AI in eCommerce: personalisation without creeping people out

Personalisation used to mean putting a first name in an email. Now it can shape everything a customer sees: products, prices, content, even the timing of each message. For eCommerce, that kind of intelligence promises higher conversion rates, bigger baskets and better retention.

It can also make people deeply uncomfortable.

Most of us have had that moment where a site seems to know a little too much. The advert that follows you across three devices. The product recommendation that reveals something you browsed once, late at night. The eerie sense that your data is being used in ways nobody ever explained.

For Irish and European businesses, there is another layer. Between GDPR, changing privacy expectations and growing scrutiny around automated decision making, it is not enough for personalisation to be effective. It has to be respectful, transparent and fair.

The good news is that these goals are not in conflict. With the right approach, AI in eCommerce can feel like a helpful shop assistant, not an overly intense salesperson who knows your entire search history.

Where personalisation starts to feel creepy

Most customers are comfortable with some level of personalisation. People appreciate being shown items in their size, language and price range. They like it when a site remembers that they prefer click and collect, or that they never want push notifications.

The unease creeps in when three things happen at once.

First, the logic is invisible. If recommendations feel opaque or over precise, customers start to wonder how you know what you know. When people cannot see a sensible reason for what you are showing them, they fill the gap with suspicion.

Second, the timing is wrong. Overly aggressive retargeting, immediate follow up messages after a brief visit, or repeated reminders about sensitive categories can all feel intrusive, especially on shared or work devices.

Third, the value is one sided. If personalisation seems to serve only the retailer, pushing high margin products or unnecessary extras, customers feel used rather than understood.

These are all design and strategy problems, not purely technical ones. They can be addressed with clear principles and thoughtful implementation.

Principles for human friendly personalisation

It helps to treat personalisation as part of your customer experience, not just a feature of your marketing stack. A few simple principles go a long way.

Start with respect for context

Not all categories are equal. Recommending more books to someone who bought a bestseller is very different from targeting products related to health, finances or family circumstances.

Be extra cautious with categories that could reveal sensitive information or create embarrassment. Build rules that limit how long certain products are used for recommendations and how prominently they appear on shared channels such as social ads.

Make the logic feel understandable

You do not need to expose your full algorithm, but you can give understandable cues. Labels like “Because you bought X” or “Popular with customers who chose Y” help users see why they are seeing a particular product.

If customers can mentally follow the link between their behaviour and your suggestion, they are far less likely to find it unsettling.

Match personalisation to the relationship

A first time visitor should not receive the same level of personalisation as a long term customer who has an account and a purchase history. Start gently and increase sophistication as trust builds.

This might mean beginning with broad category suggestions and only introducing more specific recommendations once someone has created an account, opted into marketing or returned several times.

Give people simple controls

Nothing reassures users more than a clear way to opt out or dial things down. This could include the ability to hide certain recommendations, clear a viewing history, mute specific product types or adjust email frequency.

These small control points send a powerful signal that personalisation is being done with the customer, not to them.

Practical ways to use AI in eCommerce without overstepping

With those principles in mind, there are several areas where intelligent personalisation can add real value to both customers and businesses.

Smarter product recommendations

Product recommendations are often the first use case for AI in eCommerce. Instead of static “related products”, more advanced systems can take into account browsing patterns, basket combinations and popularity trends.

To keep this helpful rather than intrusive, focus on relevance and diversity. Offer a mix of complementary items, alternatives at different price points and genuinely useful add ons, rather than simply pushing the most expensive option.

Testing is important here. Use A/B experiments to see which recommendation approaches actually improve conversion and customer satisfaction, not just revenue per session. Combine quantitative data with feedback from support teams, who often hear first when something feels “off” to customers.

Search that understands intent

On site search is a natural place to apply more intelligent approaches. Instead of simple keyword matching, you can interpret intent, correct spelling, recognise synonyms and surface richer results.

For customers, this feels like a smoother, less frustrating shopping experience. They find what they need faster, which in turn boosts conversion and reduces strain on customer service.

To avoid the “black box” feeling, show clear search suggestions and allow customers to refine or undo filters easily. Make it obvious when results have been narrowed by category, price or attributes.

Email and lifecycle journeys that feel timely, not stalkerish

Behavioural triggers, such as abandoned basket emails or follow ups after a download, are powerful tools. They can also become a source of irritation if overused.

Set sensible limits on frequency and duration. One or two reminders over a few days is usually enough. Combine behavioural signals with lifecycle data so that loyal customers are treated differently from casual browsers.

Always make sure there is a clear benefit in each message. Exclusive content, helpful guides, genuine offers or reminders about real value carry more goodwill than constant “Last chance” prompts.

Service and support that stay human

Intelligent tools can help triage customer queries, suggest answers and free up human agents to handle complex cases. In eCommerce, this often appears as chat widgets, help centres and automated replies.

Be honest about what automated tools can and cannot do. Use friendly cues to show when customers are interacting with a scripted flow and provide an easy route to a human when needed. Personalisation in support should prioritise clarity and speed over pushing products.

Governance, consent and long term trust

Behind every personalisation feature sits a set of choices about data. For European organisations, those choices are shaped by strict privacy laws and growing expectations around transparency.

That means you need more than clever technology. You need governance.

Treat consent as a conversation, not a one time banner. Explain, in simple language, how you use customer data to improve their experience. Provide accessible privacy settings, not just legal documents.

Align your experiments with your organisational values. If a particular tactic feels effective but uncomfortable, listen to that discomfort. Long term trust is harder to rebuild than short term revenue.

Involve legal, security, design and marketing teams in decisions about new personalisation features. A cross functional view helps you spot risks early and design safeguards into your systems.

Handled with care, intelligent personalisation in eCommerce becomes a way to respect customers’ time and attention, not a way to extract as much value as possible at any cost. It shows that you understand both the power of data and the responsibility that comes with using it.

For organisations that want to grow online without losing their values, that balance is where the real competitive advantage lies.

Get in touch