How to Do Email Marketing Step by Step (2026 Edition)

A step-by-step walkthrough of the email marketing system I build for clients — deliverability first, then relevance, then automation — with the numbers, the code, and the failure modes.

A Tuesday morning in Dubai, and the phone call that reshaped how I build marketing systems. An e-commerce client — kitchen and home goods, roughly 8,400 subscribers — had just fired off their biggest promotional broadcast of the month. Twenty-four hours later, the founder wasn't celebrating. He was panicking.

The campaign had "gone out." The dashboard said 8,400 sends. But the open rate was 1.9 percent. Let me translate that number for you: an open rate that low does not mean people ignored the email. It means most people never saw it at all. I asked the obvious questions — when did you last check your deliverability? When did you last prune the list? Silence. I ran a quick audit. Sixty-one percent of my test sends landed in spam. There was no DKIM record on the domain at all. A third of the list hadn't engaged in nine months. The founder had spent two weeks writing a gorgeous email to an audience that mostly never received it.

That call is why this article exists. Email marketing in 2026 is not a copywriting exercise. It is a delivery problem, a relevance problem, and an automation problem — in that order. If you get delivery wrong, nothing else matters, because none of your carefully written words ever reach an inbox. Over the next eleven weeks I rebuilt that client's entire program, and the open rate went from 1.9 percent to 34 percent while revenue per send tripled. In this guide I will walk you through the exact sequence I use, step by step, with the code, the numbers, and the failures included.

Here is the definition that keeps me honest: email marketing is the practice of reaching people who have given you explicit permission to contact them, through a channel you control, in service of a relationship that produces repeat revenue. The word that carries the whole definition is permission. And email is one of the few channels you actually own — the subscribers live on your domain, not inside someone else's feed.

I think about email as three layers, and this ordering is the entire point of the article:

1. Delivery — can your message physically reach the inbox?
2. Relevance — does the right message reach the right person?
3. Automation — does the system act without you standing over it?

Most advice you read online lives in layer two. Most real-world failures happen in layer one. So we start at the bottom and work up.

Step 1: The Technical Foundation (Get Delivered)

You cannot buy your way into an inbox, and you cannot charm your way in either. Inbox placement in 2026 is decided largely by three DNS records that most small businesses have never touched: SPF, DKIM, and DMARC. Receivers — Gmail, Outlook, Yahoo — check these before they decide where your message lands, and they check them on every single send.

Here is what a correct set looks like. In your DNS provider, add three TXT records:

# SPF — which servers are allowed to send for your domain
v=spf1 include:spf.example.com ~all

# DKIM — the signing key published by your email provider
v=DKIM1; k=rsa; p=MIGfMA0GCSqGSIb3DQEBAQUAA4GNADCBiQKBgQC...

# DMARC — what receivers should do when SPF/DKIM misalign
v=DMARC1; p=quarantine; rua=mailto:[e-mail]; pct=100

Do not hand-type the DKIM key — your email platform generates it and gives you the exact record to copy. The three records do different jobs. SPF lists which mail servers are allowed to send mail for your domain, so spammers cannot easily pretend to be you. DKIM cryptographically signs each message with a private key, so a receiver can verify it genuinely came from you and was not tampered with in transit. DMARC tells receivers what to do when the checks disagree — quarantine it or reject it outright — and it reports back so you can see what is failing.

The detail that matters most: these are per-domain records, and if you send from a subdomain like mail.example.com, they must exist there too. This was the client's exact problem. SPF existed on the apex domain, but the platform was sending from a subdomain that had nothing published, so every message looked unauthenticated to the receivers.

After we fixed the records and warmed the domain (more on warming below), my test sends went from 61 percent spam placement to 96 percent inbox placement in three weeks. Same copy, same subject lines, same list. The only change was infrastructure. If your open rate is stuck below 15 percent, run this test before you touch a single subject line.

Step 2: Build a List You Own (and Clean It)

With delivery fixed, the second problem was the list itself. The client had accumulated subscribers for four years and had never cleaned a single record. List decay is real and it is brutal: industry figures put average annual churn at roughly 22 percent — people change jobs, abandon old addresses, or simply stop reading. If you keep sending to dead addresses, you accumulate hard bounces, and enough hard bounces drag your sender reputation down until even your good subscribers stop seeing you.

Three rules govern the list:

• Double opt-in only. The subscriber confirms via a link in a verification email before they are added. It cuts initial signups by about 20-30 percent, but the people who survive are actually interested, and your complaint rate collapses.
• Prune on behavior, not on time. Anyone who has not opened or clicked in 120 days goes into a re-engagement flow; if they still do not respond after two nudges, they leave. The client's 8,400 list became 5,900 after pruning — smaller, but the sends got cheaper and the engagement metrics stopped lying.
• Never buy a list. Purchased lists arrive with zero consent, produce spam complaints on day one, and can get your sending domain blacklisted before you have sent a hundred emails. A complaint rate above 0.1 percent is the danger zone at Gmail; purchased lists blow through it instantly.

Step 3: Segment Before You Write a Word

Here is the relevance layer, and this is where most marketing advice starts — wrongly. Relevance only works on top of a clean, deliverable list. Once you have one, segmentation is the highest-leverage thing you can do, because a segmented campaign routinely beats a broadcast on every metric that matters.

The taxonomy I default to is four buckets:

• Cold — subscribed but never engaged. Feed them your best content slowly.
• Warm — opened or clicked in the last 30 days. Your primary audience for offers.
• Hot / VIP — purchased, high-frequency engagers, high lifetime value. Get them exclusive access and early drops.
• At-risk — used to engage, now silent. Win them back with a direct question or a strong reason to stay.

