How to read a board pack you didn't build
A board pack is not a report. It is an argument, assembled by people with something to protect. Read it that way and it becomes far more useful.
Nick Ayton · 13 August 2026 · 6 min read
The short answer: stop reading it as a report. A board pack is an argument — assembled selectively, by people with positions to defend, to support conclusions that were reached before the document was written. That is not corruption. It is human, and it happens in every organisation including well-run ones.
Once you read it as an argument, you start asking the right question about every page: what is this trying to get me to conclude, and who benefits if I do?
Start with what isn't there
The most informative part of any board pack is the omission. Numbers that appeared last quarter and do not appear this quarter. Segments that used to be broken out and are now aggregated. A metric that was the headline six months ago and is now a footnote.
Keep the previous three packs open next to the current one. Nothing exposes curation faster than a side-by-side. People are careful about what they put in and careless about what they quietly remove, because removal feels like tidying rather than hiding.
Then ask directly, in the meeting: we used to report X — where has it gone? The answer is rarely the interesting part. The pause before it is.
The five signals of a curated number
The denominator moved
Percentages are where most polish happens. Win rate improved from 22% to 31% — against what base? If unqualified leads have been excluded from the denominator this quarter, the business has not improved at all. Always ask what the number is a percentage of, and whether that was true last time.
The comparison period shifted
Year on year when year on year flatters, quarter on quarter when it doesn't. If a chart's comparison basis has changed since the last pack, someone chose the basis that told the better story. That is not fraud, but it is a decision, and you should know it was taken.
Aggregation hiding a decline
Three regions reported as one. Two product lines merged into a category. This is the single most effective way to conceal a problem in plain sight, because the aggregate can grow while one component collapses. Ask for the split. If the split is unavailable, that itself is the finding.
A new adjusted measure appeared
Whenever a new "underlying", "adjusted" or "like-for-like" figure enters the pack, something has gone wrong with the unadjusted one. The adjustment may be entirely legitimate. Ask what it removes and why that removal is fair, and ask to see both.
The commentary is doing the heavy lifting
When the narrative is longer than usual and full of context, causes and one-off factors, the numbers are not speaking for themselves. Strong performance is described briefly. Weak performance is explained at length.
Not paranoia. Nobody is trying to defraud you. They are trying to keep their job, protect their team, and avoid being the person who brought bad news into a room where bad news is punished. If that is happening, it is a culture problem you own — not a character flaw in them.
Every number has an author
Take three figures from the pack at random and trace them. Who compiled it? From which system? What was the raw extract before anyone touched it? How many hands did it pass through?
You will find one of two things. Either the chain is short, documented and reproducible — in which case you can trust the pack broadly and get on with the business. Or you will find that a critical number is assembled manually each month by one person in a spreadsheet nobody else has seen, in which case you have found something much more important than whatever the number said.
Do this once, early. Doing it repeatedly makes you look like an auditor rather than a chief executive. Doing it once tells everyone that you might.
The fear in the room
Here is something worth understanding before your next meeting. Everyone around that table shares the same fear — being asked a question about the pack that they cannot answer.
That fear shapes the document more than any strategic intent. It is why numbers get smoothed, why the difficult slide goes near the back, why the commentary pre-empts the obvious question. And it is why the most useful thing you can do is make it safe to bring a problem in early, rather than sharp to be caught out late.
You can be demanding about accuracy and generous about bad news at the same time. Most boards manage the opposite.
Everyone in the room wears a mask, including you. The skill is spotting when it slips.
Build your own set
The permanent fix is not better scepticism. It is having numbers of your own.
Not a parallel finance function — a small set of measures you defined, pulled from source, that describe the actual flow of the business rather than its accounting outcome. How long from customer decision to cleared cash. What proportion of orders go through clean. Where value is sitting still. That is Cash Conversion Velocity, and its great virtue is that it belongs to no function and therefore cannot be defended by one.
Once you have it, the board pack becomes what it should have been all along: useful, and no longer the only thing you know.
Five questions for the meeting
- What changed in how this was calculated since last time?
- What did we report last quarter that isn't here?
- Can I see this split rather than aggregated?
- Who built this number, and from what source?
- What would have to be true for this to be wrong?
Ask them calmly, ask two or three per meeting rather than all five, and never ask them to make a point. The purpose is not to catch anyone. It is to establish, permanently and without drama, that the numbers will be looked at properly from now on.