01Chapter 1 of 12
What resource management is (and is not)
Resource management is the weekly practice of matching the people you have to the work you have committed, or expect to commit, so delivery stays on time without burning people out or leaving paid hours idle.
Every services business, product team and internal delivery group runs on the same two curves: supply (hours people can actually work) and demand (hours the work needs). Resource management keeps those curves close together, week by week, and makes the gaps visible early enough to act: hire, borrow, re-sequence, renegotiate a date or sell more work.
Many teams still run this out of spreadsheets. In Wellingtone's State of Project Management survey, 44.3% of organisations built resource plans in Excel and only 19% used PPM or resource management software.[3] Spreadsheets can hold a plan; they rarely hold the loop of plan, actuals and leave that makes the plan trustworthy.
It answers four questions, and a good process answers each in under a minute: Who is available? Who is working on what? What is coming? Did we do what we planned? The first three are planning; the fourth is the feedback loop that makes the next plan better.
| Resource management is | It is not |
|---|---|
| A recurring decision loop between supply and demand | A staffing plan written once at kickoff |
| Measured in hours per person per week | Headcount on an org chart |
| Owned jointly: PMs bring demand, a resource owner manages supply | Whoever shouts loudest on Monday morning |
| About who can do the work, and when | Task management (what needs doing) or a Gantt chart |
| Planned hours compared with actual hours | Time tracking as surveillance |
Project manager vs resource manager
A project manager optimizes one project: scope, date, budget. A resource manager optimizes the portfolio: the best use of every person across all projects. The roles conflict by design, and that is healthy as long as the conflict is settled by priority rules agreed in advance (client tier, contractual dates, margin) rather than by negotiation every week. In small teams one person wears both hats; the rules still help.
Where is your team today?
Most teams move through five recognizable levels. Tick what is true for your team to see where you are and what the next level asks for.
Self-assessmentYour team
Resource management maturity in eight questions
Tick the statements that are true today. Each level builds on the one before it.
- Level 5Optimizing
- Level 4Forecasting
- Level 3Measured
- Level 2Visible
- Level 1Ad hoc
Spreadsheets per PM, staffing decided in chat, nobody sees the whole team.
Next: Visible. All bookings live in one shared place, visible to every PM. Bookings are in hours, not just names on a project.
Best-practice checklist
- One shared place where every booking lives; no private spreadsheets per PM
- A named owner of supply (resource manager, delivery lead or ops)
- A weekly 30-minute planning meeting with a 4 to 6 week horizon
- Written priority rules for staffing conflicts
- Bookings in hours, not just names on a project
Pitfalls
- Treating it as admin. If the schedule is updated after decisions are made, it documents the past instead of shaping the future. Decide in the tool, not in chat.
- Optimizing one project at a time. Each PM grabbing the best people for their project produces a portfolio where two projects are overstaffed and three are stuck.
02Chapter 2 of 12
Capacity planning: available vs booked vs actual
Most overbooking starts with the wrong denominator. Plan against the hours people can really work, not the hours in their contract.
Capacity planning separates four numbers that are often blurred into one. Keep them apart and most planning arguments disappear, because everyone is finally talking about the same hours.
- Contracted capacity
- Contracted hours per week × weeks in period
- The theoretical ceiling. Never plan against it.
- Available capacity
- Contracted − public holidays − planned leave − recurring internal time
- Internal time: meetings, admin, support rotas, training, hiring.
- Booked hours
- Σ hours allocated to projects (confirmed + tentative, tracked separately)
- The plan.
- Free capacity
- Available − booked
- Negative means over-allocation.
- Actual hours
- Σ hours logged in approved timesheets
- What really happened; feeds every KPI in chapter 8.
The gap between contracted and available hours is larger than most plans assume. Walk through it for your own team: change the inputs and watch where the hours go.
Capacity waterfallExample data
Where a quarter of contracted hours really goes
One quarter (13 weeks) for a team. Adjust the inputs to match yours; the bars show what is left to plan against.
Available
73%
of contracted hours
Available hours
3,041
380 h per person
Free capacity
456 h
about 11.4 person-weeks
Less than three quarters of contracted time is available for project work. Plans built on contracted hours will slip.
