Your Resource Plan is Lying to You.
I want to talk about the most expensive leak in BC delivery. Resource planning that may as well be printed on the back of a cocktail napkin.
You see it every week. The planning tool spits out a number. 70%. The CFO shrugs and enters it. Margin assumptions get baked in. Hiring conversations start. Board prep begins.
Then timesheets get submitted. Write-downs hit. The customer pushes back on hours from three Tuesdays ago. The number lands at 62%. And that's if it’s a decent week. That’s utilization vs. realization. (We talked about this).
It happens in every BC partner I have worked with. The gap is real, it’s predictable, it’s infuriating. And the cost of it is about to go up.
Why the gap shows up only after it’s too late to fix it
The planned number counts hours that were assigned. The month-end number counts hours that were billed. Those have never been the same thing.
Take a typical consultant on a Monday morning: 36 hours of billable work assigned. Yay! Fully allocated.
Except when you look closer.
Task 1: The requirements haven’t been approved. Task 2: The customer still hasn’t provided access. Task 3: The design decision is sitting in someone’s inbox. Task 4: The upstream task is already behind schedule. Task 5: “WTH is this? Fixed Assets aren’t in scope!”
The work is scheduled, but the consultant simply can’t act on it. Not today at least. So, they pivot. Check a few emails. Have an internal Teams call. Ping a few PMs. Maybe they jump to another project for a couple of hours or reach out to customer support to see if they can grab a ticket or two.
The timesheet gets filled. 4 hours “billable”, 1 hour “internal meetings”, and 3 hours “admin” (every practice manager’s fave).
Enter the napkin.
The fix isn’t a new tool
This is the part that bothers me, because I have been saying it since before “AI” was on anyone’s radar.
The fix is that “assigned” and “actionable” are tracked as different things. In real time. As proactive resourcing adjustments instead of postmortems.
The Definition of Ready is the gate for the work. It keeps sales from throwing a poorly scoped SOW over the cubicle wall. It supports your high-level backlog.
The Definition of Actionable is the gate for the assignment. It tells you whether the resource can actually do this task today.
How do I define “Actionable”? I answer these five questions before flipping the status and scheduling the resource:
Are the prereqs in place? Environment built, data accessible, right customer SME confirmed available?
Is the required skillset a match? (An available consultant is not the same as the right consultant.)
Is scope confirmed? If not, are change orders approved? Internal (SA holding firm to MVP standards) and External (customer stakeholder)? In writing? With timestamps?
Are upstream dependencies clear? No red task blocking this one.
Has the status been updated in the last 24 hours? (If it’s stale, it’s usually wrong.)
That’s the framework. Five questions, every active assignment, every end-of-day or early morning, before the consultant logs in.
If you’re an ops leader reading this thinking “my PMs already do this”, yes. The good ones do. In OneNote. In their head. In a shadow resource planning sheet you have never seen. But do they have a documented single source of truth that confirms it? Could an AI agent read it?
Spit balling here, if I built a resource planning automation, I’d need four things. A single Dataverse foundation where PSA, Planner, and ADO/GitHub all land. A status flag on every active assignment that says actionable right now, not just assigned. A nightly sweep that updates the flag from what the data is saying. And a 7am notification to each PM listing what’s blocked before the consultant logs in.
“What about when the devs refuse to participate and the PMs half-ass the assignment fields when they get busy?”
Good question. That’s the design pressure.
We don’t read what devs type. We read what they do (commits, pull requests, work item state changes). If nothing’s moved in three days, the system surfaces it. The PMs tag a skill requirement at assignment creation. Ten seconds. Skip it and the assignment stays yellow until they fix it.
But point taken. Ungoverned data, shadow systems, and undisciplined adoption put your cool automation tool right back on the napkin with the margin leak.
In other words, AI automation doesn’t accommodate half-assing. It surfaces it.
Why this matters more now than it used to
Five years ago, a 40-consultant practice running at 70% with potentially a 10-point gap between planned and actual could still hit margin. The CFO lived with it. Margin absorbed it. Customers tolerated the staffing delays. Competitors had the same problem. And padded T&M budgets, combined with decent change order enforcement, kept project budgets “greenish yellow”.
Today, the market looks different.
Fixed-fee engagements are now standard. Customers expect outcomes, not effort. Competitors are pricing tighter while still hitting margin. The slack that used to absorb resource-planning sloppiness is gone.
Five years from now, the practice next door will be running at 85% because they let AI read their Definition of Actionable and reshuffle resources before 9 am. They are not smarter than you. They do not have better PMs than you.
They made their first “AI project” operational maturity and let AI agents elevate from there.
AI doesn’t replace operational maturity. It rewards it. Big time. A 10-point bump in utilization on a 40-consultant practice (at $250 an hour) is an extra $2 million in revenue without adding headcount. Push that bump to 15 points (85% utilization, totally doable) and you’re at $3 million. Annually.
That’s the competitive threat. AI is about to widen the gap between mature delivery organizations and everyone else. And there is no shortcut to closing that gap.
One final thought
If this seems like a lot of moving parts and it is all making you want to put in for early retirement and tip toe straight to the golf course, please don’t. Just take it one step at a time. One leak at a time.
Or as Satya Nadella put it in Microsoft’s 2025 Annual Report:
“To succeed, we must continue to think in decades but execute in quarters…”
What can your team execute in the waning weeks of Q2? What about Q3? Q4?
Will operational maturity be your first AI project? Need an extra pair of hands?
Let's elevate. DGM