A definition of done checklist for each section of an event brief
A definition of done checklist gives every brief section a short list of objective items that must hold before it can be approved. Three to five per section works. Each item has to be answerable yes or no by someone who did not write the section, which rules out wording like clear, compelling or sufficiently detailed.
Somebody asks the designer to start on the visual identity for the autumn edition. She opens the brief, goes to the audience section, and finds two paragraphs describing a confident, ambitious buyer who values quality and peer connection. The section is marked approved. She cannot start.
A definition of done checklist is the thing missing there. Not a longer section, not a better writer, and not another round of review. A short list of objective items that had to hold before anybody was allowed to mark that section approved, and which would have failed it in about fifteen seconds.
What approved currently means
On most teams, approved means a person with authority read the section and did not object. That is a real signal and it is worth something, but it varies with who read it, how busy they were, and whether the section arrived on a Thursday.
The consequence shows up downstream rather than at the moment of approval. Design cannot start. Sales quotes a positioning line the marketing lead does not recognise. Two shows in the same portfolio describe the same buyer in incompatible language and nobody notices until a joint campaign goes out.
None of that is a review quality problem. It is the absence of a written standard against which review happens, which is what a definition of done supplies.
What goes on the list
Three to five items per section. Every item answerable yes or no by a person who did not write the section, holding nothing but the section and the item.
Take the audience section. The items are: named segments are listed and each one appears in the show's own segment taxonomy; a size estimate is given for each segment; the source of each size estimate is named and reachable; at least one segment is marked as the priority for this edition.
Four items. All four are presence or source questions. None requires anyone to decide whether the audience is well defined, because that is the question two reviewers will answer differently and neither will be able to show the other they are wrong.
Compare that with what teams usually write. Audience is clearly defined. Segments are relevant and prioritised. The section provides sufficient detail for downstream teams. Those are aspirations. They read like standards and they cannot fail anything.
Do the same for the rest. The positioning section: a single sentence claim exists; the claim states something the competing show cannot say; a named person owns the claim. The pricing section: every rate card line has a number; the number differs from last edition or the section says explicitly that it does not; a source for any change is named.
A useful test when you are drafting an item is to imagine two reviewers arguing about it. If the argument can be settled by pointing at the section, the item is objective. If it can only be settled by seniority, throw the item away and write a different one. That test kills most of what teams instinctively want to put on the list, and killing it is the point.
The other habit worth adopting is to write the items as things that are present rather than things that are absent. Named segments are listed reads better than no segment is missing, because a reviewer can confirm presence in a glance and cannot confirm the absence of something they have not thought of. Presence items also survive being read quickly, which is how they will be read in the last fortnight before an edition goes live.
Why does a checklist work at all?
It works because it converts a memory task into a reading task, and people are much better at the second one.
The strongest published evidence for that sits in medicine. Haynes and colleagues reported in the New England Journal of Medicine in 2009 on a nineteen item surgical safety checklist introduced across eight hospitals in eight cities, chosen to span a wide range of economic settings. They collected outcomes on 3,733 consecutively enrolled patients before the checklist and 3,955 after. The rate of death went from 1.5 per cent to 0.8 per cent, and inpatient complications went from 11.0 per cent to 7.0 per cent.
Two details of that study are the ones worth carrying across. The checklist was nineteen items for an entire operation, which is short. And the items were things like confirming the patient's identity and confirming that antibiotics had been given, which is to say they were presence checks on facts, and none of them asked the surgeon whether the plan was good.
An event brief is not surgery and the stakes are not comparable. What transfers is the shape of the instrument: short, objective, applied at a defined moment, and aimed at the omissions rather than at the judgement.
How checklists fail
They fail by not being used, which is a different problem from being wrong.
Degani and Wiener's NASA contractor report on flight deck checklists, CR-177549, published in 1991, makes the point plainly in its opening: improper use, or the non-use, of the normal checklist by flight crews is often cited as the probable cause in accident investigations. The checklists in those cockpits were correct. Nobody ran them.
The event equivalent is a checklist that lives in a wiki page nobody opens. If the items are not rendered next to the section at the moment of approval, with a tick box each, the checklist is a document rather than a control. Where the items live matters more than how good they are, and that is a decision about your brief and publishing workspace rather than a decision about the wording.
