As legal AI companies scale and raise capital, a familiar metric is coming under new scrutiny: annual recurring revenue.
A recent discussion, sparked by Spellbook CEO Scott Stevenson and explored in depth by Richard Tromans of Artificial Lawyer, is raising questions about what “recurring” actually means in the context of modern legal AI businesses.
At the centre of the debate is a growing disconnect between traditional SaaS metrics and the evolving business models underpinning legal AI.
Historically, legal technology companies have relied on predictable, per-seat subscription pricing, making ARR a relatively clean and consistent measure of growth. But the rise of generative AI is complicating that picture. Many legal AI platforms now blend subscription access with usage-based pricing, pilot programs, and enterprise deployments that don’t always fit neatly into conventional definitions of recurring revenue.
That shift has prompted industry insiders to ask whether ARR, long considered the gold standard for valuing software companies, still provides an accurate reflection of performance in this new category.
Tromans, writing in Artificial Lawyer, highlights how legal AI vendors are increasingly navigating hybrid models, where revenue may fluctuate based on usage, expand rapidly through enterprise adoption, or begin as short-term trials before converting into longer-term contracts. In this environment, the line between recurring and non-recurring revenue can blur.
Stevenson, whose Toronto-based company Spellbook has emerged as one of Canada’s leading legal AI startups, helped bring attention to the issue, pointing to the need for greater clarity and consistency in how companies report growth metrics. As more legal AI vendors enter the market and investor interest accelerates, those definitions are becoming more consequential.
The implications extend beyond accounting semantics. ARR plays a central role in how startups are valued, how investors assess performance, and how companies benchmark themselves against peers. If the underlying assumptions behind that metric are shifting, it could have ripple effects across the legal tech ecosystem.
For law firms, the debate also reflects a broader transition already underway. As AI-powered tools move from experimental pilots to embedded workflows, pricing models are evolving alongside adoption. Fixed subscriptions are increasingly complemented—or replaced—by models tied to usage, outcomes, or value delivered.
That evolution may ultimately require a new set of standard metrics better aligned with how legal AI products are built, sold, and used.
For now, the conversation initiated by Stevenson and brought into focus by Artificial Lawyer signals an industry in transition—one where even its most foundational metrics are being re-examined.