Personalisation has become one of those things in email marketing that we almost automatically assume is good. More data. More relevance. More individualisation. More sophisticated experiences.
And now AI is making it possible to personalise at a level and scale that would have been prohibitively expensive just a few years ago.
But there’s a question I don’t think we ask often enough: When is more personalisation actually worth it?
I started thinking about this recently during a free consultation with a US-based founder. He was planning an outbound email programme and wanted his emails to feel genuinely relevant to each recipient, rather than looking like another mass-produced sales email.
We started talking about the information he could use to personalise them. Some of it was easy. He could know who the person was, where they were based, what they did and other information readily available within his prospecting platform. But none of that necessarily made the email more valuable. So we started exploring what he could learn by looking at each prospect’s actual behaviour.
His product helps real estate agents create and edit social video, so what if he knew how many videos each agent was publishing every month? What if he could identify the type of videos they were producing? Perhaps even estimate how much time they were spending creating them?
Now we weren’t just talking about inserting personal data into an email. We were talking about using information to make the message more useful to the recipient.
But there was a catch. That information was harder to obtain.
And suddenly we had arrived at a question that applies far beyond cold outreach: Personalisation has a cost. So how much is it worth?
Personal doesn’t necessarily mean valuable
We’ve spent years talking about first-party data, zero-party data, behavioural data and increasingly sophisticated personalisation. But simply knowing something about somebody doesn’t mean using it creates value.
I see this myself in cold emails all the time. Someone will tell me they’ve noticed I’m the author of Holistic Email Marketing, mention an article I’ve written or refer to something else they’ve clearly pulled from my public profile.
That’s personal information. It may even be highly specific. But it often has absolutely no value to me.
When the connection between the personal detail and the reason for contacting me is tenuous, it can have the opposite effect. Rather than making the message feel thoughtful, it makes the machinery behind the personalisation more visible.
I know you’ve scraped the information.
The fact that personalisation is possible doesn’t mean it’s useful.
So the question shouldn’t simply be: What can we personalise?
It should be: What can we personalise that creates meaningful value for the customer?
And then we need to ask: What does it cost us to create that value?
Not all personalisation costs the same
Think about personalisation in three broad categories.
1. Available personalisation
This is information you already have and can easily activate: a customer’s name, location, products purchased, loyalty status, categories browsed or stated preferences.
Once the data and integrations are in place, using these fields can be relatively inexpensive and highly scalable. But cheap doesn’t automatically mean valuable.
Putting someone’s first name in a subject line may cost virtually nothing. That doesn’t mean it materially improves their experience. Likewise, telling me I bought a pair of shoes six months ago isn’t useful simply because the information is accurate.
The value comes from what you do with the information, not from the fact that you possess it.
2. Derived personalisation
This is where things become much more interesting. Instead of simply retrieving a piece of data, we derive an insight from it.
Perhaps we use several behavioural signals to determine what a customer is most likely to need next. We might analyse previous purchases to identify replenishment patterns, or combine browsing, purchasing and engagement behaviour to determine that one customer needs reassurance while another is ready for a stronger reason to act.
This isn’t just: We know X about you.
It’s: Because we know X, Y and Z, we can make this interaction more useful to you.
Historically, this kind of analysis could be expensive, requiring analysts, complicated rules, additional technology or manual intervention.
AI has the potential to change those economics dramatically. It can classify, summarise, identify patterns and derive information from unstructured data at a scale humans simply couldn’t manage.
3. Investigative personalisation
This is personalisation that requires additional work specifically for an individual customer or prospect.
In my conversation with the founder, this could have meant someone visiting an estate agent’s social profiles, reviewing their videos, identifying the types they created and assessing how frequently they posted.
That could create an incredibly relevant email. But imagine doing it 10,000 times. Or 100,000 times. Suddenly the economics matter enormously.
The same principle applies in retention marketing. We can create increasingly sophisticated individual experiences by bringing together more data, analysing more behaviour and applying more intelligence. But every additional layer can introduce costs in technology, data, processing, people, integrations, creative complexity and ongoing management.
So we need to stop assuming that the most sophisticated form of personalisation is automatically the best one.
The most personalised experience doesn’t automatically win
Imagine two versions of an email.
Version A uses data that can be gathered and processed automatically at a cost of pennies per recipient. Version B includes an additional insight that requires significantly more processing or perhaps human intervention.
Version B performs better.
Success? Not necessarily. The important question is how much better.
If the additional personalisation increases conversion by 2% but costs more to create than the incremental revenue it generates, we’ve created better-performing email and worse-performing marketing.
That’s the part of the personalisation conversation I think we’ve been missing.
We measure opens, clicks, conversions and perhaps revenue per email. But how often do we measure the incremental cost of relevance?
