Create an Email Automation Health Check Before Journeys Quietly Break

An email automation can stay switched on while quietly doing the wrong thing. A form stops adding a tag. A delay changes from two days to two hours. A branch sends new customers down the prospect route. An unsubscribe reaches one list but not another. Nothing looks dramatic from the dashboard, yet real subscribers receive late, repeated or irrelevant messages.

An email automation health check is a short, repeatable review of whether a live journey still starts, decides, sends, stops and records outcomes as intended. It is not another campaign report and it is not a reason to rebuild every workflow. The aim is to find operational drift early, repair the same journey safely and leave evidence for the next review.

This guide gives a small business a platform-neutral way to monitor one live journey. It complements a pre-launch email QA checklist and an automation change log. It does not replace the rules of your email provider, data-protection obligations or qualified legal advice.

What an automation health check should prove

A useful check answers one question: can an eligible person still travel through the journey you approved, without receiving a message they should not receive? That requires more than looking at opens and clicks.

For each important automation, the review should prove six things:

1. the correct event can start the journey;
2. eligibility and exclusion rules use the right data;
3. delays and waiting periods behave as intended;
4. each message is sent once, to the right address, with working preferences and opt-out routes;
5. exit rules remove people when the journey is no longer relevant; and
6. the intended business outcome and audit record are created.

Those checks separate operational health from content performance. A low click rate may mean the offer or message needs work. Zero new entrants, a sudden pile-up at a delay or customers receiving prospect emails points to a broken route. Treating every weak metric as a copy problem wastes time and can hide the actual fault.

Start with a one-page journey baseline

You cannot detect drift if no one can say what “healthy” means. Before choosing alerts, create a compact baseline for the live version.

Journey name and owner:
Business purpose:
Approved live version and last change:

Entry event:
- exact event, source system and required fields

Eligibility:
- who may enter
- who must never enter
- re-entry rule

Path:
- branch conditions, delays and messages

Exit rules:
- purchase, reply, unsubscribe, status change or manual removal

Expected outcome:
- record, tag, task, sale or hand-off created

Health signals:
- starts, completions, stuck contacts, exits, delivery and complaints

Test contact:
- approved address, safe test data and cleanup steps

Review:
- weekly owner, monthly test date and escalation contact

Keep the baseline beside the workflow, not buried in a private notebook. Link to the current configuration and source records rather than copying personal data into the document. If the platform is being replaced, preserve this baseline in the email platform migration checklist so the new journey can be compared with the old one.

Check the six points where journeys drift

1. Entry

Confirm the trigger still fires from the real source. A healthy form submission count does not prove the automation received the event. Compare a small, recent sample across the form, ecommerce platform, CRM and email system using authorised records.

Look for missing fields, renamed events, disconnected integrations and unexpected duplicates. If the journey permits re-entry, verify the cooldown or repeat rule. A test contact that enters twice may expose a duplicate trigger before customers receive duplicate messages.

When this check exposes repeat-entry uncertainty, use the email automation re-entry test guide to compare a genuine new event with a duplicate trigger and inspect both entries and queued messages.

2. Eligibility and exclusions

Review the rules that decide who can enter. Check positive conditions and exclusions: active consent or another valid basis, customer status, product ownership, geography, age restrictions where relevant, staff or seed addresses, previous completion and suppression status.

Do not treat “not in the unsubscribe segment” as a complete permission check. An address may be suppressed because of an opt-out, complaint, hard bounce or internal safety rule. Keep those meanings separate using documented email suppression rules.

3. Delays and branches

Inspect contacts waiting at each delay and compare the oldest waiting time with the intended rule. A growing queue can indicate a changed time zone, impossible branch condition, missing field or paused step.

For every branch, sample at least one contact that took each active route. Confirm the source value at the decision time, not just its current value. A customer might have been a prospect when the branch ran and become a customer later; the audit needs that distinction.

4. Message delivery and links

Check sends, successful deliveries, bounces, complaints, unsubscribes and the destination of important links. Mailchimp’s official automation-flow reporting, for example, exposes delivery, bounce, abuse and unsubscribe data as well as message-level analysis. Brevo’s official automation statistics also separate starts, active contacts, finishes, removals and message delivery. The names vary by platform, but the operational questions are the same.

