Lead with the number that disappoints
When you report a partly disappointing result to someone who could not watch the work, the shortfall goes in the first paragraph with its real number, and every headline figure carries the method that produced it.
Use when
- Use this when you are writing the closing message for work that mostly succeeded but missed one target, and you are deciding how prominently to say so.
- Use this when a figure you promised came in under its target and only you can say how that figure was produced, because nobody watched the work being done.
- Use this when the brief named in advance the deliverable most likely to be quietly under-delivered, and you are the one now answering that clause.
The failure
The shortfall sits in paragraph nine, or a padded figure stands in for the real one — a count reported at target while part of it was filled in after the source call failed. Nobody was watching, so nothing catches it at the time. The reader accepts the work believing it complete, and the gap between what they think they have and what they have is found later by a stranger.
The outcome
A report whose opening paragraph names what could not pass and why, whose headline figures each carry the method that produced them, and whose failing checks appear exactly as the tool printed them with an owner beside each red line.
The Practice — 5 rules
Report the number most likely to be padded, with its method
Before you write, name the one deliverable most likely to be quietly under-delivered — usually a count, a coverage figure, the number someone will glance at and move on from. If the brief named that risk in advance, answer it by name rather than leaving the reader to hunt for your answer. Report the real figure together with how it was produced: a list reported at 1039 entries is stated as 75 hand-curated entries plus 995 fetched from a public index, deduplicated — so the reader can see which part came from where. If the figure fell short of its target, the real number stands and the shortfall is explained beside it.
Why The reader could not watch the work, so a number without its method cannot be audited by anyone but you — which makes padding free at the moment it happens and ruinous when a stranger finds it.
Lead with what could not pass
The first paragraph names the item that did not pass and why it did not. Not the second paragraph, not a section near the end, and not as an abstract count of things outstanding — name them. Write it flat: the item, its state, the reason. Then read it back and take out the softening words before you send.
Why A shortfall placed after the accomplishments is read as a footnote by someone who has already decided the work is done.
Separate the substance from the blocked last hop
Where the work is built and tested but its final step cannot run yet — the notification that writes to a log instead of sending, the intake that is still an email address rather than a form — say both halves in one sentence: what is proven, and exactly where it stops. A stubbed step never sits unmarked inside a list of finished items.
Why A reader who cannot tell a finished mechanism with an unwired last step from unfinished work stops trusting both claims.
Run the check that will fail, and paste it verbatim
Where an external check is known to fail today, run it against the real target anyway and paste the output exactly as printed, failures included, with an owner named beside each failing line. Do not annotate inside the block: every caveat of yours goes in the prose beside it, so what you forward stays the record rather than your edit of the record. A run that reads twelve of fourteen green, the two failures owned by environment and by deployment, is worth more than a summary that leaves them out.
Why Verbatim output with owners turns a disappointment into a work list, while a paraphrase turns it into a mood — and a record you edited on the way past has stopped being evidence.
Do not state a total you have not recounted
Any tally, subtotal, or claim about what a given check covers must be recomputed from the pasted output before it goes into your prose. If you cannot recount it, give the raw result and leave the total out. Never describe the scope of a check you did not read.
Why The report earns its credibility from the parts a reader can verify against the block below, so a slipped total is the one sentence they check first and the reason they start doubting the honest parts.
Limits — do not use this when
- Do not use this for an interim update while the work is still running. A step not yet reached is not a shortfall, and leading with it turns a progress note into a false alarm.
- Do not use this when what fell short is a check that could not run rather than a result that missed its target. An absent check belongs in the closing record of the session, named against the brief that asked for it.
- This governs how a result is reported, not whether the result is acceptable. A report can be entirely honest about work that should be rejected.
- Naming a shortfall does not discharge it. An item reported honestly is still open until someone owns it, and assigning that owner belongs to the closing record's triage rather than to the opening paragraph.
- Leading with the failure is a rule for reports of finished work, not for every message you send.
Verify — evidence that it helped
- The first paragraph of the message you actually sent names the item that could not pass.
- Every headline figure can be traced to how it was produced by someone who was not there and cannot ask you.
- If the brief named a deliverable at risk of being padded, the report answers that clause by name.
- The output pasted for a failing check matches, character for character, what the tool printed, with no annotation inside the block.
- Every failing line in the pasted output carries a named owner.
- Any total or tally stated in the prose can be recomputed from the pasted output and comes out the same.