Email marketing has a measurement problem and a writing problem. They are related and almost nobody is talking about both at once. This is the argument for why that matters and what happens when you fix both simultaneously.
Most brands are making email decisions based on wrong numbers. Default attribution windows calibrated for a median use case with nothing to do with their specific product. Open rates inflated by bot traffic. Revenue figures that overcount by 20-40% because open-based attribution is still switched on. The data looks like data and it sits in dashboards and gets reported in monthly meetings. Problem is, a meaningful proportion of it is fiction.
Fix this and you get accurate numbers and better decisions. But it isn't the whole argument.
Most ecom email is built on visual logic. It's templated with a header image, product photo, headline, body copy, button and a footer. The text exists to fill the spaces between design elements. It labels the buttons and describes the products. It is functional infrastructure inside a visual container, written competently and efficiently, indistinguishable from every other brand in the same category sending the same kind of email to the same kind of list on the same day.
Fix this and you get better open rates and higher conversion. But again - it isn't the whole argument.
There are email marketers who are good at data - rigorous measurement, significance-tested experiments, calibrated attribution. The copy is functional. Nobody is going to forward it to a friend saying "you gotta read this!"
On the flip side, there are email copywriters who are good at story. Warmth, specificity, a voice that makes readers feel they know the person behind the email. The measurement, however, is superficial - open rates are checked, click rates are checked, and if someone makes a purchase then whoohoo - icing on the cake. But nobody is running a holdout group to test whether any of it is actually working.
These two worlds produce different practitioners with different skills and different blind spots, and they rarely overlap. Both are missing something.
The highest-performing email programmes in ecom are built at the intersection of scientific rigour and genuine literary craft. Almost nobody operates there because the two worlds don't usually produce the same person.
I wrote a seven-email welcome sequence for a UK baby goods brand. It was text-based. There was no template, no countdown timer, no scarcity. The discount code appeared once in email one and was never mentioned again. The remaining six emails contained a story about a man who had never eaten tiramisu, a conversation about Forrest Gump, a frank account of a baby attempting to bite his mother's forehead, and a meditation on the difference between parenting rules and parenting guidelines.
The industry benchmarks are 50-60% open rate on email one, declining to 25-35% by email four. Conversion sits around 2-4%. This sequence held 94.47% across seven consecutive emails. Some of those emails had 100% open rates. Why? Because the writing was genuinely good enough that people wanted to find out what happened next. That is a narrative mechanism; it has nothing whatsoever to do with subject line optimisation or button colour.
I wrote a three-email abandoned cart sequence for the same brand. The first email contained a 400-word story about a man sweating at airport check-in because he had forgotten to apply for visas for his family of six. The story was true and was written as a postscript. It had nothing to do with baby stuff.
The visa story wasn't chosen from a list of possible, industry approved, let's-put-a-story-in-the-ps-of-an-abandoned-cart-email anecdotes. They don't exist. Because putting a story in the ps of an abandoned cart email isn't something taught at marketing school.
It was a real memory of the store owner (who was writing the email) that happened to contain an emotional parallel to the experience of abandoning a cart because life got in the way. Four children; chaos; the store owner trying to hold it together; the husband making it worse, but urviving anyway. That is the emotional landscape of the customer. The story put the reader there before asking them to buy.
Shannon entropy is an eighty-year-old mathematical formula from information theory – used in signal processing, data compression, cryptography. I applied to email subject lines to measure how predictable your format is across five dimensions: capitalisation, word count, punctuation, opener type, tone signal.
Most accounts score under 0.2. Their lists have habituated to their patterns. The brain categorises each incoming email before it is opened – they think "I've seen this before, I know what it is, I'll give it low priority" – and open rates decay as a result. Not necessarily because the content is worse (the content could be the best thing you've ever written), but because the signal looks identical to every previous signal.
This is nothing to do with intuition; it's actual scientific measurement. The same instinct that identified signals in atmospheric data identifies the patterns causing open rate decay in Klaviyo accounts. The methodology transfers; the data may be different, but the rigour is identical.
The data thing is cool, but without the writing it's a well-calibrated machine producing forgettable content. The writing without the data is entertaining work operating without feedback, improving slowly and unmeasurably. Together they produce something that neither produces alone.
The subject line entropy experiment is the clearest example. The measurement identifies the problem – high predictability, habituating list, decaying open rates. The creative work fixes it – systematic format variation, controlled entropy, subject lines different enough from the previous ones that the brain cannot pre-categorise them. The measurement then tells you whether the fix worked. The data and the writing are solving the same problem from different directions.
Most email marketing advice is optimisation advice: test your subject lines, improve your send time, segment your list, yahdahyahdahyahdah. Yes, all of this is useful but none of it is sufficient. Optimisation assumes there is something worth optimising. If the underlying email is indistinguishable from everything else in the inbox, optimising it produces marginal improvements on a very low baseline.
The counter-argument is "be more human in your emails" - but is a platitude. Everyone and their uncle says it. Nobody proves it works or explains the mechanism.
The counter-argument is more specific: the highest-performing email programmes are built by people who can do two things simultaneously – run significance-tested experiments on every major decision, and write creative emails that function as narrative, with character arcs, emotional landscape mirroring, and the kind of specific true story that makes a reader feel they have been somewhere rather than been sold something.
When you do that you get 94.47% average open rates, 10.98% conversion rates and 17% abandoned cart recovery rates. This is what happens when you stop choosing between rigour and craft and start insisting on both.