Open data can still help with trend context, but it should not be the only health signal. Privacy features, image blocking and automated security scanning can change opens or clicks without changing the journey. Prefer evidence from entry events, delivery, workflow logs, verified link destinations and downstream outcomes.

5. Exit and stop rules

Test the events that should stop the journey: purchase, booking, reply, unsubscribe, complaint, status change or manual removal. A journey that starts correctly but fails to stop can be more damaging than one that never begins.

Google’s current sender guidance requires a clear unsubscribe route for marketing mail and, for higher-volume senders, one-click unsubscribe for marketing and subscribed messages. Its subscription guidance says requests should be honoured within 48 hours. The UK ICO’s electronic-mail guidance also requires a simple opt-out in every marketing communication and makes the sender responsible for compliance. Check the rules that apply to your organisation and audience rather than assuming a platform setting covers every list.

6. Outcome and record

Follow the journey beyond the last email. Did it create the promised task, customer tag, sales record, hand-off or preference update? Did the record appear once, with the correct owner and date?

A completion count inside the automation may only mean the contact reached the final node. It does not prove the CRM task was created or the preference centre updated. The health check should connect the email route with the real business outcome.

Use a 20-minute weekly review

For a small portfolio of important automations, a brief weekly review is usually more useful than a large dashboard nobody owns. Adjust the rhythm for volume, risk and seasonality.

1. Confirm the live version. Match the workflow’s modified date and status with the change log.
2. Scan starts and completions. Compare the latest period with an appropriate recent baseline. Investigate zeros, sharp step changes and unexplained gaps.
3. Inspect queues. Look for contacts waiting longer than the approved delay or concentrated at one branch.
4. Review delivery signals. Check delivery failures, bounces, complaints, unsubscribes and provider warnings.
5. Sample exceptions. Read the logs for failed, removed or manually moved contacts and classify the cause.
6. Verify one outcome. Trace a completed contact into the CRM, shop, booking system or other destination.
7. Record the decision. Mark healthy, watch or investigate; name the owner and next action.

A percentage alone is not a universal alert. Five failed sends may be serious in a workflow that normally sends ten messages, and routine in a temporary test segment of thousands. Use your own baseline, minimum sample size and business impact. Google advises keeping the spam rate reported in Postmaster Tools below 0.3%, but that provider requirement is not a general purpose threshold for every other journey metric.

Run a monthly synthetic journey test

Metrics tell you what the platform recorded. A controlled test tells you what a subscriber experiences. Use a dedicated test contact and safe fictional data; do not reuse a real customer record.

1. capture the live workflow version and take a configuration snapshot;
2. place the test contact in the required starting state;
3. trigger the journey through the real authorised entry route;
4. confirm entry time, eligibility data and selected branch;
5. check each message on desktop and a narrow mobile view;
6. test important links, reply handling, preferences and unsubscribe;
7. confirm the exit event stops every remaining promotional step;
8. verify the downstream task or record; and
9. remove test artefacts without deleting the audit evidence.

Do not shorten live delays or edit branch rules just to make the test faster. Use a provider’s test mode, clone or safe staging route where available. If none exists, test the smallest possible live path and document the limitation.

Worked example: a three-message welcome journey

Imagine a fictional training business with a welcome automation. A confirmed subscriber receives a guide, waits two days, receives a lesson, then receives an invitation unless they have already bought the course. A purchase should remove them immediately and start the customer onboarding route.

The weekly dashboard shows normal opens, but starts have fallen. A health check compares the website form with automation entries and finds that recent submissions contain a renamed consent field. The form saves the contact, but the eligibility rule still looks for the old field. Copy changes would not solve the problem.

After the field mapping is repaired, a synthetic test finds a second fault: the purchase exit works for card payments but not for invoices marked paid in the accounting system. The team pauses the invitation step for affected contacts, fixes the event mapping, re-tests both payment routes and records the change.

This example is illustrative. It does not claim measured results from a real business. Its value is the sequence: compare systems, isolate the failed checkpoint, protect subscribers, repair one route and verify the outcome.

