Can Automation Make Emails More Personal ?

Most businesses already know email personalisation matters. Far fewer have actually moved past inserting a first name into a subject line, because doing it properly used to require a data team most companies don’t have. AI has changed that equation — not by replacing email marketing, but by making the kind of individual-level personalisation that once needed a dedicated analytics function available through tools a small marketing team can run directly.

Why This Actually Moves Conversions, Not Just Opens

Personalisation gets talked about mostly in terms of open rates, but the more meaningful shift is further down the funnel. An email that recommends a product based on what someone actually browsed, arrives at a time they’re genuinely likely to check their inbox, and uses phrasing that’s been tested against what that type of recipient responds to, doesn’t just get opened more often — it converts more often, because every part of it is closer to what that specific person needed at that moment. That compounding effect across open rate, click rate, and conversion rate is where most of the real return sits, and it’s also the part that’s hardest to achieve with manual segmentation alone.

What AI Models Are Actually Doing Here

It helps to separate two different jobs AI is doing in modern email marketing, because they require different tools and different setups.

Predictive Models: Deciding Who Gets What, and When

The first job is prediction — using a customer’s past behaviour to estimate what they’re likely to want next and when they’re likely to engage. This is the same underlying technique that powers product recommendations on major retail sites, applied to email instead. It needs a reasonable amount of historical data to work well: purchase history, browsing behaviour, past email engagement. A business with a thin customer database won’t get much value from this piece yet, and that’s worth being honest about before investing heavily in it — the payoff comes once there’s enough behavioural history for a model to actually learn patterns from.

Generative Models: Writing and Testing the Content Itself

The second job is generation — using language models to write and test subject lines, preview text, and body copy at a scale a human team can’t match. Rather than running one A/B test between two subject lines, a generative system can produce a dozen variations styled for different tones or audience segments and learn which perform best, often within a single send. This is the part of AI email marketing that’s developed fastest recently, and it’s also the part most likely to produce generic-sounding output if it isn’t paired with a clear brand voice guiding what “good” looks like for that business specifically.

Platforms Worth Knowing About

The practical starting point for most businesses is whichever email platform they already use, since most mainstream tools have built AI features directly into their existing product rather than requiring a separate system. Klaviyo, popular with eCommerce brands, has predictive analytics and AI-assisted content generation built into its core platform and integrates tightly with product catalogues, which makes its recommendations unusually accurate for online stores. Mailchimp’s AI features cover subject line generation and send-time optimisation and tend to suit smaller teams wanting something usable without much setup. HubSpot’s AI tools lean more toward the content and workflow side, useful for businesses already running HubSpot for their broader marketing and sales activity. Salesforce Marketing Cloud’s AI capabilities are considerably more powerful but are built for larger organisations with the data volume and technical resource to make full use of them. For businesses wanting AI-generated and tested copy specifically, standalone tools like Persado specialise in that one job and can sit alongside whichever platform is already sending the emails.

Choosing Between Them

The right choice depends far more on existing customer data and team size than on which platform has the most features on paper. A small business with a modest but growing email list is usually better served starting with the AI features already built into a mainstream platform like Klaviyo or Mailchimp, rather than adding a specialised tool before there’s enough data volume to make it worthwhile.

How to Actually Start Using This

The businesses that get real value from AI-driven email personalisation almost always start with the data, not the tool. Before choosing a platform, it’s worth making sure customer behaviour is actually being tracked properly — purchase history, email engagement, site activity where relevant — since every AI feature above depends entirely on having something real to learn from. From there, the sensible order is usually: get predictive send-time and basic behavioural segmentation working first, since that has the clearest and fastest payoff; layer in AI-assisted subject line and copy testing once there’s a steady sending cadence to test against; and only move toward fully dynamic, per-recipient content once the earlier layers are proven to be working. Trying to do all of it at once tends to produce a messy setup nobody on the team fully understands or trusts.

Where This Needs Care

AI personalisation in email marketing sits directly on top of customer data, which means data protection isn’t a side consideration — it’s central to doing this properly, particularly for businesses operating under GDPR. Consent, data minimisation, and being able to explain to a customer what’s driving the personalisation they’re seeing all need to be built in from the start, not added afterward. Our GDPR compliance team can help make sure a personalisation setup is sound on that front before it scales. There’s also a brand-voice risk worth watching: AI-generated copy that’s optimised purely for click-through can quietly drift toward generic, high-performing phrasing that no longer sounds like the business it’s representing. The strongest setups keep a human reviewing and shaping what the AI produces, rather than letting performance metrics alone decide the final copy.

Where This Is Heading

The direction is toward emails assembled individually at the moment of sending, rather than written once and distributed to a whole list. Dynamic content blocks that swap imagery, offers, and tone based on a real-time read of the recipient are already standard among larger retailers, and that capability is steadily becoming available to smaller businesses as the underlying platforms mature. It’s a reasonable expectation that AI handling first-line reply triage — drafting responses for a human to approve — becomes a normal part of email marketing for growing businesses, not just large enterprises with dedicated support teams. Salesforce’s State of Marketing research is a useful ongoing reference for how quickly adoption of these tools is actually moving across businesses of different sizes.

The Practical Takeaway

AI hasn’t changed what makes email marketing effective — relevance, timing, and a voice people recognise still matter as much as they always did. What it’s changed is how achievable real, individual-level personalisation is for businesses that don’t have a data science team on staff. Getting the data foundation right first, choosing a platform that matches actual team size and customer volume, and keeping a human shaping the final output tends to separate the businesses getting a genuine lift in conversions from the ones just adding AI features for their own sake. Our copywriting and content strategy and data and analytics teams work on exactly this combination for clients getting started with it.