What 8 Years of Email Marketing Taught About the Inbox
The first campaign I sent was to 3,400 people. The subject line read something like "Exciting News Inside!" The open rate was 9.4 per cent. It was bad - it was not. But nobody understood because the vast majority of email benchmarks people cited in 2016 were inflated, outdated, or taken from industries that had nothing to do with ours. I learned the hard way, with split tests and subscriber complaints and one particularly humbling quarter review.
The subject line problem no one discusses
Everyone tells you to A/B test your subject lines. What they don't tell you is that most A/B tests in email are statistically meaningless unless you have a large enough list - usually north of 5,000 engaged subscribers, minimum. If you are sending to 800 people, you are not doing a test. You're doing a coin flip with extra steps.
But subject lines are important. But not for the reasons most guidebooks suggest. The open rate isn't just a measure of how clever or curiosity-gap-inducing your subject line is. It is influenced by the sender's reputation, timing of delivery, inbox placement, and, critically, how well your past emails have performed with that audience.
What moves the needle is specificity, not cleverness. Almost every time, with almost every audience, "How we cut client churn by 34% in one quarter" trumps "The secret to keeping clients". Not because it's shorter, not because it uses a number (though that helps), but because it indicates a real payoff. The reader knows immediately what they're gonna get.
Segmentation Is Not What Most People Think It Is
Segmentation is talked about as if it's just a technical feature - break up your list by demographics, by location, by purchase history, run different campaigns. That's part of it. But a better way to think about segmentation is behavioural: what has this person done recently, and what does that tell me about where they are in the relationship with us?
One caveat: don't segment too early. Twelve different segments in a list of 2,000 subscribers is no help. You'll dilute your sending volume, confuse your analytics, and spend more time managing audiences than writing. Begin with 2 or 3 behavioural differences that matter. Only make it more complex when the data tells you to.
Frequency: More is not necessarily more.
Many marketers still think that the more emails they send, the more revenue they will make. That may be true in some cases - flash sales, time-sensitive promotions, transactional contexts. It's a short-term tactic that usually degrades list quality over time.
The right frequency is always the frequency your audience told you they wanted, or the frequency your data shows they can sustain. If you don't know what that is, test lower, not higher.
The fundamentals aren't flashy, but they're non-negotiable: correctly set up SPF, DKIM and DMARC authentication. Clean list with regular purging of hard bounces and long inactives. It has to be from a domain which had a warm history, not a cold one. Avoiding spam trigger language, and more importantly, keeping real engagement rates, as inbox providers are increasingly using behavioural signals, not keyword flags, to decide placement.
The Copy Question: Voice, Length, and the Template Trap
Most email marketing content doesn't fail because it's badly written, but because it sounds like it was written by a committee or from a template. Readers sense it immediately, even if they can't articulate the reason. It's flat, no real person making a real argument or a real experience.
Length matters less than most guides would have you believe. A 600-word email can seem short when it is engaging. When it's padding, a 90-word email can feel interminable. Write until you have nothing left to say. Cut from habit, not intention. That edit, cutting the line you wrote because you thought the email needed one more line, is often the difference between good email copy and forgettable email copy.
Deliverability: The Hidden Barrier to Performance
You might have the best subject line in the world, perfectly segmented audiences, and copy that is actually well-written - and still have campaigns that quietly underperform because your emails are not hitting the inbox. Deliverability is the unseen ceiling most marketers ignore until something visibly goes wrong.
Which Metrics To Really Trust
Apple's Mail Privacy Protection, rolled out in late 2021, made open rates far less reliable. Now, a large percentage of opens are pre-fetched by Apple's servers, not real human opens. That doesn't mean open rates are useless -- they still indicate rough trends and can flag deliverability issues -- but to treat them as a precise measure of engagement is no longer defensible.
More reliable engagement signals are the click-through rate, the reply rate (if you ask for replies), and the conversion rate on a particular action. Revenue per email and list growth rate are helpful at the macro level. Unsubscribe rate is a lagging indicator, so by the time it spikes, the damage is usually already done.
The most underrated metric: what happens post-click? If 300 people click your CTA and 280 of those bounce from the landing page right away, the email did its job, but something else in the funnel didn't. Email marketers who believe their job ends at the send button are missing out on a lot of attribution and optimisation opportunities.
Automation: When It Helps and When It Creates Distance
Automation is hugely powerful for triggered sequences -- welcome flows, abandoned cart reminders, post-purchase follow-ups, and re-engagement campaigns. These are contexts where the timing is driven by behaviour and the communication is anticipated. When done well, automation is almost invisible: the subscriber gets something relevant at a relevant moment, and it feels thoughtful, not mechanical.
What starts to break down with automation is when it's used as a replacement for real editorial judgment. I've seen brands build out these massive nurture sequences - fifteen emails over ninety days --- and then just leave them for two or three years. The content contains products that are no longer available, offers that are no longer valid, or messaging from an older brand era that no longer reflects the company's current positioning. Automation does not maintain itself. It's a lesson that costs people more than they think.
There is also a more subtle risk: too much automation can make a brand seem cold, when a little warmth would be better. At the right moment in the customer relationship, a founder or named team member sending a handwritten-style email (not literally, but tonally) can outperform a polished automated sequence because it feels real. When you shouldn't automate.
The Harder Truth About What Works With Email In The Long Term
After all the testing, optimisation, and deliverability management, I keep coming back to something simpler and less technical than any of it. Email works when the person who is sending it has something worth saying to the person who is receiving it. That seems clear. That's obvious. But it's really hard to keep it going at scale, across a team, over the years.
There are a few things brands whose email programs I've seen hold up over time have in common. They view their list as a relationship, not a distribution channel. They'd rather send fewer emails than compromise on quality. They've got somebody-not a system-who really cares whether the email is good. And they are brutally honest with themselves when something isn't working, rather than rationalising the numbers.
That is not a strategy. That is discipline. And it's harder to build than any automation stack or segmentation logic. But it's also the base for everything else.
FAQs
Q1. How often should I send marketing emails?
There is no one-size-fits-all answer. It's all about what your audience expects and how well you can continually deliver value. "A good starting point for most companies is one or two emails a week. The more important principle is consistency.
Q2. What is a good open rate for email marketing?
Benchmarks can vary widely by industry, by list size, and by the way the metric is measured. After Apple MPP, open rates are inflated in Apple Mail environments, making cross-industry comparisons less meaningful than before.
Q3. How do I reduce unsubscribes?
Unsubscribes are generally a symptom, not a cause. This happens when the email frequency is too high for the value offered, when the content doesn't align with what the subscriber signed up for, or when the onboarding process sets the wrong expectations from the start.
Q4. How effective are personalised emails?
Basic personalisation -- like using a first name in the subject line or greeting -- has diminishing returns and can feel hollow if the rest of the email is generic.
Q5. How frequently should I clean my email list?
Yes, and more aggressively than most people want. You might think that keeping a large number of inactive subscribers is protecting your potential reach, but it's actually hurting your deliverability and skewing your performance data.