Show the numbers as a table
| Step | Hours | Share of contracted |
|---|---|---|
| Contracted hours | 4,160 | 100.0% |
| Public holidays | −192 | 4.6% |
| Planned leave | −384 | 9.2% |
| Sickness and unplanned | −128 | 3.1% |
| Internal time | −415 | 10.0% |
| Available capacity | 3,041 | 73.1% |
| Booked on projects | 2,585 | 62.1% |
| Free capacity | 456 | 11.0% |
Plan in horizons
- Next 2 weeks: day-level bookings, firm. Changes are expensive; protect focus.
- Weeks 3 to 6: week-level bookings, confirmed work plus tentative holds. This is where the weekly meeting earns its keep.
- Next quarter: role-level demand (for example "2 senior frontend, 1 designer from May"), compared with available capacity to drive hiring and sales decisions.
Best-practice checklist
- Store individual working patterns: part-time, 4-day weeks, compressed hours
- Load public holiday calendars per country or office
- Subtract recurring internal time before booking project work
- Book in hours (or hours per day), never just “on the project”
- Keep tentative and confirmed bookings visibly different
- Review the next 6 weeks weekly and the next quarter monthly
Pitfalls
- The 40-hour fiction. Booking everyone to their contract ignores meetings, leave and holidays, so the plan is late before it starts.
- Forgetting ramp-up. New joiners and people switching projects are not at full speed for the first weeks. Book them lighter.
- Invisible non-project work. Presales, support, hiring and internal tools consume real hours. Book them like projects so they show up in capacity.
03Chapter 3 of 12
Utilization: definitions, healthy targets and why 100% is a trap
Utilization is the most quoted and most misunderstood number in resource management. Define it precisely, set targets by role, and never aim for 100%.
There are three useful utilization numbers. They share a denominator, and you must write that denominator down: most disagreements about utilization are really disagreements about whether leave and holidays count as available time.
- Billable utilization
- Billable hours ÷ available hours
- The revenue driver in services businesses.
- Productive utilization
- (Billable + non-billable project hours) ÷ available hours
- Includes internal products, presales, approved investment work.
- Scheduled utilization
- Booked hours ÷ available hours
- Forward-looking: tells you about next month before it happens.
Targets by role, not one number
For context, SPI Research reports average billable utilization in professional services of 68.9% in its 2025 benchmark,[2] falling to 66.4% in the 2026 edition, the lowest in the survey's history.[1] Averages across firms are not targets for your team, but they are a useful reality check for plans that assume 90%.
A single company-wide target punishes the people who hold the company together. Leads review code and mentor; PMs run ceremonies and talk to clients; managers hire. Set a band per role, check it monthly, and look at the spread, not just the average. Try the calculator with the role presets below; the bands are common starting points, not benchmarks, so tune them to your margin model.
Utilization calculatorExample data
Billable and productive utilization for one person, one month
Pick a role to load a starting-point band and typical hours, then adjust. Bands are common starting points, not industry benchmarks.
Billable
80%
Within the band
Productive
89%
billable + project work
Role band
75 to 85%
Engineer or designer
Show the numbers as a table
| Measure | Value |
|---|---|
| Available hours | 140 |
| Billable hours | 112 |
| Non-billable project hours | 12 |
| Billable utilization | 80.0% |
| Productive utilization | 88.6% |
| Role band (billable) | 75 to 85% |
Why 100% utilization breaks delivery
Knowledge work arrives unevenly and takes uneven time. Queueing theory describes what happens when a resource with variable work gets close to full: waiting time does not grow in a straight line, it explodes. In the classic single-server model, the time a new request waits before anyone starts on it is proportional to u ÷ (1 − u), where u is utilization. At 50% that factor is 1; at 80% it is 4; at 90% it is 9; at 95% it is 19.
Queueing modelModel
Why waiting time explodes as utilization approaches 100%
How long a new request waits before anyone starts on it, as a multiple of the average task length. Drag the utilization slider.
Wait before work starts
5.7×
the average task length
At 85% a new request waits 5.7 times longer than at 50%.
