Email Automation Strategy: Workflows That Run Themselves
Which flows to build first based on what they return, and the review schedule that stops them drifting.
On this page
The short version
- Automated flows are the highest-return work in most programmes and the least revisited. A welcome sequence written two years ago is still shaping first impressions.
- Build in order of return per hour: welcome, then whatever your commercial trigger is, then re-engagement. Everything else is optional.
- The review schedule is the part that gets skipped and the part that decides whether flows stay true. Set a date before you build anything.
Automation has an unusual economics. A campaign costs its production every time it runs; a flow is written once and sends indefinitely. Over two years a welcome sequence may deliver more messages than every campaign combined, at a fraction of the effort.
That is also its risk. A flow keeps sending long after the product, the pricing and the people described in it have changed, and nothing about it will announce that it has gone out of date.
Build order, by return per hour
Welcome first, always. It reaches people at the only moment they have actively chosen to hear from you, it is read at a rate nothing else matches, and it runs for years.
Second is whatever your commercial trigger is — abandoned cart for a shop, trial onboarding for software, enquiry follow-up for a service. This is the flow with a revenue number attached, and it is the one that justifies the rest.
Third is re-engagement, because it protects deliverability for everything else and it is short.
After those three the returns fall sharply. Browse abandonment, anniversaries, behavioural triggers and nurture sequences are all worth having eventually, and none of them is worth building before the first three are working.
Flows by return and effort
| Flow | Effort | Return | Build |
|---|---|---|---|
| Welcome sequence | A day | Highest | First |
| Commercial trigger | A day or two | Highest, measurable | Second |
| Re-engagement | Half a day | High, indirect | Third |
| Post-purchase | A day | High for repeat businesses | Fourth |
| Nurture | Days | Slow, long cycles only | When the cycle warrants it |
| Browse abandonment | Half a day | Moderate | After the above |
| Anniversary or birthday | Hours | Low | Optional |
| Behaviour triggers | Varies | Varies wildly | Only where the signal is strong |
Exit conditions are part of the design
Every flow needs a defined way out, and it is the thing most often left unspecified.
A nurture sequence that does not exit on purchase keeps selling to customers. A cart recovery flow that does not exit on the order completing sends reminders for a completed order. A re-engagement sequence with no end keeps mailing people who have made their position clear.
Write the exit condition at the same time as the trigger, not afterwards. And test it — a flow with an exit condition that has never been verified is a flow with an untested exit condition, which is functionally the same as not having one.
Collision between flows
Once there are three or four flows, a subscriber can be enrolled in two at once — finishing a welcome sequence while a cart recovery fires, or receiving a re-engagement message during onboarding.
Two mechanisms handle it. A precedence rule saying which flow wins, and a per-subscriber frequency cap that applies across everything regardless.
The cap is the more important, because it fails safe. Precedence protects against the collisions you anticipated; a cap protects against the ones you did not, and there will be ones you did not.
Before building a flow
- The three highest-return flows exist and work
- The trigger is defined precisely enough to test
- An exit condition is written, and has been verified
- It cannot collide with an existing flow, or precedence is stated
- A frequency cap applies across all flows
- A review date is in a calendar with an owner
- Someone has walked the whole flow with a real address
Frequently asked questions
How many flows should a small programme run?
Three, done properly. Welcome, one commercial trigger, and re-engagement. A fourth is worth adding only when those three are measured and current.
Should flows pause during campaigns?
Suppress campaigns for anyone currently in a flow, rather than pausing the flow. A welcome sequence and a promotion arriving the same morning read as two different senders.
How do I measure a flow's return?
Per flow, over a period, against the cost of building it once. That figure climbs indefinitely because the denominator is fixed — which is exactly why flows outperform campaigns and why the comparison is worth making explicitly.