AI agents and the economics of mass personalisation

Zenith’s CRM makes tailored follow-up practical

At Zenith, we have recently used AI to build a bespoke CRM that lets us add agents at key stages of the lead process. It supports more personalised follow-up, with approval steps built into the workflow. The purpose is to make better use of what we know about someone’s interests and needs when deciding what to send them and when.

The broader significance is the change in what becomes practical. When tailoring a service requires substantial manual effort, businesses tend to reserve it for their most valuable customers or rely on broad segments. Reducing that effort can make a more individual service viable for a larger group.

Mass personalisation means adapting a service to an individual’s circumstances while serving many customers. It depends on relevant context, such as the customer’s objective, previous decisions or progress, and a practical way to use that context in the next interaction. An agent adds value when it helps the customer make a decision or complete a task more successfully. Generating more messages is a poor measure of that benefit.

Personalisation can change the service itself

Duolingo’s Video Call feature illustrates this opportunity in language learning. The company describes conversations that adapt to a learner’s ability and use information from earlier calls. This adds individual speaking practice to a digital learning product, without arranging a teacher for each conversation. It shows how AI can extend the service a customer receives.

L’Oréal Paris takes a different approach with Beauty Genius. Its assistant offers personalised beauty routines, product recommendations and virtual try-ons. Advice that helps a customer choose and use a product becomes part of the brand’s offer. The commercial question is whether that additional service improves the customer’s experience enough to influence what they buy or their choice of brand.

Both examples go beyond inserting a customer’s name into a message. They use information about the individual to change the advice, interaction or support provided. For a mid-market business, the equivalent opportunity might be helping customers specify a purchase, choose between options or get more value from something they already own.

A better service may take a business into adjacent markets

Consider the work a customer does before and after buying. They may pay another provider for advice, spend time comparing alternatives themselves, or receive little support once the transaction is complete. If a business can provide useful assistance at a viable cost, it may be able to include some of that work in its core offer or sell it as an additional service.

This can change where value is created and captured across an industry. A supplier that helps customers make a better decision may gain a closer relationship with them. An existing adviser or intermediary may face competition from a business that previously supplied only the product. That is connected to the value-chain questions raised by United Airlines’ restrictions on AI agents: who helps the customer choose, who controls access to that customer, and who earns the resulting margin?

The commercial test for an additional service

The ability to personalise a service does not establish demand for it. Start with a specific customer problem and the way people address it today. Determine whether better advice or support would change their decision to buy, increase their use of the product or justify a separate fee. Test that proposition with customers before assuming that more interaction means more value.

The economics also need to include the cost of maintaining accurate information, reviewing exceptions and correcting mistakes. Set clear limits on what an agent can decide and when a person should intervene. A useful pilot measures the customer outcome alongside the cost to serve, so management can judge whether the proposition deserves further investment.

The economics depend on additional contribution

A personalised service can earn a return through an explicit fee, additional sales, better retention or lower support costs. Management should identify which of these is expected to pay for the investment. Count the contribution from additional sales after the cost of fulfilling them, and separate it from purchases the customer would have made anyway. More usage of an assistant is evidence of activity, not by itself evidence of a commercial return.

Then deduct the continuing cost of the service. That includes access to models and software, maintaining product and customer information, human review and remediation when advice is wrong. Setup costs also need to be recovered over a realistic period. A useful test compares similar customers receiving the existing service and the proposed offer, with quality and commercial measures agreed before launch.

• Specify the customer decision the service should improve and the information needed to make the advice relevant.
• Choose a commercial measure, such as contribution per customer or cost per resolved request, alongside a measure of customer benefit.
• Record review and exception costs as part of delivery. Include a clear route to a person when the agent cannot resolve the customer’s need.
• Agree what evidence would justify expansion, further testing or withdrawal before making the service widely available.

From a new proposition to a value creation plan

In The AI Playbook for Mid-Market Leaders, the AI Strategy Assessment examines new customer propositions, industry boundaries and profit pools. The AI Value Creation Plan then identifies the capabilities and workflows required to deliver the chosen proposition. Reviewing a function’s remit includes considering useful work the business has never had the capacity to perform.

Mid-market businesses may have an advantage where they can make decisions and change working practices quickly. A focused test of a useful new service can move faster than a broad technology program. The opportunity is to use that speed to learn what customers value, build a proposition around it and earn a stronger position in the market.