Show the numbers as a table
| Utilization | Low | Typical | High |
|---|---|---|---|
| 50% | 0.5× | 1.0× | 2.0× |
| 60% | 0.7× | 1.5× | 3.0× |
| 70% | 1.2× | 2.3× | 4.7× |
| 80% | 2.0× | 4.0× | 8.0× |
| 85% | 2.8× | 5.7× | 11.3× |
| 90% | 4.5× | 9.0× | 18.0× |
| 95% | 9.5× | 19.0× | 38.0× |
In practice that means the urgent bug, the client change request and the sick colleague all land on people with no slack, so they queue, overtime rises and quality drops. Slack is not waste: it is the capacity that absorbs variability. Healthy teams run their delivery people at roughly 75 to 85% billable and keep the rest for the work that keeps them good at their jobs.
Best-practice checklist
- Write down the utilization formulas and the denominator, and use them everywhere
- Set utilization bands per role and seniority, not one company number
- Review monthly trends; daily and weekly utilization is mostly noise
- Look at the distribution: who is at 110% and who is at 40%?
- Never tie individual bonuses to raw utilization; it rewards padding timesheets
Pitfalls
- The healthy average. An 80% average can hide two people at 115% and two at 45%. Report the spread.
- Chasing 100%. Full utilization removes the buffer that absorbs urgent work, so lead times balloon and overtime becomes normal.
- Shifting definitions. Changing what counts as billable mid-year makes trends meaningless. Version your definitions.
04Chapter 4 of 12
Forecasting and the pipeline
Confirmed work tells you about the next few weeks. The sales pipeline tells you about the next quarter, if you weight it by probability instead of hoping.
Demand comes from four places: confirmed projects, extensions of current work, pipeline deals that have not signed yet, and internal initiatives. Confirmed work rolls off as projects end, so a team that looks fully booked today can be half idle in eight weeks. Forecasting is about seeing that cliff early.
- Weighted demand
- Σ (estimated hours of deal × win probability)
- Per week and per role, not just in total.
- Forecast gap
- Available capacity − (confirmed + weighted demand)
- Positive: sell or reassign. Negative: hire, subcontract or move dates.
- Forecast accuracy
- 1 − |actual hours − forecast hours| ÷ forecast hours
- Track it; it tells you how much to trust next quarter.
Use tentative bookings for deals above an agreed stage. They make likely work visible on the calendar without committing people, and they turn the weekly sales and delivery conversation into a look at the same chart. Move the probability sliders to see how the gap changes.
Pipeline forecastExample data
Confirmed work plus probability-weighted deals vs capacity
A 10-person team over the next 12 weeks. Change the win probability of each deal to see the forecast gap move.
- Confirmed
- Weighted pipeline
- Available capacity
- Over capacity
Weeks over
5
demand > capacity
Weeks 5 to 12
94%
scheduled utilization
Weighted demand exceeds capacity in 5 weeks, peaking at 44 h over. Start hiring or line up a subcontractor now, or agree later start dates.
Show the numbers as a table
| Week | Confirmed | Weighted pipeline | Total demand | Capacity |
|---|---|---|---|---|
| W1 | 280 | 0 | 280 | 320 |
| W2 | 280 | 0 | 280 | 320 |
| W3 | 280 | 0 | 280 | 320 |
| W4 | 280 | 84 | 364 | 320 |
| W5 | 280 | 84 | 364 | 320 |
| W6 | 160 | 183 | 343 | 320 |
| W7 | 160 | 183 | 343 | 320 |
| W8 | 160 | 183 | 343 | 320 |
| W9 | 60 | 183 | 243 | 256 |
| W10 | 60 | 183 | 243 | 320 |
| W11 | 60 | 183 | 243 | 320 |
| W12 | 60 | 156 | 216 | 320 |
Decision triggers
- Hire or contract when weighted demand exceeds available capacity for the same role for 4 or more consecutive weeks, and the gap is beyond what overtime or re-sequencing can absorb. Start early: hiring takes weeks to months.
- Sell or reassign when scheduled utilization for weeks 4 to 8 drops below your role band. Tell sales which skills are free and when.
- Renegotiate dates when a confirmed project cannot be staffed without breaking another commitment. Earlier is cheaper.
