Most agency owners can recite revenue per billable headcount from memory. Ask them how their middle managers are doing and you’ll get a longer pause. The metrics tell a clean story about revenue and utilization and miss almost everything that determines whether the business is still standing in two years.
That gap is the through-line of Episode 2 of Billable. Grant and Kurt spent 25 minutes pulling apart the numbers agencies rely on, and what those numbers don’t see. Not a “throw out your KPIs” argument. More like a “your KPIs are answering a question you stopped asking five years ago” argument.
Revenue per billable headcount, utilization, gross margin. Those are fine. They tell you how the business performed last month. They don’t tell you whether the team carrying that performance is two months from a resignation wave, whether the project that made the margin number look healthy was actually a write-off in disguise, or whether the growth that smoothed over last quarter’s bad call is about to stop being a cushion.
Most agency leaders weren’t trained for the management problem
A small group of agency owners chose this path on purpose. Most didn’t. They were designers who got good at client management, engineers who built something around a CMS, account managers who kept getting promoted until the title said “partner.”
No MBA. No operations training. Nobody handed them a playbook for running a 40-person company.
Grant’s version of this: he started in account management, moved into project management, and gradually ended up as COO. Not a career plan. A series of situations that required someone to figure it out.
Easy to read that origin story as a weakness. It’s not. Agencies that stay nimble, that adopt new models faster than their enterprise counterparts, do so precisely because their leaders aren’t anchored to a textbook framework. But it also means the measurement systems those leaders build tend to be borrowed. Pulled from a Deltek report or an industry benchmark without much interrogation of whether those numbers actually describe the firm in question.
What comes out the other side is a dashboard that looks rigorous and feels reassuring, built on assumptions nobody chose deliberately. Worth asking, periodically, whether the questions your systems are answering are still the questions you need answered.
Five numbers on a desert island
Grant proposed an exercise partway through the conversation. If you were on a desert island and could only see five to ten numbers about your organization, which ones would tell you whether it was healthy?
Revenue per billable headcount makes the list. Revenue per total headcount. Utilization rates. Fine. Standard.
But Grant’s point was sharper than the exercise. Those numbers don’t tell you whether your team is burning out, whether the person carrying three projects is two months from quitting, or whether the senior PM who’s been holding the agency’s worst client together is starting to look at her LinkedIn for the first time in years. They can’t measure culture erosion. They definitely can’t tell you whether the growth that looks so good on the dashboard is actually sustainable.
You can build vanity KPIs that say you’re winning while the organization underneath is hollowing out. Growth in revenue, growth in headcount, growth in new logos, growth in conference appearances. Any of it can mask structural problems that only surface when the growth slows down. Kurt’s line on this one landed hard: growth covers up a lot of problems.
What changes isn’t the KPI set. It’s how you read the gap between what your numbers show and what your week feels like. When those two stories stop agreeing, the numbers are usually wrong. Or asked the wrong question.
Communication is the infrastructure, not the tool
Communication came up three separate times in the episode, which tells you something about how central it is to the argument.
Kurt’s framing: most problems in a business are communication problems. Not a new idea. The layer he added was useful, though. Having Slack doesn’t mean communication is happening. Having a weekly standup doesn’t mean the right information is moving to the right people. The tools exist. Confirmation that anyone received, processed, and acted on the message usually doesn’t.
His rule of thumb, borrowed from his father: something has to be read seven times, heard three times, and felt once before it sticks. If you said it in a Slack message and assumed it landed, you communicated at yourself, not at your team.
This is the part most agencies miss when they invest in operational maturity. More tools get bought. More standups get scheduled. Nobody closes the loop on whether the cadence is actually producing alignment. A weekly resourcing meeting that ends with “we’ll figure it out” is not a resourcing meeting, it’s a calendar event. The same logic applies to every other recurring touchpoint on the calendar.
Micromanagement is inconsistency, not oversight
Sharpest reframe in the episode. Kurt defined micromanagement not as too much oversight, but as inconsistent oversight. He called it the seagull maneuver: fly in, make a lot of noise, leave a mess, disappear.
Teams resist that pattern, not the manager who checks in regularly with clear expectations. They resist the one who shows up unpredictably, reacts to whatever catches their eye, and vanishes until the next crisis.
Consistency is the differentiator. Predictable check-ins, clear definitions of what good looks like, follow-through on the things you said mattered last week. Teams don’t push back against structure. They push back against randomness.
