Value-based contracts: A Computer Weekly Downtime Upload podcast

IT decision-makers and people in IT procurement have seen various types of software licensing evolve. On-premise software licensing, normally charged per user or per processor, has moved to a subscription-based licensing model with the growth of Software as a Service (SaaS). But as Prem Ananthakrishnan, global software practice lead at Accenture notes, there is a new model that is now emerging, based on the idea of measuring the value that the software or service provides.

Looking at the evolution of software licensing, he says: “The perpetual licensing model, where you would pay a one time fee for the software and own that version forever, has been rapidly declining. The dominant model today is where the customer pays a recurring fee. This could be per user or per tier for a given term.”

This model is predictable for software firms, who, he says, are able to extract value on a fixed term basis.

While there was some resistance from IT buyers over the move from buying software as capital expenditure to an OpEx (operational expenditure) model, he says, it is helping them avoid buying shelfware or being locked in. “They also recognise that they can pay, see value and then continue to invest incrementally in software,” he adds.

Another model that has been gaining traction over the last decade is based on usage, where an IT buyer purchases consumption credits. “Think of the cell phone type plans popularised by the hyperscale cloud providers,” Ananthakrishnan says.

Accenture is now seeing the emergence of outcome-based or value-based pricing models as Ananthakrishnan explains: “The idea here is that you pay for the actual results or business impact that the software creates and not just for the access.”

“When Gen AI came to the scene with ChatGPT in 2022, people quickly realised that AI makes software dynamic.” Unlike traditional software where the value sits in dashboards and reports Ananthakrishnan says agentic AI and AI software is smart: “It thinks, it acts and it reasons.”

It also consumes compute resources in real time and is unpredictable. As Ananthakrishnan points out,  software providers find it difficult trying to apply a static software licence to price something that is dynamic.

From an IT buyer perspective, Ananthakrishnan says the big shift is a fundamental change from understanding that software cannot just be purchased as a tool. “Think about software as a collaborator that’s driving an outcome for the business,” he says.

For now, IT buyers will have seen the use of tokens as a way of measuring AI’s value. In effect, this offers a form of usage -based licensing that approximates value based on the number of tokens - or phrases - a generative AI engine handles. Ananthakrishnan says: “The tokenisation or using computing credits or GPU credits, are all just proxies for value.”

Accoridng to Ananthakrishnan, the reason most AI providers have gravitated towards token-based pricing is because it offers a simple way for customers to understand value. He says: “Truly understanding value points and attributing the value generated by software to a particular [value metric] is a very, very hard problem to solve.”

Nevertheless, for Ananthakrishnan, this is where the software market is heading.