Best-practice checklist
- Tentative bookings for every deal past an agreed stage, with start date, duration and roles
- One weekly 20-minute sales and delivery sync on the same forecast view
- Stage probabilities based on your own historic win rates, reviewed quarterly
- Roll-off dates visible for every current project
- Forecast accuracy tracked monthly
Pitfalls
- Booking the pipeline as confirmed. Creates phantom over-allocation and pushes people onto work that may never sign.
- Ignoring roll-offs. The bench appears “suddenly” only because nobody looked at when projects end.
- Sales promising start dates blind. If sales cannot see capacity, every deal starts “next Monday”.
Further reading on the Teamdeck blog
05Chapter 5 of 12
Allocation and skills matching
Allocation is choosing the right person for the work, at the right load, for the right length of time. Skills data makes it a decision instead of a habit.
When several people could take a piece of work, decide in a consistent order. A simple, shared order removes most staffing arguments and protects the people who are always “available” because they never say no.
- Required skills at the needed level. Non-negotiable.
- Availability in the period, after leave and existing bookings.
- Cost and seniority fit for the budget and the risk.
- Continuity: client or codebase knowledge shortens ramp-up.
- Growth: a stretch assignment, with support, for someone who wants to learn.
Skills coverage and single points of failure
A skills matrix lists people against skills with a proficiency level, for example 0 (none) to 3 (can lead). Coverage per skill is the number of people at level 2 or above. Any critical skill with coverage of 1 is a single point of failure: one resignation, one holiday, one illness, and work stops. The example below shows how quickly that becomes visible.
Skills coverageExample data
Skills matrix with coverage and single points of failure
Eight people, seven skills. Select a skill to see who can take the work next month, ordered by free hours.
| Person | Free h | |||||||
|---|---|---|---|---|---|---|---|---|
| Ana | Can lead: 3 | Independent: 2 | None: 0 | Basic: 1 | Basic: 1 | None: 0 | None: 0 | 24 |
| Ben | Independent: 2 | Can lead: 3 | None: 0 | None: 0 | Basic: 1 | Independent: 2 | Basic: 1 | 0 |
| Chloe | None: 0 | None: 0 | Can lead: 3 | Basic: 1 | Basic: 1 | None: 0 | None: 0 | 40 |
| Dev | Basic: 1 | Independent: 2 | None: 0 | None: 0 | None: 0 | Can lead: 3 | Independent: 2 | 12 |
| Eli | None: 0 | None: 0 | None: 0 | Can lead: 3 | None: 0 | None: 0 | None: 0 | 56 |
| Farah | Independent: 2 | Basic: 1 | None: 0 | Independent: 2 | None: 0 | None: 0 | None: 0 | 32 |
| Gus | Basic: 1 | Basic: 1 | None: 0 | None: 0 | Can lead: 3 | Basic: 1 | None: 0 | 8 |
| Hana | None: 0 | Independent: 2 | None: 0 | None: 0 | None: 0 | None: 0 | Can lead: 3 | 64 |
| Coverage | 3 | 4 | 1 (single point of failure) | 2 | 1 (single point of failure) | 2 | 2 |
iOS: 1 person at level 2 or above
- Chloe · level 3 · 40 h free
Single point of failure. Pair someone with Chloe on the next project to build a backup.
Fewer, larger allocations
Every extra concurrent project costs focus. Focus is already scarce: in Microsoft's 2023 Work Trend Index, 68% of people said they do not have enough uninterrupted focus time during the workday.[5] Splitting someone across many projects makes it worse, because every switch brings a different client, codebase, stand-up and Slack channel. Prefer allocating whole days to a project over a few hours on many, and cap concurrent projects per person (two or three is a common ceiling for delivery roles).
Best-practice checklist
- Keep a skills matrix with levels, reviewed at least quarterly
- Name a backup for every critical skill; plan pairing to build coverage
- Cap concurrent projects per person
- Allocate in days per week for deep work, not scattered hours
- Record why a staffing choice was made when it overrides the default order
Pitfalls
- The always-available hero. The same strong person lands on every escalation, burns out, and takes the knowledge with them.