The second failure mode is length. A twelve section brief with eight items per section is 96 checks per approval. Nobody does 96 checks properly, so people start ticking in blocks, and a checklist that is ticked in blocks tells you nothing at all. Four items per section across twelve sections is 48 checks, which a reviewer will actually work through.
Run it against something already approved
The diagnostic is the interesting part, and it takes an hour.
Take twenty sections from briefs your team has already signed off, ideally spread across three or four shows. Apply the new checklist to each one, four items apiece, so eighty checks in total.
Suppose 23 of the 80 fail. That is a 28.8 per cent item failure rate on work that is formally approved. More useful than the item rate is the section rate: count how many of the twenty sections fail at least one item. If it is 14, then 70 per cent of your approved sections would not pass the standard you are about to adopt.
Now split the failures by item. If 9 of the 23 are the same item, missing source for a size estimate, that one item is carrying most of your defect load and you have found where to spend the next fortnight. If the failures are spread evenly across all four, the sections are broadly thin rather than specifically broken, and the fix is upstream in how the brief gets filled rather than in review.
Do not retro-fix the twenty. The point of the exercise is calibration. If your checklist fails 95 per cent of existing sections, it is describing a future state and adopting it will stall everything for a quarter. If it fails 5 per cent, it is not catching anything and you have written a formality.
What does the checklist cost?
Adding 48 checks to an approval adds real minutes. Call it eight minutes per brief for a reviewer who is reading properly, and on a portfolio publishing thirty editions a year that is four hours of reviewer time annually, which is nothing.
The time comes back at handoff. A design team that can start on the day the brief lands, rather than sending three clarifying questions and waiting two days for answers, saves the calendar days that actually cost you. One avoided round trip per edition, at two days each, is sixty days across thirty editions.
Those numbers are illustrative and yours will differ. The shape of them will not. Checklists cost minutes at the point of approval and save days at the point of use, which is why the person who benefits is never the person paying.
Where this stops
A definition of done checklist catches omissions and cannot catch wrongness. A section can list four named segments with sizes and sources and pass every item while being about the wrong buyer entirely, and the checklist will wave it through with a clean sheet.
It is also poorly suited to the sections that carry the hardest thinking. Positioning either says something or it does not, and the item that tests whether the claim distinguishes this show from the one down the road requires a judgement no tick box will make for you. On those sections, keep the checklist to the mechanical items and accept that the substance still rests on the reviewer.
The last limit is that the checklist depends on somebody maintaining it. Segment taxonomies change, rate card structures change, and an item that points at a list which no longer exists will be ticked anyway because the reviewer assumes it is fine. Put a review date on the checklist itself and treat a stale item as a defect, in the same way a stale status on a section is a defect.
This week, write the four items for one section only. Pick audience, because it is the section most downstream teams are blocked by, and because it is the easiest to make objective. Then run those four items against ten already approved audience sections and count the failures. You will know within an hour whether the checklist is worth extending to the other twelve sections, and you will have a number to show the person who decides. That number is also the honest starting point for scoring a brief before any work begins.
Questions people ask about definition of done checklist
- What makes a checklist item objective rather than subjective?
- A second person, reading only the section and the item, reaches the same verdict as the first. Named segments present is objective. Audience clearly defined is not, because two reviewers will disagree and neither can show the other they are wrong. If an item needs a judgement call, rewrite it as a question about presence, source or count.
- How many items should a definition of done checklist have per section?
- Three to five. Below three the checklist stops catching anything useful. Above five, reviewers begin skipping items and the pass rate becomes meaningless, because you cannot tell a genuine pass from a fast one. Keep the total across a twelve section brief under sixty checks, which is roughly what a person will do properly.
- Should the checklist be applied to sections that are already approved?
- Yes, once, as a diagnostic. Take twenty sections your team has already signed off and run them through the new checklist. The failure rate tells you what approved has meant in practice. It also tells you whether the checklist is calibrated, because a checklist that fails everything is describing an aspiration.
Related reading
- Section status definitions that stop draft and in review meaning the same thing
- Percent complete reporting lies about briefs and here is the arithmetic
- A brief quality score you can apply before work starts