How software pricing will help prove the ROI of AI — or not
A recent MIT study put everyone on edge: generative AI is not returning ROI in 95% of cases, the researchers found, despite $30 billion to $40 billion in enterprise investment.
Processing Content
But the research also revealed this isn’t AI’s fault per se. Companies don’t have all the pieces in place to make AI effective, and most gen AI systems “do not retain feedback, adapt to context, or improve over time.”
What’s more, software leaders are clearly anticipating an AI upside, McKinsey reports. Its research finds that “40% of software leaders expect AI to unlock more than 20% revenue growth beyond their current trajectory, with 11% anticipating gains of over 50%.” On top of that, McKinsey notes that “AI-centric organizations” are reporting 20 to 40% reductions in operating costs.
Proving ROI
So, how will AI really prove itself when it comes to ROI?
Having real-time data is key for any AI initiative to be able to learn and adapt, which the MIT study noted is often lacking. On this front, pricing and billing automation can play a big role in making real-time data available to increase the ability to showcase the value of any AI software. Here’s how:
Pricing linked to value: Fast-growing software companies, especially those with AI-infused products, are adopting at least some level of variable pricing instead of straight subscriptions or seats. With the traditional subscription or seat model, companies pay monthly and use the software as much or as little as they want. With variable pricing, they might pay based on what they actually use — or for actual outcomes. Zendesk, for instance, offers outcome-based pricing for its AI agents. Businesses only pay when their customers’ issue is fully resolved by the AI. Others may charge for how often an AI agent answers a call or completes a conversation, etc.
The point is that, with variable pricing versus a straight subscription model, the software buyer gets more granular and real-time data and insight into the impact the software is having. Outcome-based pricing, in which a customer pays for software when that software delivers an agreed upon outcome, is a clear way to prove the value of the AI software. If software is priced per month or per user, it’s very hard to tie an actual outcome that drives ROI back to that software.
The variable part of the pricing can help clarify the ROI of AI, if and when it exists. No surprise. In the MIT study, the ability to quantify an “outcome” was often cited as something that those who were winning with gen AI were focused on and were able to do.
Today, many software companies offer both subscription pricing and variable pricing — either usage or outcome — for some of its services. Nine out of 10 of our customers use a hybrid model, for instance. Industry research mirrors that result with 77% of large SaaS players now having adopted some level of usage-based pricing. McKinsey also notes that, “between 2015 and 2024, the number of consumption-based software companies more than doubled, and leaders such as Salesforce, Zendesk, Intercom and LexisNexis are already monetizing their AI capabilities through these models.”
Billing transparency: With usage or other variable pricing, companies will get a ton of data about how and when the AI software is being used. If the AI software is replacing manual processes, it’ll gather a lot more data and that data will also be more reliable. More data and more reliable data will lead to more reliable measurement of software effectiveness. All of this can be presented to customers in real time via billing dashboards, increasing billing transparency and giving customers more data to pinpoint the value from AI-infused solutions.
Automation and AI-infused pricing tools will also help reduce revenue leakage. Software companies with complex pricing typically lose 1%-3% of revenue to under-billing, and sometimes more. They don’t bill correctly, they take too long to update pricing, etc. We’ve had customers who weren’t even invoicing some of their long-standing software users.
With automated and AI-infused pricing tools, such revenue leakage will likely be more easily spotted and prevented. That, too, will prove the value of an AI tool.
Pinpointing infrastructure ROI
Proving the ROI of any new technology is always a challenge and the stakes are especially high given the massive amount of investment going into AI. It’s also hard to pinpoint the ROI of software infrastructure that’s more under the hood than top-line applications.
But usage data linked to variable pricing for AI software will go a long way to providing the underlying data to help enterprises make more informed choices about whether that software is making a measurable difference, or not.