- Resource hoarding. PMs keep good people booked “just in case”. Central visibility and priority rules fix it.
- Stale skills data. A skills matrix nobody updates is worse than none: it sends people to work they no longer do.
Further reading on the Teamdeck blog
06Chapter 6 of 12
Leave, holidays and part-time realities
Time off is not an exception to the plan. It is a predictable, seasonal part of capacity, and the teams that plan it in advance rarely miss a summer deadline.
Annual leave entitlements, public holidays and part-time contracts vary by country and by person. In the European Union every worker is entitled to at least four weeks of paid annual leave,[8] England and Wales have eight bank holidays a year,[9] and in the United States paid vacation is not required by federal law,[10] so it depends on company policy. A distributed team can easily have three different holiday calendars in the same sprint.
- FTE
- Contracted hours ÷ full-time hours
- A 24-hour contract on a 40-hour standard is 0.6 FTE.
- Working days available
- Weekdays in period − public holidays − approved leave
- Per person, per country.
- Leave-adjusted capacity
- Working days available × daily hours × FTE − internal time
- The number the plan should use.
Leave clusters. In most teams it peaks in the summer months and around the end of the year, exactly when clients want things shipped before their own holidays. The example below shows a 10-person team: contracted hours are flat, available hours are not.
Seasonal capacityExample data
Contracted hours are flat. Available hours are not.
A 10-person team over a year, with leave clustering in summer and December. Select a month for details.
Aug available
1,264 h
75% of contracted
Best month
Mar
1,616 h available
Lowest month
Aug
22% below Mar
- Available
- Leave
- Public holidays
Show the numbers as a table
| Month | Available | Leave | Public holidays | Contracted |
|---|---|---|---|---|
| Jan | 1,536 | 64 | 80 | 1,680 |
| Feb | 1,600 | 80 | 0 | 1,680 |
| Mar | 1,616 | 64 | 0 | 1,680 |
| Apr | 1,504 | 96 | 80 | 1,680 |
| May | 1,408 | 112 | 160 | 1,680 |
| Jun | 1,536 | 144 | 0 | 1,680 |
| Jul | 1,376 | 304 | 0 | 1,680 |
| Aug | 1,264 | 336 | 80 | 1,680 |
| Sep | 1,584 | 96 | 0 | 1,680 |
| Oct | 1,600 | 80 | 0 | 1,680 |
| Nov | 1,536 | 64 | 80 | 1,680 |
| Dec | 1,280 | 240 | 160 | 1,680 |
Part-time and flexible patterns
Store each person's real weekly pattern (for example Monday to Thursday, or 6 hours a day) instead of an FTE percentage spread evenly. A 0.8 FTE colleague who never works Fridays cannot cover a Friday release, however their percentage looks on paper.
Best-practice checklist
- Leave requests and approvals in the same system as bookings
- Approve with the capacity impact visible, not from an email thread
- Holiday calendars per country or office, loaded a year ahead
- Plan summer and year-end capacity a quarter in advance
- A minimum-coverage rule per team or client (for example one person who knows the system is always in)
- Track allowances so unused leave does not pile up into December
Pitfalls
- Leave by email. Approvals in inboxes never reach the schedule, so someone is booked while on a beach.
- Even FTE spreading. Treating 0.6 FTE as 60% of every day books people on days they do not work.
- December surprise. Unused allowance gets taken at the end of the year, all at once. Nudge people to plan it earlier.
07Chapter 7 of 12
Estimation vs actuals: the feedback loop
Estimates are forecasts, and forecasts improve only when you compare them with what happened. Time tracking is the measuring instrument, not the goal.
People consistently underestimate how long their own work will take, even when they know similar tasks ran late before. Psychologists call it the planning fallacy. Large projects show the same bias at scale: in a study of 1,471 IT projects the average cost overrun was 27%, and one project in six overran its budget by 200% on average.[7] The fix is not to try harder at guessing; it is to calibrate against your own history.
- Estimate variance
- (Actual hours − estimated hours) ÷ estimated hours
- Track by project type, phase and role.
- Calibration factor
- Median of (actual ÷ estimate) for similar past work
- Multiply new estimates by it: reference class forecasting.
