The deliverability work that comes before any retention flow

Most retention plateaus are deliverability problems disguised as content problems. A welcome flow cannot outperform an inbox that filters it to spam, and a clever subject line cannot rescue a sender domain with a bad reputation. This is for retention operators and ecommerce founders who suspect the email program is under-delivering and want to fix the layer underneath before touching another flow. The deliverability work that comes before any retention flow is week one of any program we run.

The signal that deliverability is the problem and not the copy is usually visible in the data. Open rates lag the category benchmark by five to ten points across every flow simultaneously. Click rates look proportional to opens, which means the copy is fine and the audience is not seeing it. Unsubscribe rates are low, because the audience that would unsubscribe is not getting the email in the first place. Spam complaints from a small fraction of recipients are dragging the sender reputation. Fixing the welcome flow copy in this scenario does nothing. The flow lands in the same compromised inbox and the numbers do not move.

What deliverability work means before any retention flow

Deliverability is the chain of technical and operational decisions that determine whether an email lands in the primary inbox, in promotions, or in spam. The chain has roughly six links: domain authentication, subdomain isolation, list hygiene, send reputation, content quality, and recipient engagement signals. Most retention programs touch only the last two and assume the first four are someone else taking care of them. They are usually not.

The work splits into authentication (SPF, DKIM, DMARC), structural isolation (a dedicated sending subdomain), hygiene (suppressing bounces and the long tail of unengaged subscribers), and reputation maintenance (warming the subdomain, monitoring complaint rates, watching the bounce rate trend). Each link in the chain has a single fix. None of them is glamorous. All of them have to be in place before the flow numbers move on their own.

Why this is week one and not week six

Sequencing matters. A welcome flow built and shipped onto a compromised sender domain trains the audience to look for the brand in spam, which is a habit that does not unwind even after the deliverability fix. The first impression of the program is the inbox folder it lands in. Fixing deliverability in week six, after the flows are live, takes longer to recover from than fixing it in week one before the flows ship. The cost of doing it later is paid in months of regression, not weeks.

What gets skipped most often

List hygiene. Teams will set up SPF and DKIM because the ESP onboarding flow makes them. They will sometimes publish DMARC. They will rarely suppress the long tail of unengaged subscribers, because the list size feels like an asset, and the unsubscribe-now decision feels like throwing away an asset. It is not. The long tail of unengaged subscribers actively suppresses the sender reputation by reducing the engagement-rate signal mailbox providers use to grade incoming senders. Cutting them out is the single highest-leverage deliverability move available, and it is the one most teams skip.

The four pillars of deliverability before any retention flow

Authentication. SPF, DKIM, and DMARC alignment is the floor. SPF tells receivers which servers are authorised to send for the domain. DKIM signs each message so receivers can verify it has not been tampered with. DMARC sets the policy for what to do when SPF or DKIM fail and the receiver should reject or quarantine. All three published, all three aligned with the visible from-address, no exceptions. This is foundational and most ESPs make it discoverable, but the work to actually publish the records correctly still falls to the brand.

Subdomain isolation. Subdomain isolation matters more than people think. Mixing transactional sends and marketing sends on the same subdomain means a marketing complaint spike pollutes your transactional reputation. The cost of separation is one DNS record. The benefit is that the order confirmation emails still land in the primary inbox even when the marketing program has a bad week. Use a subdomain like send.brand.com for marketing and a different one for transactional, both authenticated, both isolated.

List hygiene. A list-hygiene pass that suppresses bounced and unengaged segments older than 90 days. List hygiene is the lever most teams skip. The mechanic is straightforward: anyone who has not opened or clicked in 90 days is excluded from broadcast sends. The flows can still target them with reactivation messages, but the broadcast cadence does not hit them, because their lack of engagement is what is dragging the engagement-rate signal that mailbox providers use to grade the sender. Cutting them out lifts the rate visible to receivers, which lifts inbox placement for everyone else.

Reputation maintenance. Once authenticated, isolated, and hygienic, the subdomain has to be warmed and the reputation has to be maintained. Warming means starting send volume low and ramping over a few weeks, especially on a fresh subdomain. Maintenance means monitoring the bounce rate, the complaint rate, and the gradient of the engagement rate week over week. A reputation graph that is trending down is the early warning that something in the program is regressing, and it shows up before the revenue graph does.

