Personalisation has been part of email marketing for a long time. But for much of that time, it hasn’t been particularly personal.

Adding someone’s first name to a subject line or opening an email with “Hi Sarah” technically qualifies as personalisation. However, it doesn’t demonstrate much understanding of who Sarah is, what she wants or why she might be interested in the email.

AI is giving marketers the ability to go significantly further. Instead of simply inserting information about a recipient into otherwise generic content, increasingly sophisticated email platforms can use behavioural and customer data to influence what someone receives.

That distinction is important because not everything marketed as AI personalisation delivers the same value.

The most useful applications tend to have one thing in common: they personalise according to something meaningful about the customer and their behaviour.

Moving beyond surface-level personalisation

Using someone’s name can make an email feel more conversational and customer information such as their location, company or job title can sometimes make content more relevant. But inserting a known data point into a message is not the same as tailoring the experience.

Consider two subscribers called Sarah.

One visited a website yesterday, looked at the same product three times and then left without buying it. The other purchased that product six months ago and hasn’t visited the site since.

A line such as “Hi Sarah, take a look at our latest products” technically personalises the email for both people. In practice, it ignores almost everything that differentiates them.

This is where some of the hype surrounding AI personalisation can become distracting. Generating hundreds of slightly different versions of essentially the same email isn’t necessarily useful simply because AI makes it possible.

The better question is whether that personalisation makes the communication more relevant.

Behavioural personalisation is where AI becomes more useful

Behaviour gives marketers evidence of intent. Products viewed, purchases made, baskets abandoned, links clicked and frequency of engagement can all tell you considerably more about what somebody might want than their first name ever could.

This kind of personalisation is where AI becomes much more useful. Many email marketing platforms can now use behaviour and engagement data to decide which products, offers or content are most relevant to an individual. That means recommendations can be shaped by what someone has actually viewed, bought or interacted with, rather than relying only on static profile information.

This creates a fundamentally different kind of personalisation. Instead of saying, “Hello Sarah”, the system is effectively saying, “Based on what you’ve actually done, we think this is what you’ll be interested in.”

Certain platforms can also use event data within email flows. That means personalisation can respond to a particular action, such as a product being ordered or left in a basket, rather than relying exclusively on static information stored against a customer profile.

That’s much closer to what marketers should be aiming for: relevance based on signals rather than decoration.

The limitations of behavioural personalisation

There is an important catch. Behaviour tells you what somebody did, but it doesn’t always tell you why.

A customer might repeatedly browse an expensive sofa because they’re considering buying it. Or they might be an interior designer researching a project. They could already own it. They could even dislike it and simply keep returning to show somebody else.

AI models have access to signals, not a complete understanding of the person producing them.

There is still an important limitation, though. Most of these systems are strongest at interpreting customer, catalogue and behavioural data. Marketers can usually apply rules around factors such as product category, price, availability, customer status or previous purchases, but the technology is still working primarily from the signals it has been given. While that’s valuable, it isn’t the same as understanding a brand.

You might know that a particular product shouldn’t be promoted heavily because it conflicts with an upcoming campaign, doesn’t represent where your brand wants to position itself or simply isn’t something the business wants to prioritise. Behavioural data alone can’t be expected to understand that wider context.

The lesson is that marketers still need to provide the strategic guardrails around behavioural personalisation.

CRM data can add another layer of context

For B2B organisations in particular, some of the most valuable personalisation data may already be sitting in the CRM.

Platforms like HubSpot allow you to use CRM properties and activity data to personalise email communications, going well beyond a contact’s name. That can include lifecycle stage and contact or company information such as role, industry, company size and customer status. Behavioural signals such as form submissions, website activity and email engagement can also feed into targeting and workflows.

This matters because the appropriate message often depends on context. A prospect who has just downloaded their first piece of content probably shouldn’t receive the same email as an opportunity already speaking to sales. Similarly, the priorities of a marketing manager at a small business could be very different from those of a CMO at an enterprise organisation.

CRM-driven personalisation gives marketers the opportunity to account for those differences. Again, AI is most useful here when it’s helping marketers interpret and act on meaningful information, rather than simply producing more variations of copy.

So, what’s genuinely useful?

A useful test is to ask whether the personalisation would change something meaningful if the recipient were different.

Changing “Hi Sarah” to “Hi James” doesn’t substantially change the email. Recommending different products because James has bought from you five times while Sarah has never made a purchase, does.

The strongest personalisation therefore tends to rely on several types of information. That is: who someone is, where they are in the customer journey and, crucially, what they’ve actually done.

Better data matters more 

There’s a temptation to treat AI as the ingredient that suddenly makes personalisation intelligent. In reality, its usefulness is heavily dependent on the information underneath it.

If customer records are incomplete, behavioural tracking is unreliable or lifecycle stages aren’t maintained properly, adding AI won’t magically create a detailed understanding of the audience. It may simply automate decisions based on poor information.

Marketers therefore need to think about AI personalisation as a data challenge as much as a technology one. What customer information are you collecting? Is it accurate? What behaviours are meaningful? And which signals actually justify changing the message somebody receives?.

Effective personalisation is about combining the right technology and data with a clear understanding of your audience, your brand and what you’re trying to achieve.

At Axonn, we help businesses build digital marketing strategies that use technology with purpose. Get in touch with our team to find out how we can help you make your email marketing work harder.