Capacity planning: knowing what you can actually take on

Capacity planning is working out how much work your team can take on in a period, and comparing that against what is already committed. It sounds like something only large operations need, and the opposite is true: a company of twelve overcommits far more easily than one of two hundred, because nobody is doing the arithmetic and everyone is optimistic in the same direction.

The symptom is familiar. Every project is agreed on the assumption that the others will finish on time, deadlines slip in a chain, and the team works late without anyone having decided that they should. None of that is a motivation problem — it is a planning gap that a single sheet resolves.

What a capacity model needs

• Available hours per person per week, after holidays, and after subtracting the hours that are not project work.
• A realistic utilisation assumption. Nobody delivers forty billable hours in a forty-hour week — meetings, admin, support and context switching take a large share, and pretending otherwise is the main cause of overcommitment.
• Committed work by period: what is already promised, with its estimated hours.
• Skills or roles where they are not interchangeable — total hours mean nothing if the only person who can do the work is one of them.
• Absence: booked holiday, known leave, and a small allowance for the unknown.
• The resulting free capacity per period, which is the number the whole model exists to produce.
• A pipeline column for probable work, weighted, so you can see the squeeze before it arrives rather than during it.

Choose a utilisation figure and be honest about it. Teams that plan at 100% are planning to be late; most knowledge work sustains somewhere in the region of 60–80% on delivery once meetings, support and overheads are counted, and yours is measurable from your own timesheets rather than guessed. Plan at your real number and the model starts predicting instead of flattering.

Building it

1. Pick a period that matches how you commit — weekly for short work, monthly for projects.
2. List people, their contracted hours, and their realistic utilisation. Roles rather than names if the team is larger.
3. Subtract booked absence and a small allowance for the unbooked kind.
4. Add committed work with estimated hours, and be explicit that these are estimates.
5. Calculate free capacity per period. If it is negative for the next month, you have found something worth acting on today.
6. Add the pipeline, weighted by likelihood, to see where the strain lands if the work you expect actually arrives.
7. Check the skill dimension separately for any role only one or two people can fill — the aggregate can look healthy while a single person is the bottleneck.
8. Review weekly, and update estimates as work progresses rather than only when it finishes.
9. Compare planned to actual once a quarter. The gap tells you whether your estimates or your utilisation figure need adjusting.

What to do when you are over capacity

The model's value is that it forces the choice while options still exist. There are only four: move a deadline, cut scope, add people, or decline the work. Discovering the problem in advance means you can pick; discovering it in the last week means the choice is made for you, usually by whoever shouts loudest. Small companies underuse the fourth option — saying no to work you cannot deliver protects the work you already promised, and the alternative is delivering everything slightly late.

Ettex Sheets is where the model belongs: import an existing XLSX with formulas intact, formulas for available hours and free capacity, dropdown validation for roles and statuses, conditional formatting to flag any period where capacity goes negative, charts for the capacity curve, pivots by person or by project, and cell comments so a manager can query an estimate against the cell it sits in. Version history means last month's plan is still recoverable when you compare planned to actual. The work itself lives in Ettex Board with columns, assignees, due dates and WIP limits, longer projects can carry dates in Ettex Calendar, and the commitments you are weighing sit in Ettex CRM as a pipeline.

Plainly: Ettex does not do resource management. There is no capacity engine, no allocation of people to tasks with automatic conflict detection, no timesheet feed to measure real utilisation, and no scheduling tool that levels workload for you. The model is a spreadsheet you maintain from your own numbers. That is proportionate for a team of a few dozen; agencies scheduling many people across many projects want dedicated resource software. Where the constraint is people rather than machines, the same numbers drive employee scheduling software decisions.

Planning capacity in months is a different job from sequencing jobs on Tuesday. Production scheduling software works at the second scale, with finite capacity, sequence-dependent changeovers and tooling treated as a resource.

Why capacity plans mislead

• Planned at full utilisation, which guarantees the plan is wrong before it starts.
• Absence ignored, so August and December are quietly overcommitted every year.
• Aggregate hours only, hiding the one person everything depends on.
• Estimates never revised as work progresses, so the model describes last month.
• Pipeline excluded, so the squeeze appears as a surprise rather than a forecast.
• Built once for a plan and never used for a decision.
• Planned versus actual never compared, so the same optimistic estimates repeat indefinitely.

On a shop floor the planning becomes a sequence rather than a number. A production schedule turns the capacity figure into an order of work on named resources, and it is where an optimistic capacity assumption gets caught — usually by a changeover time nobody measured.

Capacity planning is available hours, honestly adjusted, against work already promised. Plan at your real utilisation, model the specialist you depend on, and look at the number before you agree the next deadline.