Careers
Part of What matters most in email automation: five priorities
Email automation questions: what people ask and what is true
Common email automation questions cover triggers, eligibility, journey length, collisions, message classification, testing, measurement, ownership, and safe exits.
What to take away
- A trigger must represent a defined business event, not a convenient proxy whose meaning changes across systems.
- Eligibility is a send-time decision because permission, preference, status, priority, and exclusions can change after entry.
- Questions about classification, collision, failure, measurement, and stop authority should be resolved before launch.
Common email automation questions usually concern when a journey should start, how often it should send, and what success means. The answers depend on customer context, message classification, market, data reliability, and operational risk. Verify current requirements and obtain qualified review before launch.
What should trigger an automation?
The ICO's electronic-mail key concepts define direct marketing broadly, explain what counts as electronic mail, and distinguish solicited from unsolicited marketing. Use the current U.K. guidance only for recipients and activity within its scope, then document the actual message purpose and applicable market rule.
Use a defined event or state that reliably represents the intended context. Document producer, timestamp, identity, required fields, deduplication, ordering, latency, version, and owner. Do not confuse a proxy with the completed customer outcome.
Worked example: an order-shipped trigger carries order_id, customer_id, shipped_at in UTC, carrier, tracking_number, locale, and event_version. The order service produces the event and the fulfillment data owner owns it. Deduplicate on order_id plus event_version.
When should eligibility be checked?
At entry and immediately before every send. Permission, market, topic, relationship, preference, frequency, suppression, and service state can change during a delay. Use a final gate and cancel queued messages when eligibility fails.
How many emails belong in a journey?
Use the minimum sequence needed to deliver the promised value or decision. A typical behavior-triggered journey runs three to five messages over seven to fourteen days; onboarding runs five to seven over two to three weeks. Set purpose, delay, exit, and guardrails for each message. Stop when the customer completes the task, declines, expires, or enters a higher-priority path.
How should automations avoid collisions?
Create a global hierarchy, quiet periods, maximum frequency, mutually exclusive journeys, suppressions, and recovery across marketing, sales, product, and service platforms. A per-journey cap does not protect the total customer experience.
A sample hierarchy, highest priority first: legal or regulatory notice, security or account alert, service or transactional message, requested customer journey, sales outreach, promotional broadcast. A typical cap is four marketing messages per recipient per week across all systems.
Can marketing appear in transactional email?
The current U.S. primary-purpose rule in 16 CFR 316.3 addresses mixed commercial and transactional or relationship content through the subject line, ordering, proportion, placement, and visual emphasis. It also lists defined transactional or relationship categories. Apply the rule to the complete message and current facts; this article is not legal advice.
Classification depends on applicable rules and overall or primary purpose, not the automation label. The FTC describes U.S. categories and mixed-content analysis; other markets differ. Protect essential messages and obtain qualified advice.
How should automation be tested?
Test state transitions for eligible, excluded, suppressed, duplicate, late, reversed, missing, stale, shared, wrong-market, completed, canceled, expired, migrated, failed, and recovered records. Verify events, waits, branches, exits, queues, logs, and rollback.
Which metrics matter?
Track events, entries, exclusions, sends, branch distribution, exits, duplicates, failures, complaints, withdrawal, support, verified actions, retained value, revenue quality, and contribution. Use holdouts where appropriate and treat opens as limited diagnostics.
Track complaint rate against a typical ceiling of 0.1 percent of delivered mail, adjusted to your own data. Watch exit rate at the first message: if most entries leave there, the trigger is too broad.
Preserve versions and results so changes remain auditable.
Automation question owners
| Question | Evidence | Owner |
|---|---|---|
| Why did the person enter? | Event and identity record | Data owner |
| May this message send? | Purpose, market, eligibility | Requirement owner |
| What can stop the path? | Exit, suppression, incident test | Journey owner |
| Did it create value? | Mature outcome and guardrails | Decision owner |
Resolve the question with a decision record
The GAO evaluation design guide connects evaluation questions with evidence needs and design choices. Write the evaluation question first, then name the evidence that can answer it; federal evaluation guidance does not make a local marketing result causal or transferable.
The W3C Privacy Principles statement gives web-system designers shared privacy concepts and warns against shifting privacy work to individuals. Map each data flow to a stated purpose and a retention limit, then review the governing law and configuration.
Common questions
How many emails should an automation contain?
Only the messages required by the customer need and evidence. Define the exit and expiration first, then justify each message's distinct job.
Can an automation reuse a broadcast list?
Only after reproducing eligibility for that journey, message type, market, purpose, topic, timing, preference, and current exclusions.
When should a question block launch?
Block launch when the answer could change who receives the message, what it claims, whether it is safe, or whether the team can stop and correct harm.