A Personalisation Value Equation
I’m not suggesting marketers need another complicated attribution formula. But I do think we need a different way of thinking about personalisation.
On one side, we have the value created:
Customer value × performance uplift × scalability
On the other, we have the cost:
Data cost × production effort × complexity
The goal isn’t maximum personalisation. It’s finding the point at which additional relevance continues to justify the additional investment.
That point will differ by business, customer and use case. A high-value B2B sale might justify considerable research and individualisation. A retailer sending millions of promotional emails may need the incremental cost per recipient to be fractions of a penny. A lifecycle programme may justify more upfront investment because the capability can continue generating value for years.
So rather than asking, “How much should we personalise?”, the better question is: Does this personalisation earn its keep?
Test the value, not just the execution
This is where testing becomes particularly important.
In that conversation with the founder, we’d identified several pieces of information that could make his outreach more relevant. Some were relatively easy to gather automatically. Another was much harder and might require manual research.
So rather than assuming more personalisation must be better, why not test it?
One group could receive the email using the two easily scalable data points. Another could receive the richer version containing the additional, more expensive insight.
Then measure the incremental difference.
If the more labour-intensive version produces enough additional value to justify the work, excellent. If it doesn’t, you’ve learnt something equally useful: the extra personalisation isn’t worth buying.
That’s very different from the way many personalisation programmes are developed. Too often, sophistication itself becomes the goal.
Can we build it? Can the platform do it? Can we connect this data source? Can AI generate 10,000 different versions?
Those are implementation questions. They aren’t strategy questions.
Strategy asks: What are we trying to improve, why should this particular information improve it and how will we know whether the improvement was worth the investment?
AI is changing the cost of relevance
AI’s biggest contribution to personalisation may not be its ability to write thousands of different emails. It may be its ability to make previously expensive insight cheap enough to use at scale.
Let’s go back to the founder analysing real estate agents’ videos. He had manually reviewed a sample and identified patterns. Some were simple one-take videos. Others involved multiple cuts. Some clearly required much more production effort.
That classification was the difficult part. He had done the thinking.
The next question was whether AI could reliably apply that framework across thousands of prospects.
The human creates the framework. AI scales the analysis. And the human validates whether the output is reliable enough to use.
That’s very different from asking AI: “Personalise this email for 10,000 people.”
If we don’t know what information is valuable, AI simply enables us to create irrelevant personalisation more efficiently.
Do the hard thinking before you automate
Before building a sophisticated personalisation programme, marketers should ask:
- What information would genuinely help the customer make a better or easier decision?
- Do we already have that information?
- If not, can we derive it reliably?
- What will obtaining or deriving it cost?
- Can AI reduce that cost?
- Why do we believe this information will improve performance?
- How will we test whether the additional value justifies the investment?
Notice where AI appears in that list. It’s not at the beginning.
The first job is still understanding the customer. That’s the work marketers can’t outsource.
AI can analyse enormous datasets, classify customers, identify patterns, generate content and potentially create an individualised version of an email for every person in your database.
But capability isn’t strategy. Just because we can create 100,000 unique emails doesn’t mean we should.
Meaningful personalisation is about the customer, not the data
I often define good personalisation in terms of relevance and value. The customer shouldn’t be impressed that you know something about them. They should benefit because you know it.
Knowing my birthday is data. Using it to send me a generic “Happy Birthday!” email may be pleasant, but its value is limited.
Knowing what I buy, when I buy it and what I may need next — and then using that understanding to make my life easier — is different.
And sometimes the most valuable personalisation may barely look like personalisation at all. It might mean removing irrelevant information, changing the order in which products are presented, providing additional reassurance, reducing choice, reminding a customer when something is likely to become useful again or recognising that the best email to send today is no email at all.
None of those require shouting: LOOK HOW MUCH DATA WE HAVE ABOUT YOU.
They require using what we know intelligently.
The future isn’t maximum personalisation
For years, the direction of travel in email marketing has seemed obvious. More data leads to more personalisation, which leads to greater relevance, which leads to better results.
But the reality is more nuanced.
More data can create more relevance. More relevance can create more value.
Neither is guaranteed. And both come at a cost.
AI is changing that equation by reducing the cost of analysing information and creating individual experiences. That makes this an incredibly exciting time for personalisation. It also makes strategic discipline more important.
Because when something becomes easier to do, we tend to do more of it. The temptation will be to personalise everything simply because we can.
I think the smarter opportunity is different.
Use AI to reduce the cost of creating relevance.
Use behavioural science to understand what relevance actually means.
Use testing to determine whether additional sophistication creates sufficient incremental value.
The future of personalisation isn’t maximum personalisation. It’s economically intelligent personalisation.
And that means asking a much better question than, “Can we personalise this?”
Is this personalisation worth it?
