We live in a world full of email advice. Too much advice, if I'm being honest. Every week, there’s another thread, another framework, another “10 ways to personalise your emails”, another example screenshot of a brand or business that’s allegedly doing it brilliantly.
And the message you keep hearing is basically this:
You need to send segmented emails
You need to be personalised
You need to be hyper-personalised
Cool - ok.
But no one tells you how!?!!?!
No one defines what “personal” actually means in a modern inbox. No one tells you what to look for, no one gives you a practical way to make it real in your programme or strategy without building a Frankenstein automation spiderweb that ends up creating more friction than value.
And for me, that’s one of the biggest flaws in email marketing right now. We have an industry that shouts about personalisation (and loads of crap), but keeps giving people cosmetic tactics instead of contextual strategy.
So this blog is my attempt to fix that. With a clear reframe, a practical framework, and exercises you can actually use.
Because in 2026, personalisation is not a first name token or 'personalised content'
It’s the right message landing at the best time for this person, given what they are actually signalling, given what they are actually saying, given what they are actually showing you.
And when I say time, I do not mean “send at 10:14am because your ESP said so.” Send time optimisation is stupid (end of ).
I mean: is this message arriving at the right moment in their reality?
We all love the phrase “right message, right person, right time.”
It’s become a mantra, a slogan, a tagline we put in decks (I should get it tattooed on my forehead at this point).
But most teams use it as a content prompt, not a behavioural question.
They interpret it like this: How do we tweak the creative so it feels more personal?
The evolved version is much harder, and much more useful:
Is this the right message for this person, at this moment, given what they are actually signalling?
Because people are always signalling something. Not always with words, not always with forms, often with behaviour.
They’re telling you:
they’re interested
they’re warming up
they’re hesitating
they’re distracted
they’re confused
they’re overwhelmed
they’re actively shopping
they’re actively avoiding you
it’s the wrong time
it’s the right time
And that is where personalisation lives.
Not in “Hi Sarah”.
Let’s call the old version what it is: cosmetic personalisation is the stuff people see inside the email that makes you feel like you’ve “personalised” it:
You put their first name in the subject line because you think it boosts clicks, you insert product blocks based on their last click, you throw in “recommended for you”, you add a dynamic discount timer or you say "you might like".
In B2B, the cosmetic version tends to look like:
“Hi {FirstName}, here’s a case study in {Industry}” (pulled from a dropdown somewhere). Or the classic: “As a {JobTitle}, you’ll love this.”
None of this is truly personal anymore. People know you know their name, people know you can pull their company name into a sentence. It isn’t impressive. It isn’t meaningful. It doesn’t create trust.
The reason it doesn’t work isn’t because those tactics are “bad” in isolation. It’s because they avoid the real job.
The real job is context.
Context is what makes an email feel personal.
It’s the feeling of: this landed at the right moment.
It’s the feeling of: this is what I needed.
It’s the feeling of: this understands where I am.
You can have a completely unpersonalised plain-text email with no tokens at all, and it can feel deeply personal if it reflects reality. And you can have a beautifully designed, hyper-dynamic email that feels wildly irrelevant because it ignores the moment.
Direct data is great. If someone tells you what they want, you should absolutely use it.
But behavioural data is often more honest, more current, and more predictive.
Someone can fill in a form and say they’re “interested in X”, then their reality changes a week later. Priorities shift, budgets change, feelings change, needs change and their team changes. They get pulled into another project, they solve the problem another way, they buy from somewhere else.
Behaviour doesn’t lie as easily.
If you’re lucky enough to have systems that track website activity, content engagement, lifecycle movement, pipeline progression, purchase behaviour, support interactions, sales interactions, and frequency shifts, you’re already ahead (this is why I LOVE HubSpot).
Because behavioural signals tell you what is true now.
And this is why intent beats personalisation.
Intent is the behavioural truth underneath the email address.