- Plan adherence
- Hours logged on booked projects ÷ booked hours
- Low adherence means the plan is not where work really happens.
Reference class forecasting takes the outside view: instead of building an estimate only from the task list, look at how similar past projects actually ended and adjust. Below is an example history of eight projects. Set a new bottom-up estimate to see what your track record suggests.
Reference class forecastExample data
Calibrate a new estimate with your own history
Each dot is a past project: actual hours divided by estimated hours. The median and 80th percentile turn a bottom-up estimate into a realistic booking and a quote.
Book (P50)
580 h
+16% on the estimate
Quote against (P80)
640 h
+28% on the estimate
Explicit contingency
60 h
P80 minus P50
Show the numbers as a table
| Project | Estimated h | Actual h | Actual ÷ estimate |
|---|---|---|---|
| Marketing website | 320 | 368 | 1.15× |
| Mobile MVP | 900 | 1,170 | 1.30× |
| CRM integration | 240 | 228 | 0.95× |
| App redesign | 400 | 468 | 1.17× |
| Data pipeline | 600 | 840 | 1.40× |
| Customer portal | 520 | 572 | 1.10× |
| Public API | 180 | 207 | 1.15× |
| E-commerce store | 1,000 | 1,250 | 1.25× |
Close the loop every week
- Estimate in ranges: a likely value and a pessimistic one (P50 and P80).
- Book the likely value; hold the difference as explicit contingency.
- Log time against the same projects and phases you booked.
- Compare booked vs logged weekly, while there is still time to react.
- Recalibrate the reference class at every project close.
Best-practice checklist
- Timesheets logged at least weekly, against the same structure as bookings
- Timesheet approval before data feeds billing or payroll
- Weekly booked vs logged review per active project
- A numbers-first retrospective at every project close
- Historic actuals stored by project type for reference class estimates
Pitfalls
- Timesheets as surveillance. If logged time is used to blame, people log what they think you want to see and the data stops being useful.
- Comparing at the end. A variance discovered at project close is a post-mortem, not a correction.
- Single-point estimates. One number hides the uncertainty; ranges make contingency explicit.
08Chapter 8 of 12
Reporting and KPIs, with formulas
A handful of well-defined KPIs, reviewed at the right cadence, beats a dashboard of thirty. Here are the ones that matter, how to calculate them and what to do when they move.
Most organisations are still assembling these numbers by hand: Wellingtone found that 56% had no access to real-time project KPIs and 43% spent a day or more manually collating project reports.[3] Each KPI below has a formula, a review cadence and the decision it informs. If a metric does not change a decision, drop it. Example values are calculated live from a sample month so you can sanity-check the formulas.
Billable utilization
MonthlyBillable hours ÷ available hours
Example: 1,020 billable h ÷ 1,344 available h = 75.9%
Informs: Pricing, hiring and role bands
Scheduled utilization (next 4 weeks)
WeeklyBooked hours ÷ available hours
Example: 1,090 booked h ÷ 1,280 available h = 85.2%
Informs: Sell, reassign or protect capacity
Over-allocation rate
WeeklyPerson-weeks with booked > available ÷ all person-weeks
Example: 5 of 40 person-weeks = 12.5%
Informs: Rebalance before the week starts
Bench (unassigned) capacity
WeeklyAvailable hours − booked hours, next 2 to 4 weeks
Example: 1,280 − 1,090 = 190 h
Informs: Staff pipeline work or internal priorities
Plan accuracy
Monthly1 − |logged − booked| ÷ booked
Example: |1,150 − 1,090| ÷ 1,090 = 94.5%
Informs: How much to trust the next forecast
Estimate variance
Per project close(Actual − estimate) ÷ estimate
Example: (460 − 400) ÷ 400 = +15%
Informs: Calibration factor for new quotes
Project margin
Monthly(Revenue − Σ hours × cost rate) ÷ revenue
Example: (€48k − 460 h × €62) ÷ €48k = 40.6%
Informs: Scope, rates and staffing mix. For reference, SPI Research reports an average project margin of 37.7% for 2025.[1]
Realization rate
MonthlyBilled value ÷ (billable hours × standard rate)
Example: €86k ÷ (1,020 h × €95) = 88.8%
Informs: Write-offs, discounts and fixed-price risk
Overtime ratio
MonthlyHours above contract ÷ contracted hours
Example: 64 h ÷ 1,520 h = 4.2%
Informs: Burnout risk and hiring triggers
Reporting cadence
| Cadence | What to review |
|---|---|
| Weekly (30 min) | Free capacity and over-allocation for the next 6 weeks, booked vs logged on active projects, leave requests in the window |
| Monthly (60 min) | Utilization by role (trend and spread), project margins, estimate variance, forecast accuracy, overtime |
| Quarterly (half day) | Hiring plan from the forecast gap, skills coverage, utilization bands per role, stage probabilities |
Best-practice checklist
- Every KPI has a written formula, an owner and a review cadence
- Utilization shown as a distribution, not just an average
- Forward-looking metrics (scheduled utilization, forecast gap) next to backward-looking ones
- Same data source for planning and reporting; no copy-paste into slides
Pitfalls
- Vanity dashboards. Thirty charts that nobody acts on. Fewer KPIs, each tied to a decision.