There’s a measurement angle here too. Predictable management produces predictable performance, and predictable performance is what makes forecasting possible. When leadership is reactive, the line between “this PM is underperforming” and “this PM is being thrashed” gets impossible to draw. Dashboard says one thing. Reality says another.
Where coaching effort actually compounds
Grant and Kurt landed on a framework for team performance that agencies rarely follow in practice. A bottom 10 to 15 percent exists on every team, the people who won’t get it no matter how much time you invest. They end up absorbing a disproportionate share of management attention. That math almost never works.
Kurt’s version: hiring is guessing, firing is knowing. You’ll never be 100 percent certain a hire will work out. But if someone has been on the team long enough and the pattern is clear, the decision is already made. You just haven’t acted on it yet.
Real leverage lives in the middle 80 percent. Not the stars who’ll be fine regardless. Not the bottom tier who probably need a different role. It’s the large middle group that responds to investment, coaching, clarity, and a manager who shows up consistently. Effort compounds there.
Worth saying directly: this isn’t about labeling people. It’s about where leadership time actually moves the needle. Most agency owners we talk to are spending 40 percent of their management bandwidth on the 10 percent of the team where it produces the least return.
Retainers change what you measure
Toward the end of the conversation, the topic turned to a shift that deserves its own episode. Progressive agencies are moving from time-and-materials billing to retainer-based recurring revenue. Sounds like a pricing decision. It’s actually an operations decision.
When you bill by the hour, your KPIs are built around utilization, billable hours, and efficiency per person. When you bill on a retainer, the measurement changes. Kurt pointed to agencies where the new performance signal is iteration speed. Three campaign cycles in a day instead of one. Value delivered to the client in a Tuesday test that nobody can cleanly attribute to a single role.
Old measurement infrastructure can’t capture that. And most agencies haven’t built the new one yet. They’re trying to run a recurring-revenue business on a billable-hour scoreboard, and the numbers keep telling them the work is fine even when the client is quietly disengaging.
Watch the operational signal underneath, not just the financial one. Whether the team’s planning rhythm has actually changed to match the new commercial model is what tells you if the shift is real. Pricing without an operating model behind it is just a number on an invoice.
One more thing about incentives
Grant brought up Anthropic giving unlimited Claude tokens to employees as an incentive. Not a pizza party or a plaque. Access to the tool that makes you better at your job, without a meter running.
Small example, but it says something about where incentive design is heading. Agencies that figure out how to reward the behaviors they actually want, not the behaviors their KPI dashboard happens to measure, will pull ahead. Everyone else will keep optimizing for last year’s scorecard and wondering why their best people keep leaving.
If any of this lines up with the conversations you keep having about your own agency, the full episode is worth 25 minutes. Grant and Kurt go deeper on every thread above, and a few that didn’t make it into this post.
Surviving the next two years won’t come down to who has the cleanest KPI dashboard. It’ll come down to who noticed what those dashboards weren’t showing them.
Frequently Asked Questions
Nothing, taken on their own. They tell you how the business performed. They don't tell you why, or whether the performance is durable. Utilization can look healthy while two of your senior PMs are running on fumes. Revenue per head can hit target while a client relationship is quietly unraveling. The numbers aren't lying. They're answering a narrower question than most leadership teams realize.
Simplest test: when your dashboard says one thing and your week feels like something else, the dashboard is usually wrong. Or at least incomplete. Most agency owners can name two or three problems they've been watching for months that don't show up anywhere in their reporting. That gap is the answer.
Episode 2 reframes the term. What teams resist isn't oversight, it's randomness. A manager who checks in consistently with clear expectations is not micromanaging. A manager who descends unpredictably, reacts to whatever catches their eye, and disappears until the next crisis is. Consistency is the variable that matters.
Billable-hour models reward output per person per hour. Retainer models reward value delivered over time, often through faster iteration cycles. The dashboards built for the first model can't read the signal in the second. Agencies making this shift commercially usually haven't rebuilt the operating model underneath it, which is why the numbers stop reflecting reality.
Grant and Kurt's framework: not on the top 10 percent, who'll be fine regardless. Not on the bottom 10 to 15 percent, who probably need a different role. Compounding return lives in the middle 80 percent, the team members who respond to investment, clarity, and a manager who shows up consistently.
Because the tools and the cadence don't guarantee that information is moving. Kurt's rule of thumb in the episode: something has to be read seven times, heard three times, and felt once before it sticks. The agencies that close that loop deliberately, instead of assuming it happens, are the ones whose plans actually survive contact with the team.