The deliverability runbook

1. Verify SPF, DKIM, and DMARC are published and aligned. Use a tool like mxtoolbox or the ESP-provided diagnostic. If DMARC is at p=none, leave it there for two weeks of monitoring before ramping to p=quarantine. 2. Configure a dedicated marketing subdomain. Authenticate it. Move marketing sends to it. Leave transactional sends on a separate subdomain or the root domain. 3. Run the list-hygiene pass. Identify all subscribers who have not opened or clicked in 90 days. Suppress them from broadcast cadence. Build a reactivation flow for them as a separate workstream, which is a different question from the broadcast send. 4. Warm the new subdomain if it is fresh. Start at 10 to 20 percent of expected daily volume on the first send. Ramp over two to three weeks. Watch the bounce and complaint rates as you ramp. 5. Set up reputation monitoring. Bounce rate, complaint rate, engagement rate, all tracked weekly. A complaint rate above 0.1 percent on any campaign is a flag. A bounce rate trending up week over week is a flag. 6. Document the deliverability state in the audit. Score each pillar. The output is the contract for the rest of the program.

When deliverability work is not the answer

Brands with a clean deliverability layer who are still under-performing have a different problem. Authentication is correct, the subdomain is isolated, the hygiene pass is current, the reputation is sound, and the open rates are still flat. In that case the issue is upstream of the inbox: the offer, the audience definition, the segment architecture, or the cadence. Spending another week on deliverability there is wasted work. The diagnostic step has to happen first to know which side of the line the brand is on.

Brands deliberately sending to a small, highly engaged list often get away without the full deliverability stack because the engagement signal is so high that mailbox providers grade the sender favourably regardless. This is rare, and it stops being true the moment the list scales. Below a few thousand engaged subscribers the mathematics is forgiving; above 10k it is not. Anyone planning to grow past that threshold should put the deliverability layer in place before the growth, because retrofitting it later means paying the warming cost a second time.

What success looks like

A clean deliverability layer is invisible. The brand does not see headlines about it. The flows ship, they land in the primary inbox at category-leading rates, the open rates clear the benchmark, and the program revenue moves on its own. The marker that the work is done is not a number on a dashboard. It is the absence of a problem that used to suppress every other number on the dashboard.

Quantitatively, expect the open-rate floor to lift by five to fifteen points across every flow simultaneously when the deliverability work is done correctly. The lift is uneven by audience and by category, but it is broad: every flow moves, not just one. The revenue lift follows in 10 to 14 days as the rebuilt flows hit a clean inbox, and within the first 90 days the brand should be in the published 25 to 40 percent of revenue from retention band that the foundational work supports.

FAQ

How long does the week-one deliverability work actually take? The technical work (SPF, DKIM, DMARC, subdomain) is one to two days for someone who has done it before. The list hygiene pass takes a few hours of analysis plus a Klaviyo configuration session. The warming, if a fresh subdomain is involved, takes two to three weeks but runs in the background while the rest of the program is being built.

Should we use a separate subdomain for transactional and marketing? Yes. The cost is one DNS record. The benefit is that a marketing complaint spike does not pollute the transactional reputation that order confirmations and shipping notices depend on. This is the cheapest deliverability decision available and most teams skip it.

Is suppressing the long tail of unengaged subscribers actually safe? Yes, and the brands that try it almost always wish they had done it sooner. The suppressed segment was already not converting. Removing them from the broadcast cadence raises the engagement signal mailbox providers use to grade the sender, which improves inbox placement for the engaged audience that is actually driving revenue. A separate reactivation flow targets the suppressed segment without polluting the broadcast send.

What if our flow numbers are good already? Then the deliverability layer is probably already sound and the work has been done. Audit it anyway. Most brands with good flow numbers have at least one weak link in the deliverability chain that, if fixed, lifts the floor a further few points. The audit is cheap; the answer is high-signal.

Read more

- The first 90 days of Klaviyo retention work: https://www.arthea.ai/article/first-90-days-klaviyo-retention-roadmap - The retention, CRO, and content compound effect: https://www.arthea.ai/article/retention-cro-content-compound - Email and SMS retention engagements: https://www.arthea.ai/email-and-sms

If you want a 30-minute architecture review on deliverability for your brand, the calendar is here: arthea.ai/book.