Email Marketing Automation for Shopify: Flows That Convert 24/7

A store doing 400 orders a month is abandoning roughly 930 checkouts in the same month. Recovering 5% of them at an $80 average order value is $3,720, from a flow that takes an afternoon to build.

That is the case for automation, and it is a good one. But it assumes the emails reach an inbox. A flow that lands in the spam folder is worth exactly zero, and since February 2024 the requirements for landing in an inbox stopped being folklore and became published rules with thresholds attached.

Get that part right first. The flows are the easy half.

Authenticate the domain before you automate anything

Three DNS records do three different jobs, and most people who can name all three cannot say what each one asserts.

SPF is a TXT record listing which servers are allowed to send mail on behalf of your domain. It authenticates the envelope sender, the address bounces go back to. It says nothing about the From address a human actually sees.

DKIM puts a cryptographic signature on each message, with the matching public key published in your DNS. It asserts two things: someone holding your private key signed this message, and the signed parts were not altered in transit. It does not assert who that someone should be.

DMARC is the record that ties the other two to the visible From domain. It requires that a passing SPF or DKIM result be aligned with the domain the reader sees, tells receiving servers what to do when nothing aligns, and gives you an address for aggregate reports.

That last one is where the common mistake lives. A DMARC record with a policy of p=none asks receivers to take no action on failure. It gets you the reports, which are genuinely the most useful diagnostic you will ever run on your own mail, and it satisfies a checkbox that says you have DMARC. It does not stop anyone spoofing your domain, and it does not improve your placement on its own. It is the correct place to start and the wrong place to stop. Read reports for two to four weeks, fix the legitimate senders that fail alignment, then move to quarantine and then to reject.

The 2024 Gmail and Yahoo bulk sender requirements are the practical checklist. If you send more than 5,000 messages a day to Gmail addresses you need SPF and DKIM and a DMARC record on the sending domain, one-click unsubscribe implemented in the message headers and honoured within two days, and a spam complaint rate kept under 0.3%. Google's own guidance is to aim under 0.1%.

Sit with that 0.3% number, because it is smaller than it sounds. It is three complaints per thousand delivered messages. On a 40,000-person send, twelve people hitting the spam button puts you at the ceiling.

Build the five flows in order of revenue per hour of setup

Priority order, not alphabetical order. Build them in this sequence and stop when you run out of appetite.

• Abandoned checkout. Three emails at roughly one hour, 24 hours and 72 hours. What decides whether it works: whether you captured the email address before the shopper left. No capture, no flow, and the fix is a form field, not a copy rewrite.
• Welcome. Three to five emails over ten days, triggered by a signup. What decides whether it works: whether the first email arrives within minutes. Interest decays fast, and a welcome email that arrives the next morning is a different, weaker email.
• Post-purchase. Four emails: shipping confirmation, a how-to timed to arrival, a review request seven to fourteen days after delivery, and a replenishment nudge at your median inter-purchase gap. What decides whether it works: timing everything off delivery date rather than order date.
• Browse abandonment. One or two emails, four to 24 hours after the visit, and only to identified, engaged contacts. What decides whether it works: restraint. This is the flow most likely to read as surveillance to somebody who has bought from you once.
• Winback. One to three emails starting at your churn threshold. What decides whether it works: whether that threshold came from your own inter-purchase data or from a round number somebody wrote in a blog post.

Work the abandoned-checkout arithmetic yourself

Baymard Institute's aggregate of documented studies puts cart and checkout abandonment around 70%, and that figure is stable enough to plan against.

At 70% abandonment, completed orders are 30% of started checkouts. So 400 orders a month implies 400 divided by 0.30, or about 1,333 started checkouts, of which roughly 933 are abandoned.

Recover 5% of 933 and you get 47 orders. At an $80 average order value, that is $3,720 a month and about $44,600 a year.

Now the correction that keeps the number honest. You can only mail the abandons where you have an email address. If checkout captures the address on 60% of abandons, your addressable pool is 560, and 5% of that is 28 orders, or $2,240 a month. Improving that capture rate is often worth more than improving the emails.

Why the first email carries no discount. A large share of abandonment is logistical rather than financial: a shipping cost that appeared late, a declined card, a phone call, a tab closed on a train. Those people were coming back. Discounting the first email pays them to do what they were already going to do, and it costs you the same margin on every one of them.

Why the third one usually should not either. The third email is where discounts get parked by default, and it is the email your repeat customers learn fastest. Once a segment of your buyers knows that abandoning produces a code in 48 hours, you have not built a recovery flow, you have built a self-service discount machine, and you cannot switch it off without a visible price rise.

Cost it out. Those 47 recovered orders at $80, with a blanket 10% code, hand over $8 an order, so $376 a month. If only a third of those buyers needed the incentive, roughly $250 a month bought nothing except a lower baseline that now persists. Put the third email's effort into the actual objection instead: the free-shipping threshold, the return policy, financing, or a genuine stock constraint if one exists. Save discounting for a segment you have proven to be price sensitive.

Prune the list to raise revenue, not to lower it

Cutting subscribers feels like cutting revenue. It usually is not, and the mechanism is worth understanding rather than taking on faith.

