For years, communications providers have largely sold predictable products in predictable ways. Customers bought seats, licences, minutes and services, often through contracts designed to make costs easy to forecast.

AI is starting to challenge that model.

As AI becomes embedded across UCaaS, CCaaS and business communications, providers are increasingly looking at consumption-based pricing. The attraction is obvious: customers pay for what they use, can experiment more easily, and can scale usage up or down.

But there is a problem.

Consumption-based pricing changes much more than the price of a service. It changes the relationship between the provider and the customer.

And that, as discussed by Patrick and Finbarr in our latest Cloud Conversations podcast, raises a bigger question for the communications industry: are providers ready to move from selling technology to being accountable for how well that technology performs?

The problem with simply charging for usage

Consumption-based pricing sounds straightforward until you consider what is actually being consumed.

An AI service can sit on top of several layers of technology. There may be an underlying large language model, compute infrastructure, a communications platform and then the specific service being delivered to the customer.

Different services may therefore lend themselves to different pricing approaches.

Some could work on a straightforward pay-as-you-go basis. Others may use tokens or another measure of consumption.

The important point is that there is unlikely to be one universal model.

For communications providers, this creates a significant commercial and operational challenge. The industry has spent years building systems and processes around relatively predictable products. Moving to consumption requires those systems to cope with usage that can fluctuate significantly.

That is not just a pricing problem. It is an OSS and BSS problem, a billing problem and, ultimately, a customer-management problem.

Customers have reasons to be cautious

The flexibility of consumption pricing is attractive, particularly when customers are experimenting with AI.

→ Register interest for our upcoming AI research portfolio.

A business might want to test an AI capability with a small group of users, understand how it performs and then expand it if the results are positive. That is much harder to do when the customer has to commit to a large licence agreement before it knows whether the technology will work.

But the same flexibility creates uncertainty.

Cavell research discussed in the podcast points to a lukewarm and sometimes confused response from customers to consumption-based pricing. Around half of customers were described as having a mixed reception, while approximately 25% were actively avoiding consumption-based models. A further 25% were positive, or positive provided strict transparency rules were in place.

The message is clear: flexibility does not automatically equal confidence.

For a customer, a successful AI proof of concept creates a new question. If the technology works across 100 calls, what happens when it is handling 10,000?

The customer needs to know what that success will cost before rolling it out across the organisation.

AI is also disrupting the internal budget

There is another challenge that is easy to overlook.

Traditional communications spending is often relatively easy to allocate. A certain number of users, licences or minutes can sit within a particular department or budget.

AI does not necessarily work that way.

An analytics platform could analyse customer conversations and generate insights that are useful to a strategy director, product team or customer experience function, while still relying on communications infrastructure underneath.

Who pays for that?

The answer may not be obvious.

As AI services spread across organisations, businesses may have to rethink how they allocate technology spending internally. A service that starts life as an IT or communications purchase can quickly become a business-wide capability.

That makes consumption more complicated than simply introducing another line item on an invoice.

The communications provider becomes part of the optimisation process

This is where consumption-based pricing could fundamentally change the provider relationship.

With a traditional licence, the provider can sell the technology and the customer can decide how much value to extract from it.

Consumption creates a stronger incentive for both sides to pay attention to utilisation.

If customers are not using the service, they can reduce their consumption. If the service is not performing well, they have a reason to question whether they should continue using it.

For providers, that means adoption and optimisation become commercially important.

A successful AI deployment may require ongoing configuration, tuning, data management, change management and performance monitoring. The provider is no longer simply supplying a piece of technology. It is helping the customer make sure that technology continues to deliver value.

That is potentially a much stronger customer relationship.

But it also requires a very different operating model.

The margin question gets harder

There is another reason AI pricing is different from previous shifts in communications pricing: customers can increasingly see the underlying cost of the technology.

Many AI providers publish their pricing.

That creates a straightforward challenge for a communications provider.

If the underlying AI capability costs a customer £500, but the provider is charging £1,500 for the complete service, the provider needs to explain the additional £1,000.

The answer cannot simply be margin.

The value has to come from somewhere else.

It could be integration, expertise, data preparation, knowledge management, configuration, ongoing optimisation or the ability to make the technology work within the customer’s existing environment.

This is where communications providers have an opportunity.

Their advantage may not be owning the underlying AI model. It may be their understanding of the customer’s communications environment and their ability to turn rapidly changing AI capabilities into something that actually works for the business.

Trust becomes part of the commercial proposition

That also makes trust increasingly important.

Businesses are already navigating an enormous number of AI vendors and technologies. Communications providers have an established relationship with many of those businesses and can potentially act as a trusted layer between customers and the rapidly changing AI market.

But that position comes with expectations.

Customers will want pricing to be transparent. They will want to understand what they are paying for. And they will increasingly expect providers to demonstrate that the technology is delivering value.

This could make the provider’s role less about reselling AI and more about taking responsibility for its successful deployment.

Is consumption actually the end point?

This leads to a more interesting possibility.

Perhaps consumption-based pricing is not the final evolution of AI pricing.

Perhaps it is simply the step that takes the industry towards outcome-based pricing.

The difference is subtle but important.

Consumption asks:

How much did you use?

Outcome-based pricing asks:

Did it work?

Consider an AI chatbot designed to answer a specific type of customer enquiry. Under a consumption model, the customer could pay for every interaction.

Under an outcome-based model, the discussion becomes whether the chatbot successfully handled the enquiry.

That creates a much stronger incentive for the provider to make the technology work.

It also introduces complexity. Providers and customers need to agree on what constitutes a successful outcome. Not every interaction will be resolved in exactly the same way, and a successful outcome may sometimes mean recognising that an AI agent cannot resolve an issue and escalating it to a human.

That makes outcome-based pricing particularly suited to use cases where success can be clearly defined and measured.

The industry may move faster towards outcomes than consumption

There is an irony here.

Consumption-based pricing is often presented as the more flexible alternative to traditional licensing. But for customers still uncertain about AI, paying for usage may not address their biggest concern.

They do not necessarily want to know that they can use an AI service for £X per minute.

They want to know that the service will deliver the result they need.

That could ultimately push the market towards outcome-based models faster than expected.

For providers, this represents a significant shift in accountability. They would have to understand the customer’s objectives, define success, measure performance and take some of the risk when the technology does not deliver.

For customers, however, the proposition is compelling: you are not simply buying AI. You are buying a result.

The communications industry has already experienced major changes in how technology is sold and consumed. AI could take that transformation one step further.

The next competitive advantage may not be having the best AI model or the lowest consumption price.

It may be being the provider that can confidently say:

If it works, you pay. If it doesn’t, you don’t.