Reporting mistakes that can derail your plans

A recurring report is an obligation, not a project. It goes out every period whether or not anyone thought about it, which means its faults are permanent until somebody deliberately removes them.

The nine below are faults of the obligation rather than of the arithmetic. Each one survives a perfectly correct set of numbers.

What to take away

• Almost every fault here started as a reasonable accommodation made once, under time pressure, and then inherited by everyone after.
• The pack is a product with users. Nobody measures it that way, so it degrades in the one direction nobody watches.
• The cheap fixes are subtraction: fewer metrics, fewer recipients, fewer comments, one artifact instead of two.

1. Publishing on a date rather than after a close

The calendar says the fifth working day, so the pack goes out on the fifth working day. Some sources have not settled, and the numbers move for a week afterwards.

The cost is that readers learn the figures are provisional and start waiting, which means the pack has a publication date nobody uses and a real one nobody states. Statistical bodies solve this by announcing the release in advance and holding to it, with the quality of the data stated on the page: the Code of Practice for Statistics is built around that separation between orderly release and honest quality reporting.

The fix is to define the close in writing, publish after it, and label anything issued before it as a flash with its own name. Two products beat one product with two meanings.

2. Changing the shape of the pack every period

Somebody adds a chart, someone else reorders the sections, a new metric gets a page. None of it is unreasonable in isolation.

What it costs is the reader's ability to skim. A pack whose layout is identical every period can be read in two minutes because the eye knows where to land. A pack that moves has to be read from the top every time, so it gets read less each period until it is only opened when something has gone wrong.

The fix is to freeze the layout and version it deliberately, on an announced date, with the previous shape available for one period.

3. Adding a metric without retiring one

Every addition has a sponsor and every removal has an objector, so packs only grow. A pack that started at four pages arrives at nineteen with nobody having decided that nineteen was right.

The fix is a fixed budget: a new metric enters only when a named existing one leaves. The argument this forces is the useful part, because it is the only moment anyone compares two metrics on the same terms.

4. Writing a comment on every line

It looks diligent. It trains readers to skip the commentary entirely, because most of it says that a number moved slightly and nobody knows why.

The fix is to comment only on movements outside an agreed band, and to say in the pack what the band is. Silence then carries information, which is the whole point.

5. A distribution list that only grows

People are added when they ask, when they join, and when someone is being polite. Nobody is ever removed, because removal feels like a demotion.

The damage is subtle. A large audience makes the writer cautious, so the commentary gets vaguer, and the pack drifts toward saying nothing that could be quoted against anyone. Meanwhile the people who act on it get a document written for spectators.

The fix is an annual re-subscription: the list empties and people opt back in. It is the only removal mechanism that offends nobody.

6. Two artifacts, one truth

A file goes out by email and a live page carries the same metrics. They diverge within a period, because one is a snapshot and the other refreshes.

Then every meeting begins by establishing which version people are looking at, and the reconciliation gets done live, badly, by whoever has the strongest opinion.

The fix is to make one of them authoritative and say so on the face of both. Usually the snapshot wins for a periodic pack, with the live page linked as the drill path.

7. Comparing against a plan that keeps getting restated

Performance is shown against plan. The plan is revised in the second quarter, and the revised plan quietly becomes the comparison for the whole year.

This is cherry picking with no intent behind it: the comparison that survives is the one that makes the current period readable, and the original commitment disappears from the record. A reader six months later cannot tell that anything was ever different.

The fix is to keep the original comparison on the page beside the current one, and to annotate the period when the plan changed.

8. One pack for two audiences

An operating team needs weekly detail. A board needs a short view with context. One document tries to serve both, so it is too long for one reader and too shallow for the other.

The reliable symptom is a pack with an executive summary that nobody in operations reads and thirty pages of detail that nobody on the board opens.

The fix is two products with a shared calculation layer, not one product with a summary page bolted to the front. Guidance on communicating quality, uncertainty and change is written for exactly this problem of one set of figures reaching audiences with different tolerances for detail.

9. Every number owned by the reporting team

A question about a figure goes to whoever produced the pack, who did not define the metric, does not run the process behind it, and cannot explain a movement without asking three people.

The result is a queue in front of one team, and a slow erosion of the idea that anyone in the business owns a number.

The fix is an owner name against each metric in the pack, and a rule that questions go to the owner rather than to the producer. The producer's job is production.

What these have in common

Every one of them is a small kindness extended to a stakeholder that later becomes structure: adding a recipient, keeping a metric, publishing early, softening a comment. None can be reversed by the person who made the accommodation, because reversing it now looks like a decision rather than a default.

So the useful discipline is a scheduled review of the pack as an object, with someone senior enough in the room to remove things. Twice a year is enough.

Related reading on this site

The production and close discipline this sits inside is in building a report worth keeping. The reader-side habits that make a movement questionable rather than assumed are in the review checklist, and the panel-level faults that show up on a screen rather than in a pack are in nine dashboard errors. For choosing what a metric should be before it enters the pack, see metric shapes and their traps, and for the recurring artifacts that quietly became load-bearing, see nine patterns worth studying.

Common questions

Our pack has grown to thirty pages and nobody will cut it. Where do we start?

Ask each recipient which three pages they would keep if the rest were removed. The pages nobody names are not contested, and they can go without a fight. That usually clears a third of it.

Is a flash report worth the extra work?

Only if a decision genuinely happens before the close. If nothing is decided in that window, the flash exists to relieve anxiety, and it costs you the credibility of the final number every time the two differ.

How do we stop the plan comparison from being restated?

Name the original in the pack as a separate row rather than as the default comparison. Nobody will argue for deleting a row that is already there, and its presence makes the restatement visible without anyone having to accuse anyone.

Who should own a metric if the process behind it spans three teams?

One of the three, chosen because they can change the outcome rather than because they hold the most data. Shared ownership is the same as no ownership, and it shows up as a question that circulates for a week.