To make this practical, we need to stop treating “intent” as one blob (I love the word blob).
Not all signals are equal and not all actions mean the same thing. And this is where you go wrong: they see any behaviour and assume it means “send more”.
Intent needs to be split into three buckets because each bucket should produce a different response.
Here’s what I want you to do before you build a single email.
Open a doc, or a spreadsheet. Whatever you use to think.
And build a simple matrix with three columns:
Active intent
Passive warming intent
Negative intent
Now list every signal you can track in your ecosystem.
Not what you wish you could track. What you can track right now.
If you’re stuck, start with:
website behaviour
page visits
repeat visits
product views
content consumption
form submissions
purchase behaviour
time since last purchase
support interactions
sales interactions
lifecycle stages
engagement shifts
Then, for each signal, write one sentence:
If this happens, what is most likely true?
You are not writing rules yet. You are building behavioural hypotheses.
This is the starting point for real personalisation.
Prediction is where personalisation becomes mature.
And I want to be clear here: prediction is not AI hype. Prediction is not “the system knows what everyone wants.” Prediction is not “we assume they’ll buy now.”
Prediction is disciplined pattern recognition.
It’s recognising what typically happens next based on past behaviour and real data, then using that understanding to improve timing and relevance.
There are two layers to prediction in email:
That second layer is where most email programmes fail. Because a lot of teams see a behaviour and assume the next logical step is a purchase.
But the next step is often confidence, or proof, or reassurance, or objection resolution.
Prediction done well allows you to say: “Given this pattern, what is most likely holding them back?”
This is why the smallest portion of your emails often drives the most revenue.
In B2C, most revenue usually comes from a handful of flows: welcome/orientation, abandoned basket, post-purchase, replenishment, maybe a winback. Not the daily campaigns. If your revenue is primarily coming from batch campaigns and not lifecycle flows, something is off. Either your flows are weak, your tracking is broken, or your programme has become a broadcast engine rather than a behaviour-responsive system.
In B2B, prediction is often about identifying when someone is moving from passive interest into active evaluation. It’s about knowing when it’s appropriate to reach out, and what support they need before that outreach feels welcome rather than intrusive.
And yes, AI can massively accelerate prediction because analysing patterns manually is not scalable. But prediction does not require AI to start. You can begin with the data you already have and build simple benchmarks.
The risk here is that prediction can easily turn into assumption if you’re sloppy.
If you treat prediction as certainty, you create friction, if you treat prediction as probability, you build relevance.
Everyone in your database is different. Some people need three exposures. Some need ten. Some will buy quickly. Some will buy in six months. Some will never buy, but will still be influenced by awareness and familiarity.
Prediction is about increasing the likelihood that what you send matches reality. It is not permission to presume.
Take your top five signals from your matrix. For each one, write answers to these prompts:
What is most likely true right now?
What are the top three objections that could be present?
What would make this message feel perfectly timed?
What would make it feel irritating or tone-deaf?
What should we suppress or exclude if this signal appears?
You’re not building automation yet, you’re now building judgement.
And judgement is the difference between “AI personalisation” and actual personalisation.
We’ve been taught to treat opens and clicks as the primary signals in email because for a long time, that was all we could see.
But opens and clicks are not behavioural truth. They’re surface indicators. Often messy ones.
An open can mean:
They read it and loved it
They opened it to delete it
They opened it by accident
They opened it because their inbox preview triggered it
They opened it because they were searching for something
They opened it because they wanted to unsubscribe
A click is a stronger signal, but it still doesn’t stand alone
A click becomes meaningful when you pair it with what happens next. Did they land where you intended? Did they stay? Did they take a meaningful action? Did they return later? Did their behaviour change?
This is why your signal ecosystem should include things like:
Content depth and content clustering, which tells you what topics they’re actively exploring rather than what they clicked once
Frequency shifts, which often reveal warming or cooling behaviour more reliably than one-off engagement.
