guide · 2 Sep 2026
How to build a weekly steering committee pack from Excel
Most firms of 10 to 500 people run their portfolio on 3 spreadsheets and rebuild the same pack by hand every week. The procedure below is the one that survives a partner asking where a number came from.
What the pack has to answer
A steering committee pack is not a status update. It exists so that 6 questions get answered with numbers somebody will defend.
- 01Is the portfolio making money, and how much of that figure do we trust?
- 02Which projects moved since the last cycle, and which moved the wrong way?
- 03Who is overbooked, and on which projects at the same time?
- 04Which client commitment is at risk in the next 30 days?
- 05What decision does this committee have to take today?
- 06Which numbers are we knowingly reporting without proof?
Questions 01, 03 and 06 cannot be answered from one file. They only appear when the staffing file, the budget and the tracker are put next to each other, which is exactly the work that gets skipped when the meeting is in 2 hours.
The 3 sources, and what each one hides
staffing_2026_v3_FINAL_(2).xlsx
Gives you: who is booked, on what, at what percentage, in which week.
Hides: the same consultant under 2 spellings, roll-offs left at 0%, a rate table on a tab marked DO NOT EDIT.
budget_Q1_v2.xlsx
Gives you: sold days, sold revenue, cost to date, forecast cost, margin.
Hides: a TOTAL row produced by a formula that broke 3 versions ago, and an old FX rate on a second tab.
project tracker mars.xlsx
Gives you: status, owner, deadline, risk level, the comment that matters.
Hides: a copied tab that still shows last month, and project names spelled differently from the budget.
The 4 failures that make a hand-built pack wrong
A pack is rarely wrong because someone cannot do arithmetic. It is wrong because of 4 specific failures, and all 4 happen silently.
1Identity drift
Sarah Laurent in the staffing file and S. Laurent in the same file are one person. Counted separately, she is 80% and 50%. Counted once, she is booked at 130% and the committee has a staffing decision it did not know about.
2Dead tabs treated as data
A tab called Staffing (old), a tab called copie de Tracker, a tab called Sheet7 with 0 rows. Each one is a plausible source of truth to anyone who did not build the file, and each one will produce a number that contradicts the pack.
3Totals inherited from broken formulas
When the TOTAL row of a budget shows #REF!, the row above it is still copied into the deck, or the total is retyped from memory. Portfolio revenue and portfolio margin then come from a cell nobody can reproduce.
4Reported margin instead of recomputed margin
A margin column written last quarter is not a margin. Recompute it from sold revenue and forecast cost present in the file this week, and state which projects had to be recomputed.
A build order that holds up
7 steps, in this order. Steps 03 and 04 are where the pack is actually made, and they are the 2 that get compressed when time runs out.
- 01Freeze the sources. Copy the 3 files into one folder named for the cycle, and never edit them again. Everything downstream refers to those file names, verbatim.
- 02List every tab in every file with its row count, then mark the ones you will ignore and why. This list is the pack's audit trail, and it takes 5 minutes.
- 03Build one project list. Reconcile names across files by hand, and write down each merge: ERP Migration and Migration ERP are one project, and phase 2 is not started.
- 04Build one people list the same way, merging spellings, and sum each person's allocation across all projects. Anything above 100% is a finding, not a rounding problem.
- 05Recompute margin per project from the numbers present this week, then recompute the portfolio margin from those results. Never carry a total that a broken formula produced.
- 06Write the findings. One flat sentence each, naming the file it came from. A finding is a resolved reconciliation or a flagged problem, not an opinion.
- 07Write what you would not guess. Unsigned phase, scope widened without a change order, a stale FX rate: name each one and say what you need to resolve it.
One habit changes the pack more than any template: write the merges down. Sarah Laurent and S. Laurent merged into one row is a decision, and a committee is entitled to see it.
The checklist before the meeting
- ✓Every figure in the pack traces to a file name and a tab you can open in front of the committee.
- ✓No total in the pack was copied from a cell containing an error value.
- ✓Every person appears exactly once, and every allocation above 100% is stated as a finding.
- ✓Every project name is the same string in the portfolio table, the allocation table and the findings.
- ✓Negative margins are shown with a minus sign and one decimal, not in parentheses and not in red alone.
- ✓The pack says which sources it read, and which tabs it ignored.
- ✓The pack contains a list of what it could not resolve, and that list is not empty by accident.
The last line is the one that earns trust. A pack that names what it could not resolve is read as complete, and a pack with no unresolved list is read as confident until the first question lands.
Or drop the 3 files
This is the procedure Operations Cell runs. You drop the staffing file, the budget and the tracker exactly as they are, a model reads them, and the pack comes back with the portfolio, the allocation, the findings and the list of what it refused to guess.
Read a real pack before you trust it with client numbers.
The example is a real run of the engine on an invented firm whose 3 files are as broken as yours: 2 dead tabs, a #REF! in the budget total, and one consultant spelled 2 ways.