Inbox placement at the major providers is heavily influenced by per-recipient engagement. Repeatedly delivering mail to people who never open teaches the filter that your mail is ignorable, and that judgement attaches to your sending domain, not to those individuals. The people who do want your mail start seeing it in the Promotions tab, then in spam.

The arithmetic. A 40,000-person list where 12,000 have not opened or clicked in twelve months. Remove them and volume drops 30%, while revenue per send holds or rises because placement improves for the remaining 28,000. Since email is metered, you also stop paying to send 12,000 copies of every campaign to nobody.

Sunset properly rather than abruptly:

1. Define disengaged precisely. No open or click in six months, and no order in twelve.
2. Send two or three re-permission emails with a plain subject line and an obvious way to stay.
3. Suppress rather than delete, so they cannot be re-imported by accident later.
4. Keep them excluded from campaigns but leave transactional mail untouched.
5. Watch your complaint rate for a month afterwards, because that is the number the change is meant to move.

What email cannot fix

Email will not fix a product nobody wants. If your repeat rate is low because the product disappoints on arrival, a better winback flow raises the rate at which disappointed people are reminded of their disappointment. Read your returns reasons before you build a retention programme.

Discount-led flows compound, and the damage is slow enough to miss. Track the share of your revenue that arrives at full price, monthly. If that share is falling while total revenue is flat, your flows are converting existing demand more cheaply for the customer, and the trend does not reverse on its own.

Sending to a purchased or scraped list will burn your sending domain in weeks. Those lists carry spam traps, addresses that exist specifically to identify senders who did not collect consent, and hitting one is not a slow degradation. Domain reputation is not something you can buy back, and moving to a fresh subdomain does not inherit a rescue, it inherits a warmup period and the same list.

One more, about a number you have seen. The often-quoted 4200% ROI figure, or $42 back for every $1 spent, comes from an industry association's survey of marketers reporting their own attributed returns. It is self-reported, it is a survey, and it carries every attribution generosity that email measurement is known for. Do not plan against it. Plan against arithmetic you did yourself, like the $3,720 above, where you can see every input.

What Synton does here

The Email app is the creation side. In Builder you describe the email or the series you want in plain language, and a planner produces the series structure before a composer writes each email, so a five-email welcome series arrives as five coherent emails rather than five unrelated ones. You refine by describing the change rather than editing raw HTML, then save into Templates.

Flows is the flow canvas: add a trigger, then send, wait, branch on a condition, or update the customer. Worth knowing before you plan around it, the triggers implemented today are back in stock, replenishment reminder, review request, birthday greeting, winback offer, low inventory alert, subscription renewal and browse abandonment. Abandoned checkout is not one of them. Cart recovery lives in the Commerce app instead, under Abandoned Carts, where you open a cart, see what the shopper left behind, and retrigger the recovery sequence or send an SMS.

Deliverability shows the SPF, DKIM and DMARC rows in one place. Bulk sending goes through Amazon SES with per-shop rate limiting, warmup pacing for a new sending domain, and automatic suppression of hard bounces and complaints. One caveat that is documented rather than hidden: a failed DNS lookup is currently reported as a record failure rather than as an unknown state, so if all three rows fail at once, check with an external DNS tool before you change anything.

Two of the 18 autonomous agents apply. Email Health checks SPF, DKIM and DMARC configuration. Cart Recovery analyses cart abandonment patterns, which makes it an analyst rather than a sender. Enabling either one puts it in suggest mode and does not schedule it; a recurring run is a separate switch.

On measurement, be careful with one figure: email revenue on Synton's scoreboard is modelled rather than measured, and the response marks it as an estimate. Use it to compare campaigns and flows against each other, not to reconcile against your accounting.

If you already run Klaviyo, that connection is bidirectional and real: Synton can read campaigns, flows, lists, metrics and segments, and push profiles, segment members and commerce events such as Placed Order, Added to Cart and Started Checkout. Mailchimp, Omnisend, Brevo and ConvertKit are a different story. Adapters exist in the codebase, but no supported merchant connection reaches them today, so the working path from those platforms is a CSV export into Import Subscribers. Better to know that now than after a migration plan.

Sending is gated: it needs Operator access or higher, generation consumes credits, and email itself is metered at 10 emails per credit.

What to do on Monday

1. Look up your SPF, DKIM and DMARC records. If DMARC is missing, publish p=none with a reporting address today, read the reports for two weeks, then move to quarantine.
2. Send yourself a campaign and confirm the one-click unsubscribe header is present and honoured. Pull your complaint rate and compare it to 0.3%.
3. Do the abandoned-checkout arithmetic for your own store, including your email capture rate, before you write a word of copy. Then build the three-email flow with no discount in it.
4. Compute your median inter-purchase gap from your own order history and set replenishment and winback timing off that number rather than off 30, 60 and 90 days.
5. Define your disengaged segment, run a three-email sunset, and suppress the rest.

Creating a Synton account is free and needs no card, so you can connect a store and read your own deliverability status before deciding anything. Sending and AI generation both need a paid plan.