Website behaviour, especially high-intent page patterns. Lifecycle movement, because where someone is in the relationship matters more than what they opened today.
Pipeline signals in B2B, and purchase gaps in B2C.
Support interactions and sales interactions, because those often override marketing context completely.
Engagement decay, because gradual cooling is one of the strongest predictors of fatigue, overload, or relevance failure.
This is what it means to build an email strategy around signals.
Signals are only useful if they change what you send.
If you collect a mountain of behavioural data and still send the same email to everyone, you’re not doing strategy. You’re doing admin with better reporting.
If personalisation is context, then exclusion is one of the most personal things you can do. Because recognising that it’s the wrong time is a form of respect.
I recently wrote a blog about segmentation in 2026, and the core argument is that modern segmentation is about segmenting people out, not endlessly slicing audiences thinner. That blog is worth reading alongside this one, because intent without exclusions becomes dangerous quickly.
A few months ago, I had a nightmare scenario with British Gas. We had an issue with our meters during a switch, and the credit didn’t transfer across properly. The end result was that we lost electricity and heating, even though the reality wasn’t that we had “run out of credit” in the normal sense.
But their system saw one data point: the meter hit zero. So I was dropped into a marketing flow explaining why running out of credit is risky, why I should switch to direct debit, and why I should do X and Y to avoid it happening again.
The signal was technically correct. The logic was not!!
They triggered an automation off a single intent signal without checking negative context signals that should have excluded me. They didn’t connect the dots between operational reality and marketing messaging. They didn’t account for the fact that someone in a service issue is not in a “helpful education” moment. They’re in a frustration moment.
That is what “wrong time” looks like in the real world. And your audience experiences that kind of mismatch constantly.
You see a haircare brand on TikTok, the message hits a nerve. You’ve struggled to grow your hair for ages (welcome to my world), you’ve tried things before, you’ve had that familiar cycle of hope, spend, disappointment. But something about this feels interesting, so you click through.
You land on site, you see the prices, and you hesitate. A pop-up offers you a discount code, you sign up because, in that moment, you’re curious enough to take the code and keep the door open.
You think about the money, you think about all the other products you’ve tried ,you tell yourself you should probably just stick with what you’ve already bought. You get distracted. You close the tab. You move on with your life.
Now, from the brand’s perspective, what just happened?
You signalled intent, you signalled interest, you also signalled hesitation.
You did not signal readiness.
But what do most brands do next?
They send 20% off.
They send “Still thinking?”
They send urgency.
They send countdowns.
They send the same offer again.
There is no objection handling.
There is no acknowledgement of what is most likely true: the person isn’t convinced it will work, the price feels risky, and trust hasn’t been earned yet. The right move here is not more discounting. The right move is reassurance, proof, and confidence-building.
This is exactly where TFDS becomes an intent-to-messaging bridge.
If you’ve read my blog on designing email journeys using TFDS, you already know I’m obsessed with it for a reason.
TFDS is the simplest way I know to stop designing journeys from marketer mode and start designing from human mode.
When you take the haircare example and run TFDS, you stop guessing what to send.
You start designing around reality.
They might be thinking: will this work for my hair type, what if it damages my hair, what if I waste money again, how do I know this isn’t just another marketing promise? They might be feeling sceptical, cautious, mildly hopeful, but also tired and disappointed because the problem hasn’t been solved for them yet. They might be reading reviews, comparing brands, searching on Reddit, asking a friend. They might be saying, “I’ll come back to this,” or “I’m not sure.”
Now look at the gap.
If that’s what’s happening internally, why are you sending a discount timer?
TFDS helps you design the correct next email, which is often something like: reassurance, proof, explanation, safety, and a gentle path forward.
It doesn’t need to be emotional manipulation. It needs to be emotionally aware.
B2B teams do this too. They just dress it up differently!
Someone downloads three technical blogs, they attend one webinar, they visit pricing pages once. Then they go quiet.
What does that signal?