- Rear-view only. Last month's utilization explains the past. Scheduled utilization changes the future.
09Chapter 9 of 12
Common anti-patterns, and what to do instead
Most resource management problems are not unique. They are well-known patterns with well-known fixes. Spot the symptom, apply the fix.
Anti-pattern 01
Planning on contracted hours
Symptom: Every week starts “fully booked” and ends late.
Instead: Plan on available capacity: subtract holidays, leave and recurring internal time first.
Anti-pattern 02
A spreadsheet per project manager
Symptom: The same person is booked at 100% on two sheets.
Instead: One shared calendar of bookings that every PM reads and writes.
Anti-pattern 03
The 100% utilization target
Symptom: Urgent work waits days, overtime is normal, quality slips.
Instead: Role-based bands (for delivery roles typically 75 to 85%) with slack for variability.
Anti-pattern 04
Booking the pipeline as confirmed
Symptom: Phantom over-allocation and people “reserved” for deals that never sign.
Instead: Tentative bookings weighted by probability, confirmed when the deal closes.
Anti-pattern 05
The hero bottleneck
Symptom: One expert is on every escalation and every holiday is a crisis.
Instead: Skills matrix, a named backup for each critical skill, pairing to spread knowledge.
Anti-pattern 06
Monday-morning staffing
Symptom: Decisions made in a scramble for this week only.
Instead: A weekly meeting that looks 4 to 6 weeks ahead, with priority rules agreed in advance.
Anti-pattern 07
Averages only
Symptom: Utilization “looks fine” while two people burn out and two sit idle.
Instead: Report the distribution per person and role, and flag outliers both ways.
Anti-pattern 08
Leave approved by email
Symptom: People are booked during their holiday; deadlines slip in August.
Instead: Leave requests in the planning system, approved with capacity in view.
Anti-pattern 09
Timesheets as surveillance
Symptom: Logs are filled in on Friday from memory, matching the plan exactly.
Instead: Use time data to improve plans and estimates, share it back with the team, never to blame.
Anti-pattern 10
Estimating from scratch every time
Symptom: The same type of project overruns by the same amount, again.
Instead: Reference class forecasting: calibrate new estimates with actuals from similar work.
10Chapter 10 of 12
A 30/60/90 day implementation playbook
You do not need a transformation programme. You need one shared plan, one weekly meeting and three months of steady habits. Tick items off as you go; your progress stays in this browser.
0 of 15 steps done
Days 1 to 30
0/51. Make it visible
Everyone sees the same plan.
Exit criterion: No bookings live outside the shared calendar.
Days 31 to 60
0/52. Measure it
Compare the plan with reality.
Exit criterion: Planned vs actual is a click, not a spreadsheet project.
Days 61 to 90
0/53. Look ahead
Decide before the problem arrives.
Exit criterion: Hiring and sales decisions cite the forecast gap.
11Chapter 11 of 12
Glossary
The vocabulary of resource management, defined the way this compendium uses it.
