The Metric We Have to Drop (Pricing Series, Part 3)
The Setup
I’ve been (accidentally) writing a series about how we price and measure knowledge work, and why AI is breaking the math we’ve been using. Part 1 showed me the meter was wrong. Part 2 gave me a lens for which variable to push. Both posts left me stuck on the same question: okay, so what does better measurement actually look like? This isn’t theoretical for me right now. I’ve been in pricing conversations recently where you’re in a room trying to build a KPI that captures the value of judgment work, and every metric you reach for either gets gamed or misses the point. A response-time SLA measures speed but says nothing about whether the work was good. A satisfaction score measures perception but neither side controls all the variables. A volume metric rewards activity, not impact. No single number captures it. I kept hitting the same wall.
My Approach + AI Role
So I did what I always do when I’m stuck. I went looking for who already figured it out. I brought the question into an AI session and started pulling research. Published papers, blogs, white papers… which industries wrestled with the same problem: how do you measure the productivity of people whose most valuable work doesn’t show up in the system tracking them? Software engineering had twenty years of trying. The music industry went through a complete unit-of-measurement transition. I went through all of it. The surprise wasn’t what they found. It was where they all stopped.
What Actually Happened
Every industry that tried hard enough landed on the same boundary. Software engineering is the most metrics-literate industry on earth. Google’s DORA framework and Microsoft’s SPACE framework both emerged from years of data across tens of thousands of professionals. Both are widely adopted. Both respected. And both draw the same line: measure the team, measure the system, do not score the individual. Not because they couldn’t figure out how. Because every time they tried, it broke. The proxy corrupted the behavior it was supposed to measure. Goodhart’s Law, running on repeat. When McKinsey tried to cross that line in 2023, the researchers who built the actual frameworks pushed back hard and publicly. Part of why it struck a nerve: a popular reason leaders want individual measurement is to decide who to cut. The framework authors knew it. They’d watched it happen. Here’s where it connected for me. Those pricing conversations I mentioned… same pattern. Same wall. And one thing I keep gravitating back to (still working this out): two metrics pulling against each other might be more honest than one perfect metric. When neither score alone tells the story, you have to sit in a room together and interpret what’s actually happening. That conversation is the measurement. Not the dashboard. Think about it. A response-time SLA plus a client-satisfaction score. Neither is sufficient alone. But when one goes up and the other goes down, you can’t just celebrate the dashboard. You have to ask what’s actually happening. You have to talk to each other. And that forced interpretation, the one where two people who see different parts of the picture have to reconcile what the numbers mean… that might be closer to honest measurement of judgment work than any single KPI we’ll ever build. I don’t have this fully formed yet. But I think the principle is: design for productive tension between metrics, not for the perfect metric. The conversation the tension forces is where the real signal lives. Now look at how this plays out in practice. If you’ve been following this series, you’ve already seen the pieces. Part 1 showed where the margin actually lives in the individual rate structure. Part 2 showed what happens at the project level when you push the wrong variable. They’re the same system: The individual rate builds the team. The team shapes the SOW. The SOW determines what gets measured. And what gets measured is what gets optimized. The client reinforces every step. They’re buying the thing they can hold. The deliverable, the asset, the output. Strategy? That’s just meetings where people talk about ideas (never mind that it determines whether the output is worth producing). Account management? That’s just meetings and recap notes right? (Never mind that it’s the reason the project didn’t go sideways three times last quarter.) Creative concepts? Just pretty pictures that aren’t the thing you can hold (never mind that it’s what makes an insight come to life in a way you can actually feel). So those lines get squeezed in the SOW. And quietly, the work that actually differentiates you from the pure executional shops disappears from the price. So when AI shows up and a leader asks where to apply it, the dashboard points at the only thing it can see: execution. Output. Volume. And on the dashboard, that looks like progress.
The Real Insight
I expected to find a smarter formula. A better way to count. Instead I found a consensus… and a connection that’s hidden. The software engineers figured out that individual productivity metrics destroy collaboration. The agency world is still trying to build the perfect individual KPI for judgment work. These are the same problem. I can tell you: the measurement consensus from engineering hasn’t made it into a single pricing conversation I’ve been part of. The most valuable work in any knowledge organization (judgment, the ability to hold coherence across people who disagree, the instinct that says “this brief is off” before a single asset gets built) doesn’t fit in a cell on a spreadsheet. It never will. And every attempt to force it into one eventually destroys the thing it was trying to measure. Here’s where I’ve landed after three posts of pulling this thread. I think the first step toward any new model, whether that’s outcome-based pricing or capacity bands or whatever we haven’t named yet, is being willing to let go of the individual productivity metric. Not because people shouldn’t be accountable. Because the metric was never measuring what we actually value. It was measuring what was easy to count.
Try This If…
If you’re leading a team through AI adoption and someone asks something like “how do we measure individual productivity now?”… pause before you answer. Ask instead: who on your team does the work that never shows up on a report but somehow holds everything together? The person who reads the room before the meeting starts. The one who catches the misalignment before it becomes a redo. The one who translates between the client’s language and the team’s language so the brief actually lands. Can you name that work? Can your dashboard see it? If not, that’s the thing to figure out first. Before you optimize anything. Before you point AI at any workflow. Because that’s the work you’ll miss when it’s gone, and by then you won’t know what you lost.
Systems Lens
This is where the series lands. Three posts. One thread. Part 1: The meter is wrong. The leverage model says AI compresses your most profitable hours first, not your cheapest. Part 2: Here’s a diagnostic lens. Are you pushing quantity or value? And are you giving away the thinking that makes production worth doing? Part 3 (this post): Here’s what happens when you try to fix the meter. Everyone who went deep enough drew the same line. Measure the system. Protect the conditions. Let go of scoring the person. I don’t have the replacement model. I’m not sure anyone does yet (do let me know). But the music industry didn’t get from albums to streaming with a formula. It started with the willingness to let go of the old unit of measurement before the new one existed. I think we’re at that point. And the longer we hold onto individual productivity as the primary measurement, the longer we delay whatever comes next. I’m sharing this because this curiosity journey helped me. It gave me language for something I’d been feeling in those conversations. If you’re in budget meetings or staffing conversations and something feels off about how the work gets measured while everything’s changing with AI… it’s not just you. The research says the same thing.