No AI, no promotion and no clear picture of the bill: Companies struggle to track AI usage

When Accenture CEO Julie Sweet told employees that mastering AI tools is now a prerequisite for career advancement at the consulting giant, it sent a clear signal about how seriously corporate America is taking artificial intelligence. But as companies race to embed AI into their operations, a new survey reveals a striking paradox: most businesses have little idea how much AI they are actually using or what it is costing them.

According to a KPMG survey cited by the Wall Street Journal, only 26 per cent of companies say they have a comprehensive view of their AI costs, while 50 per cent have some visibility and 22 per cent report no visibility whatsoever. The findings expose a widening gap between the pace of AI adoption and the financial controls companies have in place to manage it.

At the heart of the challenge is a fundamental shift in how AI vendors price their products. The industry is moving away from predictable, flat-fee software subscriptions toward usage-based pricing measured in tokens, which is the basic unit of AI computing.

For CFOs long accustomed to fixed technology budgets, this creates a cost structure that is both volatile and difficult to forecast.

The problem is compounded when companies scale up AI-driven workflows. Affirm, the financial-services firm, discovered this when it significantly increased the share of code written by AI agents during its March quarter.

Finance chief Rob O'Hare told the WSJ the company experienced what amounted to an almost overnight step-change in token consumption. To stay on top of it, Affirm now monitors AI usage on a near-real-time basis and conducts weekly cost reviews with its leadership team - a level of financial vigilance that most companies have yet to build.

Even where companies deploy AI tools, there is no guarantee employees will use them. Reckitt, the UK-based consumer-goods giant, rolled out 12 AI solutions for its marketing team, only to find that usage dropped sharply after a few weeks as staff drifted back to familiar ways of working. CFO Shannon Eisenhardt told the WSJ the company is now closely scrutinising how employees engage with AI and adjusting spending where adoption has stalled.

The consequences go beyond wasted subscriptions. Reckitt also discovered that one of its marketing AI tools was generating inaccurate and insufficient data, prompting the company to slow its rollout. Eisenhardt acknowledged that the delay has pushed expected savings further out, with returns on AI investments arriving more slowly than planned.

The KPMG findings arrive at a critical moment. Corporate investment in AI is accelerating, and vendors are increasingly designing pricing models that reward heavy usage and shifting financial risk squarely onto the enterprise customer. For companies that cannot see where AI is being used, how much is being consumed, or whether it is actually delivering value, that risk is flying blind.

Sweet's mandate at Accenture - use AI or forfeit your promotion - reflects the urgency with which companies are trying to drive adoption. But the KPMG survey suggests that the harder task may not be getting employees to use AI, but building the financial infrastructure to understand what that usage actually means for the bottom line.