- Allocation
- Assigning a person to work for a period, usually in hours per day or week.
- Availability
- The hours a person can work in a period after their working pattern, public holidays and leave.
- Available capacity
- Contracted hours minus public holidays, planned leave and recurring internal time. The denominator for planning.
- Bench
- People or hours with no project booking in a period. Some bench is healthy; a growing bench is a sales or forecasting signal.
- Billable hours
- Hours that can be invoiced to a client under the contract.
- Billable utilization
- Billable hours divided by available (or contracted) hours, with the denominator stated.
- Booking
- A planned block of time for a person on a project or activity.
- Capacity planning
- Comparing available capacity with expected demand over a horizon to decide on hiring, sales and scheduling.
- Confirmed booking
- A booking for signed or approved work that people should plan around.
- Contracted capacity
- Hours in employment contracts for a period, before any time off.
- Demand
- Hours of work needed, from confirmed projects, extensions, pipeline deals and internal initiatives.
- Estimate variance
- The difference between actual and estimated effort, as a share of the estimate.
- FTE (full-time equivalent)
- Contracted hours divided by the full-time standard; 0.6 FTE is 24 hours on a 40-hour week.
- Forecast gap
- Available capacity minus confirmed plus probability-weighted demand.
- Non-billable hours
- Work that is not invoiced: internal projects, presales, training, admin, support rotas.
- Over-allocation
- Booking a person for more hours than they have available in a period.
- Planning fallacy
- The tendency to underestimate how long one's own tasks will take, even with experience of past overruns.
- Productive utilization
- Billable plus approved non-billable project hours, divided by available hours.
- Realization rate
- Value actually billed divided by the value of billable hours at standard rates.
- Reference class forecasting
- Estimating by adjusting a bottom-up estimate with the actual outcomes of similar past projects.
- Resource leveling
- Changing project dates so that demand fits available capacity, accepting a later finish if needed.
- Resource smoothing
- Re-distributing work within existing slack so the end date does not move.
- Roll-off
- The date a person finishes their current project and becomes available.
- Scheduled utilization
- Booked hours divided by available hours for a future period.
- Skills matrix
- A table of people against skills with proficiency levels, used for allocation and coverage.
- Tentative booking
- A provisional booking for likely but unsigned work, kept visibly separate from confirmed work.
- Timesheet
- A record of hours a person worked, by project and activity, usually approved weekly.
12Chapter 12 of 12
Frequently asked questions
Short answers to the questions project and resource managers ask most often.
What is resource management in project management?
It is the ongoing practice of matching people (supply) with project work (demand) over time: knowing who is available, booking them on the right work at a sustainable load, forecasting future demand and comparing planned with actual hours to improve the next plan.
What is a good utilization rate?
It depends on the role. Delivery roles in services firms are commonly planned at about 75 to 85% billable utilization of available hours; leads, project managers and managers lower, because part of their job is non-billable by design. Aim for a band per role, never 100%.
Why should we not target 100% utilization?
Because knowledge work is variable. Queueing models show waiting time growing with u ÷ (1 − u): at 90% utilization the wait is nine times longer than at 50%. Without slack, urgent work queues, overtime becomes normal and quality drops.
What is the difference between capacity planning and resource allocation?
Capacity planning compares total available hours with total demand over weeks or quarters to decide on hiring and sales. Resource allocation picks specific people for specific work in a specific period.
How far ahead should we plan resources?
In layers: day-level bookings for the next two weeks, week-level confirmed and tentative bookings for weeks three to six, and role-level demand against capacity for the next quarter.
How do we include sales pipeline work in the resource plan?
Book likely deals as tentative, record start date, duration and roles, and weight their hours by win probability when you compare demand with capacity. Confirm the booking when the deal signs.
Which resource management KPIs matter most?
Billable and scheduled utilization by role, over-allocation rate, bench capacity, plan accuracy, estimate variance, project margin and overtime. Each should have a formula, an owner and a review cadence.
Do we need time tracking for resource management?
You need actual hours to know whether the plan worked, to calibrate estimates and to calculate utilization and margins. Time tracking provides them. Keep it lightweight, approve weekly and use it to improve plans, not to police people.