Segmentation can be as simple as a rule against a database. On a plain Postgres subscriber table, "warm" looks like this:

SELECT email, last_purchase_at
FROM subscribers
WHERE opened_at >= now() - interval '30 days'
  AND last_purchase_at IS NULL
  AND bounced = false
  AND unsubscribed = false
ORDER BY last_opened_at DESC;

Run that before every campaign and send each segment its own version. When I finally convinced the kitchenware client to split their broadcasts this way, the warm segment opened at 38 percent while the cold segment opened at 9 percent. Both numbers became useful in a way the blended 34 percent never was — one tells you to keep selling, the other tells you to start earning attention again.

Step 4: Automate the Flows That Print Money

Broadcasts are the surface of email marketing. The compounding lives in the automated flows — triggered sequences that run forever without you touching them. Four flows cover most of the value for any product business:

1. Welcome flow — the moment someone subscribes. Open rates here are 50 percent or higher because the subscriber just asked for something. This is your chance to set expectations and make the first offer.
2. Abandoned cart flow — the highest-converting flow in e-commerce. Someone added to cart and left; two or three emails over 24-48 hours recover a meaningful slice of that revenue.
3. Post-purchase flow — deliver value after the sale, request a review, and tee up the next product.
4. Win-back flow — the at-risk segment above, with a final, honest ask.

A flow is just a state machine. Here is a welcome flow in the shape I write them:

flow: welcome
trigger:
  event: subscriber.created
  condition: source == "signup_form"
steps:
  - send: template/welcome
    delay: 0m
  - branch:
      - if: opened(welcome) == false
        send: template/best_of_series
        delay: 48h
      - if: opened(welcome) == true
        send: template/first_case_study
        delay: 24h
  - stop_when: clicks_cta == true

The technical part is boring — every serious email platform has a flow builder with exactly these primitives. The part people get wrong is measurement: they build a flow, watch the first month's revenue, and then never look again. Every flow decays as your product and audience change, so review each one monthly and kill anything that has not paid for itself. For the automation layer itself, I have been using a free email marketing tool I've used across client projects this year rather than pushing a five-person company onto a $300-a-month enterprise plan before they have the volume to justify it. The tooling matters far less than the sequence, and the sequence matters far less than the list.

Step 5: Write and Send for the Inbox

Now the copy. A few rules I enforce because they are measurable:

• Subject lines have one job: get the open. Under 40 characters, one clear idea, no deception. A subject that promises something the body does not deliver is the fastest way to accumulate complaints.
• Preview text is the second subject line. Most clients never set it, which means the "should I open this" decision gets made on a random fragment of body text.
• One ask per email. An email that asks for a reply, a purchase, and a referral all at once does none of them well.
• Cadence is a floor, not a target. Three campaigns a month to a cold list will burn it faster than one excellent campaign. When in doubt, send less.

Step 6: Measure What Matters

You will see a lot of vanity dashboards. The metrics that actually run a program are these five, with the benchmarks I use for a healthy program:

Metric	Healthy range	What it tells you
Open rate (engaged list)	30-45%	Interest in the message
Click rate	3-8%	Interest in the action
Hard bounce rate	<1%	List hygiene
Spam complaint rate	<0.1%	Permission quality
Reply rate	>0.3%	Relationship strength

Reply rate is the sleeper. A reply is the strongest possible signal an algorithm can read, and a reader who replies is a reader who will buy. Ask questions that invite replies; your inbox placement will thank you.

Production Reality: Where This Breaks

Let me give you the failure modes so you don't have to collect them like I did:

1. Domain warming. A fresh sending domain has no reputation. You cannot send 50,000 emails on day one no matter how clean the list is; ramp gradually over weeks or you will be throttled into irrelevance.
2. Sending too often. Frequency is a reputation input, not just a nuisance input. Unengaged subscribers who receive twenty emails a month become the complaints that drag your placement down for the engaged ones.
3. Transactional mail in the same pool. If your password-reset emails ride the same IPs as your newsletter blasts, one spam spike can pollute your critical transactional mail. Separate them.
4. Ignoring DMARC reports. DMARC reports arrive whether or not you read them. If a scammer is spoofing your domain, they are burning your name every single day — read the reports, or at least let your provider's dashboard read them for you.

When NOT to Do Email Marketing

Honesty, as always: email is not the answer to every marketing problem. Skip it when your list is under a few hundred people and your product does not repeat — a one-off service sale to strangers is a job for outbound and content, not a newsletter. Skip the automation until you have something the automated flow can act on; automating a welcome email for a product nobody buys yet is polishing a door nobody knocks on. And if you have a genuinely viral product with a discoverability engine, email is the retention layer, not the growth layer — invest accordingly.

The Practitioner's Checklist

• [ ] SPF, DKIM, and DMARC published and verified for the actual sending domain
• [ ] Domain warmed before the first big send
• [ ] Double opt-in enabled
• [ ] List pruned: no dead addresses, at-risk segment separated
• [ ] Broadcasts segmented into at least cold / warm buckets
• [ ] Welcome flow live and reviewed monthly
• [ ] One ask per email, subject under 40 characters, preview text set
• [ ] Complaint rate under 0.1%, hard bounce under 1%
• [ ] DMARC reports read at least once a month
• [ ] Every flow has a kill criterion and a revisit date

The Part I Keep Repeating

That client in Dubai now opens at 34 percent because the system works in order: the emails arrive, the right people get the right message, and the automated flows run in the background whether anyone remembers them or not. The founder stopped asking "how do I write a better subject line" and started asking "what is my list telling me this month" — and that shift in question is the whole game. Email is a compounding asset. Every clean subscriber, every authenticated domain, every working flow is infrastructure you own forever, and unlike any social feed, nobody can change the algorithm under you overnight. Build the delivery, then the relevance, then the automation — in that order, and never skip the first.

---

*Gulshan Yad