Real-Time Personalisation in Sales with AI: Tailored Experiences

Real-time personalisation in sales means that at the moment of their enquiry, a customer gets a reply that fits them rather than a standard text. With AI, this is now feasible for small businesses too. This article shows which forms of personalisation make sense, where the line lies between helpful and uncomfortable or legally tricky, and how to start with little data.

What real-time personalisation in sales means

Personalisation comes in several levels. Everyone knows the simplest: the name in the greeting of an email. The next level uses behaviour, for example a reminder when someone has opened a quote but not accepted it. Real time means the response is adapted during the ongoing contact, in other words in the chat, on the phone or in a WhatsApp conversation.

An example from a beauty salon: a regular client sends a WhatsApp message in the evening saying she needs “the usual again”. An AI employee without context would have to ask which treatment she means. One with access to the customer history can see that she books a particular facial with the same beautician every six weeks, and immediately suggests two suitable appointments with her. For the client it feels like talking to a receptionist who knows her.

It is similar in a car repair shop. When a customer calls, the system recognises the vehicle from the phone number, knows that the statutory vehicle inspection is due soon and can mention it while booking the tyre change. Or in an online shop: the chat knows which product page the visitor is on and answers questions about exactly that item instead of reeling off a generic greeting.

Three levels at a glance

Level	Example	Data required	Risk
Situational	The chat refers to the product page being viewed or the question asked	Only the current contact	Low
Historical	The bot knows previous appointments, the vehicle or the favourite treatment	Customer history from the CRM or calendar	Medium, mind data protection
Profile-based	The system predicts purchase intent and adjusts the offer or price	Behavioural data over a longer period, tracking	High, legally and for trust

For most small and mid-sized businesses, the greatest benefit lies in the first two levels. They save customers having to answer follow-up questions, lead to quicker appointments and come across as attentive without anyone feeling watched. The third level, by contrast, requires volumes of data that many smaller businesses simply do not have, and it brings the most obligations.

Personalised prices: proceed with caution

Some articles on real-time personalisation recommend adjusting prices according to the customer profile, for example a cheaper entry price for visitors who were previously on a competitor's site. Technically that is possible. For a small business it is usually not a good idea.

For online contracts with consumers, you must disclose when a price has been personalised on the basis of automated decision-making. Then there is the question of fairness: if a regular customer finds out that her neighbour paid less for the same service, trust is quickly gone. Transparent differences that anyone can understand make more sense, such as discounts for regular customers or prices based on order size. You can read more about the possibilities and limits of dynamic pricing in the article on AI-supported price optimisation.

Data protection and trust

The more data goes into personalisation, the more the ground rules matter. Website tracking that goes beyond what is technically necessary usually requires consent via the cookie banner. The processing of customer data in the CRM must be described in your privacy notice. And under the GDPR, decisions that significantly affect customers may not be made solely by automated means unless an exception applies.

The customers' feelings matter at least as much. There is a line between “they know me” and “they know too much about me”. A bot that reminds a regular client of her usual appointment comes across as attentive. A bot that mentions how often she has looked at the pricing page over the last few days comes across as creepy. Use what you know about customers to save them effort, not to show off how much you know.

Anyone talking to an AI should also be told so. The EU AI Act requires this transparency, and a short note at the start of the conversation is enough.

How to get started with real-time personalisation

1. Bring customer data together in one place. Personalisation only works if the AI employee accesses the same data as your team. This is usually the CRM or the appointment and job management software. The basics are in the guide to AI CRM integration.
2. Start with situational personalisation. The bot should know what the customer is talking about right now and which page they came from. This needs hardly any additional data.
3. Recognise existing customers. Match the phone number or email address against the CRM and use basic data such as the last appointment or the vehicle. Check beforehand whether your privacy notice covers this.
4. Align the channels. A customer who calls in the morning and follows up via WhatsApp in the evening should not have to give the same details twice. The article on multi-channel AI sales describes how to achieve this.
5. Read along. In the first weeks, check samples of conversations regularly. If a personalisation comes across as intrusive or wrong, take it back.
6. Write down the rules. Define which details the bot may use and which it may not, for example: previous appointments and booked services yes, payment arrears or the team's notes about the person no. This list also helps when a customer asks how the bot knows something.

When it is not worth it

If most of your customers are new and have a one-off need, as with a locksmith, there is little history for personalisation to draw on. In that case situational personalisation is perfectly adequate. And if your customer data is spread across several Excel lists and notebooks, tidy up first before asking a bot to build personal replies from it.

Frequently asked questions

Do I need a customer data platform for real-time personalisation?

For small and mid-sized businesses, generally not. A well-maintained CRM or appointment software with customer history is enough for the first two levels. A dedicated data platform only pays off with a very large number of customers and channels.

Can the bot automatically recognise existing customers on the phone?

Matching the phone number against your customer data is common practice but should be described in your privacy notice. In sensitive areas such as medical practices, the bot should also check that the right person is really calling before giving out any information.

How is this different from personalised mass outreach?

Mass outreach personalises outgoing messages, such as email campaigns. Real-time personalisation concerns the response to a contact the customer started themselves. You can read more about the former under Personalising sales outreach at scale.

Can the AI sort my customers into groups?

Yes, and that is often a sensible intermediate step. Instead of looking at each customer individually, you work with a few groups that are easy to understand. The article on AI customer segmentation explains how.

Conclusion

Real-time personalisation in sales benefits small businesses most when it saves customers from follow-up questions: the right appointment with the usual staff member, the answer about the product they are looking at, the vehicle that is already known. Personalised prices and profiling, on the other hand, carry legal obligations and risks to trust that are rarely in proportion to the benefit.

The page AI assistant for customer retention shows how an AI employee recognises existing customers, sends appointment reminders and maintains contacts.