Software vendors keen to monetize AI should tread cautiously, since they risk inflating costs for their customers without delivering any promised benefits such as reducing employee head count.
The latest report from McKinsey & Company mulls what software-as-a-service (SaaS) vendors need to do to navigate the minefield of hype that surrounds AI and successfully fold such capabilities into their offerings.
According to the consultancy, there are three main challenges it identifies as holding back broader growth in AI software monetization in the report "Upgrading software business models to thrive in the AI era."
One of these is simply the inability to show any savings that can be expected. Many software firms trumpet potential use cases for AI, but only 30 percent have published quantifiable return on investment from real customer deployments.
Meanwhile, many customers see AI hiking IT costs without being able to offset these by slashing labor costs. The billions poured into developing AI models mean they don't come cheap, and AI-enabling the entire customer service stack of a typical business could lead to a 60 to 80 percent price increase, McKinsey says, while quoting an HR executive at a Fortune 100 company griping: "All of these copilots are supposed to make work more efficient with fewer people, but my business leaders are also saying they can't reduce head count yet."
Another challenge is scaling up adoption after introduction, which the report blames on underinvestment in change management. It says that for every $1 spent on model development, firms should expect to have to spend $3 on change management, which means user training and performance monitoring.
The third issue is a lack of predictable pricing, which means that customers find it hard to forecast how their AI costs will scale with usage because the pricing models are often complex and opaque.
To address these, McKinsey focuses mainly on how software firms should structure their pricing in the age of AI, rather than the wisdom of infusing AI into everything in the first place.
The report considers it unlikely that the traditional per-user monthly subscription model will disappear entirely, but expects that vendors will have to incorporate some form of consumption-based pricing into the mix.
Many are starting with hybrid models, where "additional" consumption that goes beyond a capacity cap is treated in different ways, such as metered throughput that limits the number of tokens processed daily, weekly, or monthly for certain models.
However, firms with hybrid models will need to revisit their choices frequently, it warns, as the rapid pace of AI evolution means that capabilities that are cutting-edge at launch can quickly become table stakes.
Vendors also need to choose their pricing unit carefully, whether that is a per-user flat fee with a capacity cap, like Microsoft Copilot, on a per-task basis, or perhaps on an outcome basis, such as per qualified lead for sales tools.
However, McKinsey also claims that the cost of inferencing is dropping rapidly, and so vendors need to consider carefully how they balance charges with growing adoption. The cost of large language model (LLM) delivery has declined by more than 80 percent per year over the past two years, it says.
Many SaaS companies believe they need to encourage trials to increase adoption, by offering free initial usage allocations for AI capabilities, for example. Once customers adopt and see value, the thinking goes, the firm can then look to upsell to a higher allocation for additional use cases. The problem with that, of course, is that one MIT study found that many enterprise organizations have so far seen zero return from their AI efforts.
Buyers are also changing, McKinsey believes. It says purchasing decisions are shifting from the IT department to line-of-business units. These leaders are increasingly making budget trade-offs between head count investment and AI deployment, and expect vendors to engage them on value and outcomes, not just features.
That could be a tricky sell, when trials of AI tools such as Microsoft's Copilot by a UK government department reveal no discernible boost in productivity. Still, the AI firms have to recoup all those billions they've already invested somehow, don't they? ®