Most Firms Can't Prove Their AI Works. They're Scaling It Anyway.

Forty per cent of professional services firms — across legal, tax, accounting, and risk — now use generative AI in their workflows. That's nearly double the 22% recorded a year earlier, according to the Thomson Reuters Institute's 2026 AI in Professional Services report, a survey of more than 1,500 professionals now in its fourth year. Over 80% of current users engage with AI weekly or more. The adoption curve isn't bending. It's steepening.

But here's the figure that matters more: only 18% of those organisations measure whether AI is delivering a return on investment. Forty per cent of respondents don't even know if their firm tracks ROI at all. The rest say it isn't tracked. Adoption is accelerating. Measurement isn't.

Of the minority that do track AI returns, the metrics skew inward. Seventy-seven per cent measure cost savings. Sixty-four per cent track employee usage rates. Fewer than 30% measure client satisfaction, revenue impact, or new business generated.

That imbalance matters. Cost savings and usage rates are the easiest metrics to report — and the least useful for understanding whether AI is shifting your firm's competitive position. A tool can be cheap and widely used and still add nothing to a client's experience or the bottom line. The question isn't whether people are logging in. It's whether the firm is more profitable because they are.

The Thomson Reuters data makes the consequence concrete: organisations with a formal AI strategy are more than three times as likely to report positive ROI as those without one. Strategy — a documented plan for what AI is meant to achieve, which workflows it targets, and how returns are measured — isn't overhead. It's the variable that separates AI investment from AI expense. This is fundamentally a cost intelligence problem, and most firms haven't built the measurement infrastructure to solve it.

The report also tracks what's coming. Only 15% of professional services firms currently use agentic AI — systems that don't just assist with a task but complete multi-step workflows autonomously: pulling data, drafting documents, logging time, and routing approvals without human intervention at each step. But 53% are planning or considering it, and 77% of professionals expect agentic AI to be central to their workflow by 2030.

For Australian accounting and law firms, agentic AI means tools that manage entire processes end-to-end. Not a chatbot that drafts a letter — a system that pulls the relevant precedents, drafts, formats, sends for review, logs the time entry, and updates the matter file. The operational leverage is significant. But firms that can't measure the return on today's comparatively simple AI tools are not positioned to evaluate systems that are more powerful, more expensive, and considerably more opaque.

Two-thirds of corporate clients already want their professional services firms to use AI, per the same report. But fewer than 20% mandate it through formal procurement. That gap will close. When it does, the firms that can demonstrate measured AI-driven efficiency will win work. The firms that adopted AI without tracking its impact will struggle to articulate why their fees should reflect the investment.

If your firm uses AI in any form — a ChatGPT subscription, a copilot in your practice management software, an AI-powered document review tool — start tracking three metrics this quarter. First, time saved: have each team member estimate hours per week before and after AI use on specific tasks. Second, cost avoided: what would you have paid a person, contractor, or outsourced provider to do that same work? Third, revenue converted: has the time freed up actually been redirected to billable work, or has it just made non-billable tasks faster?

We've covered the adoption gap in Australian law firms and the efficiency gains accounting firms are seeing with AI-powered document processing. This week's data from Thomson Reuters puts a sharper question underneath both: are the firms that adopted AI actually capturing the value, or just accumulating subscriptions? If you don't know the answer for your firm, that's the first thing worth fixing.

Key takeaways

Can your firm prove its AI is working?

Most can't. A short conversation is enough to identify what to measure, where the gaps are, and whether your current AI spend is earning its keep. Book a call to talk it through.