Use three response levels

Healthy: the journey matches its baseline, the test path completes and no unexplained exception needs action. Record the review date.

Watch: a signal has moved, but the evidence does not yet show a broken route or subscriber harm. Name the metric, comparison period, owner and next review. Do not keep a journey in “watch” indefinitely.

Investigate: entry, eligibility, sending, stopping or the downstream outcome is wrong or cannot be proved. Pause the smallest unsafe step, preserve logs, identify affected contacts and decide whether a correction or notification is required. Do not restart until the failed checkpoint passes a controlled test.

If the issue involves consent, privacy, security or a widespread sending error, use the appropriate incident and legal process. An operational checklist helps locate the fault; it does not determine regulatory obligations.

A compact health-check log

Journey:
Live version:
Review period:
Reviewer:

Entry: PASS / WATCH / FAIL
Evidence:

Eligibility and suppression: PASS / WATCH / FAIL
Evidence:

Delays and branches: PASS / WATCH / FAIL
Evidence:

Messages and links: PASS / WATCH / FAIL
Evidence:

Exit rules: PASS / WATCH / FAIL
Evidence:

Outcome and records: PASS / WATCH / FAIL
Evidence:

Synthetic test:
- test contact
- path taken
- result

Decision:
- healthy / watch / investigate
- owner
- next action and deadline
- change-log or incident reference

A bounded prompt for reviewing the evidence

An AI assistant can organise observations, but it should not decide whether consent is valid, invent a root cause or change the live journey. Remove personal data and confidential configuration before using an external tool.

You are helping organise an email automation health-check record.
Use only the supplied baseline and observations.
Do not infer consent, legal compliance, root cause or missing data.

Return:
1. confirmed facts;
2. differences from the baseline;
3. evidence gaps;
4. checkpoints needing human investigation;
5. the smallest safe test to discuss with the workflow owner;
6. questions for the email platform or integration owner.

Keep personal data out of the response.
Do not recommend sending, deleting or changing subscriber records.

Baseline:
[Insert approved, minimised baseline]

Observations:
[Insert non-personal evidence]

Check every output against the platform logs and source systems. A tidy explanation is not proof of the cause.

Keep monitoring connected to the wider email system

An automation health check works best as one part of a connected operating routine. Use the deliverability diagnosis guide when provider or inbox signals deteriorate. Use the preference-centre plan when subscribers need better choices, and keep evidence of permission with the consent record audit.

For the broader strategy—audience, value, segmentation, campaigns and measurement—return to Email Marketing for Beginners. Monitoring should protect that strategy, not become reporting for its own sake.

Reader Q&A

What is an email automation health check?

It is a repeatable review of whether a live email journey still starts, applies eligibility rules, follows the intended branches, sends correctly, stops when required and creates its expected business outcome.

How often should I check email automations?

Review important automations on a rhythm that matches their volume and risk. A small business can begin with a brief weekly signal review and a monthly controlled journey test, then adjust using evidence.

Which metrics matter most?

Start with entries, completions, contacts stuck at steps, removals, delivery failures, bounces, complaints, unsubscribes and the final business outcome. Use opens and clicks as supporting context rather than proof that the route works.

Should I test a live automation with a real customer?

No. Use a dedicated test contact and safe fictional data. Prefer a platform test mode or staging copy, and never change live timing or rules merely to make a test finish faster.

What should I do if an unsubscribe does not stop the journey?

Pause the smallest unsafe sending step, preserve the evidence, identify the affected routes and escalate immediately to the workflow owner and the appropriate compliance process. Repair and re-test the same journey before restarting it.

Is an automation report the same as a health check?

No. A report summarises activity and performance. A health check compares the live journey with its approved baseline, tests critical controls and verifies the subscriber experience and downstream result.

Garry Knight

I'm Garry Knight, the person behind Prodify Digital. I write about email list building, email marketing, SEO, AI search and the tools that connect them. My aim is to make online marketing easier to understand, so creators and small business owners can make informed decisions about building an audience and keeping people engaged. Here you'll find straightforward guides and product reviews that explain what something does, where it fits and which limitations matter. The focus is on clear explanations and useful next steps—not hype, shortcuts or promises of easy earnings.

See Full Bio