The average customer exists only because businesses cannot afford to understand every individual.

To compensate, we create proxies, such as bronze, silver and gold tiers of service. Management decides what will be in each package. There is a degree of flexibility, but nothing like what is now possible.

Our goal is to move on from asking what should we do for customers. We want to answer what should we do for this customer today.

The best sales advice is to listen to the customer and try and give them what they want.

This is easier with large, long-term enterprise sales. You expect to negotiate length of contract, payment terms and service level agreements. You might hold back free training as a sweetener to get the deal to close.

It’s trickier when people want changes to the product. At OTAS, we made the rookie error of responding to our most engaged client. We added on every feature they requested. We ended up with a product that was too complicated to explain to almost everyone else.

Stripping back the core offering made selling it a great deal easier.

It was far better to be able to provide each customer with the product features they wanted without exposing them to every feature any client had requested.

AI makes that possible for a greater range of businesses.

At OTAS we used a small accountancy firm with a few hundred SME clients. We were at the larger end of the list and got attention from a partner when we needed it. The tier below us received regular advisory work. The rest took management accounts, regulatory reporting and an annual conversation.

The small number of accountants lacked the time to look into several hundred clients.

Yet the firm has the information it needs to manage them in a unique way. It has the data, the deadlines and the discussions. It needs a way of reviewing them every morning to find something worth pursuing.

One client may have rising debtor days. Another has rising overtime on two major contracts. A third owes a slug of corporation tax the week before a large customer is due to settle its invoice.

The first client gets a conversation about cashflow in place of its monthly accounts meeting. The second receives advice on pricing. The third has a talk about tax planning.

The accountant stops delivering the same service to every client. It now offers something much closer to a part-time finance director than an accounting package.

The application of AI is not about making it cheaper to run monthly accounts, although it might. It is about changing what the client receives.

Personalisation has meant changing the message, recommendation or offer a customer receives. AI promises something more substantial. We may now finetune the service around the behaviour and circumstances of individual customers.

Commonwealth Bank of Australia wanted to move away from treating millions of customers as segments. It built a Customer Engagement Engine that evaluates individual information and determines the next best conversation.

Pega reports that CBA was making these decisions around 20 million times a day using 200 machine-learning models and 157 billion data points. It’s been running since 2015, long before today’s AI hype. New layers of generative AI are being added, but the aim of the end of the average customer is older than the technology that refines it.

Microsoft explains how Discovery Bank used generative AI to provide clients with “a private banker in their pocket”. Clients may ask questions, get personalised recommendations and set regular budget reminders.

The business model revolves around a behavioural fingerprint for every customer. Discovery claims the personalised service is 80% more effective and produces up to a fivefold increase in customers taking positive financial actions. Adding a generative AI layer reportedly doubled engagement with the recommended next action.

Businesses spent decades standardising services because doing so improved margins. At the same time, salespeople tried to represent this as a tailored service, by tweaking the contract. There was an obvious tension because every request for a new feature cost considerable time and money.

AI now promises enough personalisation to make a material difference to what customers receive and raise margins at the same time.

This is what we mean when we talk about reorganising how a business works using AI. It’s the end of the average customer as a workaround.