Often, it signals internal discussion, stakeholder alignment, budget hesitation, or evaluation. It can also signal that they’re early in their journey and only came for education, not buying.
What do many B2B companies do next?
They escalate immediately.
They push “Book a demo.”
They send generic nurture with increasingly desperate persuasion
Again, wrong time.
The more personal response here is to support the internal journey: provide comparison guides, stakeholder-friendly decks, objection-handling case studies, ROI breakdowns, and clarity about what happens if they take the next step.
Intent-led messaging changes the tone. It reduces pressure and increases confidence.
That is personalisation.
If you want to take this out of theory and into your work, you need to stop thinking in terms of “personalised emails” and start thinking in terms of “signal-led responses.”
The build process is simple. Not easy, but simple.
First, you need your intent buckets. Active, passive warming, and negative. You already have them. Your job is to list the signals you can track and classify them honestly.
Then you need to define what each signal means. That means taking each signal and asking: what stage are they likely in, what objection exists, what are they probably thinking or feeling, and what should we not send right now? That last part matters more than most teams realise, because exclusions are what stop intent systems turning into chaos.
Next, you map each signal to a response:
If someone visits pricing three times, what is the email meant to do?
Is it meant to invite a call?
Is it meant to answer objections?
Is it meant to provide proof?
Is it meant to clarify implementation?
You decide the job first, then the copy. Not the other way around.
Then you define what success looks like and how you’ll track it. Not just clicks, but meaningful actions: a booked call, a purchase, a return visit, a reply, a trial activation, a reduction in drop-off. If you don’t define success properly, you’ll default back to open rate theatre, and you’ll optimise the wrong thing.
Finally, you add suppression rules. You cannot do intent without exclusions. It is too risky. It creates overlap. It creates collision. It creates the exact “too many emails” problem that your customers are already drowning in.
Suppression rules need to consider the entire ecosystem, not just marketing. If someone has an open support ticket, marketing should shut up. If sales is mid-conversation, marketing needs to avoid mixed signals. If someone is in onboarding, they should not be treated like a cold lead.
This is where intent-led strategy becomes a maturity marker.
I’ll say it.
Putting someone’s first name in an email does not make it personal. People do not feel “seen” because you inserted a token. They feel seen when the email reflects their reality.
The thing that makes an email feel personal is recognition, not personalisation.
It’s the moment someone thinks: “That is exactly what I was wondering.”
Or: “That landed at the right time.”
Or: “That answered the thing I couldn’t articulate.”
This is why intent + prediction works so well. It reduces the gap between what someone is experiencing and what you’re saying.
And when you build this properly, it changes outcomes.
I’ve seen this in B2B when we deployed an intent-based email trigger journey tied to both positive signals and negative context signals. A client’s pickup rate on inbound lead follow-up calls moved from 2% to 17%. Not because we spammed harder, but because the messaging and timing aligned with what was actually going on for the lead. They were warmer, clearer, and more confident when the outreach happened.
In B2C, I’ve seen it when welcome flows stop assuming that every new subscriber is excited and ready to buy. When we rebuilt a welcome flow around what people were actually signalling — including hesitation and objection states — the content changed completely. It became more educational, more reassuring, more proof-led, and far less “here’s your discount, buy now.” That shift drives revenue because it builds trust, not because it shouts louder.
This is the point.
Intent-led strategy isn’t about sending more messages. It’s about sending fewer wrong ones.
If you only do one thing after reading this, do this:
Pick one journey you already have — welcome, abandoned basket, lead magnet nurture, post-purchase, whatever.
Write down the top three signals that put someone into that journey.
Then ask yourself one brutally honest question:
What are they signalling that could mean this is the wrong message, or the wrong time?
If you can’t answer that, you don’t have personalisation. You have automation.
And if you can answer it, you’re already on your way to building the kind of email strategy that actually feels personal in 2026.
Because personal is not cosmetic, personal is contextual.