The Real AI ROI Metric: Expertise Compounding Rate

If you ask most professional services leaders how they’re measuring AI ROI, you’ll get some version of the same two answers: time saved or tasks completed. I think both of those are the wrong metric, and I want to explain why, because I think the industry is about to spend a lot of money optimizing for the wrong thing.

Activity isn’t Value

Time saved and tasks completed measure activity. They tell you whether a person or a team is moving faster than before. What they don’t tell you is whether the work getting done faster is actually better, or whether it’s building anything that lasts beyond that single task. You can be extremely efficient at producing the same shallow output over and over again.

But that’s not the outcome any of us are actually trying to buy when we invest in AI.

A Better Metric

What I’d propose instead is a metric I call the “expertise compounding rate.” This how much of the insight and expertise already sitting inside your business is actually getting surfaced, shared, and reused, rather than created once and forgotten.

Here’s the philosophy behind it: Every interaction your firm has generates value that most businesses never fully capture. Every conversation with a client, every problem your team solves, every document someone writes, every email that gets sent is an opportunity to enrich the next interaction, the next colleague, the next piece of work. Most of that value evaporates the moment the task is finished, because there’s no mechanism to surface it again when it would actually help someone else.

Start with the Basics

So how would a firm actually calculate this? Start with the basics.

Look at how often the insights and information already resident in your business are actually being leveraged, not just stored. That’s the simplest version of the measure. It’s a piece of prior work informing new work, or is it just sitting in an archive somewhere. An enriched insight, in my definition, is one that’s genuinely informed by something your firm has already done, not a generic answer that could have come from anywhere.

Going Deeper: Asynchronous Learning Across Your Firm

Then go a level deeper and look at how those insights are getting shared across boundaries that used to be hard limits. The world is genuinely global now. Firms may still be organized around geographic or business unit lines, but knowledge doesn’t respect those boundaries anymore, and it shouldn’t have to.

Can a colleague in one office learn from a project that happened somewhere else entirely, without needing to have been in the room? Can your team enrich themselves asynchronously, based on interactions and outcomes that happened while they were asleep?

This kind of asynchronous learning, at scale, is where I think the true value shows up. Data that used to be a stranded asset, sitting in a single person’s inbox or a single team’s project folder, becomes something the whole enterprise can draw from.

The Final Layer

The final layer is the one most firms haven’t even started thinking about: are you actually deriving new insight from that information, or just regurgitating what already existed?

Getting a fast, accurate answer to a question that’s already been answered before is not incremental value.

The real impact is in how that prior answer gets adjusted, improved, and enriched by new information as it’s applied again and again. That’s the difference between a firm that’s using AI to retrieve its own history and a firm that’s using AI to actually grow smarter over time.

The Dashboard You Actually Need

If I were running a large global services firm right now, this is the metric I’d be building a dashboard around: not utilization, not raw efficiency gains, but the rate at which our collective expertise is compounding.

Because in a world where every firm has access to the same AI models, the only sustainable advantage left is how well you leverage what your own people already know. That’s not something a competitor can buy off the shelf. It has to be built, deliberately, inside your own business.