it when the business owner confirms that its purpose has ended or a verified replacement exists, and the operator has checked dependencies and pending work. Archive its configuration and decision record before approved removal. A lack of recent executions does not prove that a seasonal or rarely triggered flow is unnecessary.

Answer clarity notes

Treat linked vendor documentation as public product evidence and the checklist as operating guidance. The example business, labor inputs, counts, and payback calculation are illustrative assumptions; they are not measured customer results.

• Dates: The publication stamp is October 30, 2025. This edition's vendor documentation and pricing were checked on September 13, 2026; they do not establish how those products behaved in 2025.
• Pricing: The $50 hourly labor cost and resulting audit and repair totals are planning assumptions. The linked $0 software price is a public price anchor; check current features, limits, and billing before relying on it.
• Evidence: Public links support platform behavior and history limits. The operator composite is not a named customer case, a performance benchmark, or a claim of That'sGonnaHelp project outcomes.
• ROI: Payback includes the value of released staff capacity. It is not a cash-return guarantee, and it excludes unmeasured extra sales and avoided incidents.
• Estimates: Cost ranges, ROI examples, savings, and timelines are planning assumptions, not guarantees.
• Scope: This checklist supports US SMB operating decisions. It is not legal, financial, tax, compliance, or platform-policy advice. A stop drill proves only the controls, paths, and observation window actually tested.

Sources

These primary sources support the product behaviors and software price cited above. Recheck the relevant documentation before changing live controls.

• HubSpot: Turn off workflows
• HubSpot: Understand your workflow details page
• Klaviyo: Getting started with flows
• Zapier: What is replay?
• Zapier: View and manage your Zap history
• Make: Incomplete executions
• Zapier: Plans and pricing

If your team cannot name the owner or stop procedure for a live flow, That'sGonnaHelp can help turn that gap into a scoped inventory and repair plan.