Treating the brief as source of truth when decks keep contradicting it

A production manager builds the hall plan graphic from slide 14 of the campaign deck, because slide 14 is the thing that was presented and everybody nodded at it. Three weeks later somebody opens the signed brief and finds a different hall split in the deliverables section. Both documents are current. Both were circulated by the same team. One of them has now been printed on 4,000 visitor guides.

The question of the day is which document was right, and there is no answer available, because nobody decided in advance that the brief as source of truth was the rule. The question that will actually get asked in the review is who to blame, which is a worse question and much easier to answer.

One decision, one home

The rule is short. Every decision about the edition lives in exactly one place. That place is the brief. Anything else that states the same decision is a copy, and a copy that disagrees with the original is wrong, whatever order they were written in and however senior the person who presented the copy.

Stated like that it sounds obvious and slightly authoritarian. In practice it is the only version that survives contact with a portfolio, because the alternative is reconciliation, and reconciliation does not scale. Two documents can be reconciled by a careful person in twenty minutes. Eight shows times four artefacts each, at any given moment, cannot be, and the reconciliation work has no natural owner, so it gets done late by whoever is nearest the printer.

Michael Nygard's 2011 note on architecture decision records is the cleanest small template for this outside our industry. He proposes one short record per decision, with a title, the context, the decision itself stated in full sentences and active voice, a status of proposed, accepted, deprecated or superseded, and the consequences including the bad ones. The status field is the part event teams tend to skip and the part that does the work, because it lets a decision be visibly replaced instead of quietly contradicted.

Where did this actually get decided?

Run the trace on three recent scope changes. Pick real ones from the current cycle, and for each ask a single question: where is the moment somebody agreed to this?

Change one, moving the buyer lounge from hall 3 to hall 5. Agreed in a group chat thread on 4 March, in a message that reads "yeah let's do 5", from the operations director. Never written anywhere else. Surfaced when the wayfinding proofs came back wrong.

Change two, dropping the printed pocket guide. Agreed verbally in a Tuesday stand up, typed into slide 9 of the following week's deck as a bullet reading "digital only for 2027". The brief still lists the pocket guide under deliverables, because nobody edits deliverables after sign off.

Change three, adding a second keynote stage. Agreed in an email between the content director and the show director, with the budget holder copied. The brief was updated, eleven days later, by a coordinator who was not on the email and reconstructed the decision from the thread.

Three changes. Zero recorded in the brief at the moment of agreement, one recorded eventually with a reconstruction risk attached, two never recorded at all. If your trace comes back looking different from that, the interesting number is how many of yours started in a chat thread, because that is where speed comes from and it is also where the record is lost.

The cost is countable in this example. The wayfinding reprint is 4,000 guides and two proofing rounds. The pocket guide contradiction produced a print quote that nobody needed. Neither of those failures was a bad decision. Both were good decisions with no home.

The deck problem, stated precisely

Decks are not the enemy. A deck does a job the brief cannot do, which is holding a room's attention for fifteen minutes and making an argument in a sequence.

The trouble is that building a deck requires restating decisions in compressed form, and compression drops the qualifications. The brief says the edition is aimed at plant maintenance managers at sites of 200 people or more, with budget authority to about 50,000 pounds, and explicitly not at group engineering directors. Slide 6 says "targeting maintenance and engineering leadership". Both sentences feel true to the person writing the slide. Only one of them tells a media buyer what to exclude.

Then the deck circulates, gets forwarded, and acquires authority through repetition. Six weeks later the media plan is built from slide 6 and includes a title the brief ruled out.

Bryar and Carr, in Working Backwards, 2021, describe Amazon banning PowerPoint in senior meetings in 2004 and replacing it with written narratives read silently at the start of the meeting. Their argument was about the writing rather than the format: prose forces the author to make the argument, where bullets let an author list a topic and move on. An event brief has the same property. A filled brief section with a compressed slide behind it will lose to the slide every time, because the slide is what people saw.

The workable rule is that decks get generated from the brief and never edited afterwards. If slide 6 needs to say something the brief does not, change the brief first and rebuild the slide. Teams find this irritating for about a month.

What the record has to hold

A brief that wins arguments needs four things per decision, and most templates hold one of them.

The decision, in a sentence with a subject. The date it was made. The person who made it. The status, which is where Nygard's model earns its keep: accepted on 4 February, superseded on 11 March by the hall 5 decision, with the superseded text still visible.

That last property is what makes the brief usable in a dispute. A brief that only ever shows the current answer cannot settle an argument about what was agreed in February, so the argument moves to the email archive, and once it has moved there the brief has lost its status permanently. Keeping the author and the timestamp on every field is a bigger job than it sounds and gets its own treatment in field level provenance.

Three numbers worth knowing about your own setup. How many fields in your current brief have a date on them. How many have a person's name. How many can show what they said before the last edit. In most templates the answers are one, one and zero, and all three refer to the sign off block rather than to the fields.

Why does the brief lose to the deck?

Because the deck is where the attention is, and attention decides which document people treat as real.

The brief gets written once, reviewed by four people over ten days, signed, and then never opened again by anybody except the studio. The deck gets presented to the leadership team, forwarded to the commercial team, shown to the venue, and quoted in three meetings. It accumulates social weight while the brief sits in a folder.

You do not fix that by policy. Writing "the brief is the single source of truth" into a process document changes nothing, because the person building the media plan at 4pm on a Thursday opens whatever is nearest and most recent. What changes behaviour is making the brief the nearest and most recent thing, which means it has to be live, linkable at field level, and visibly newer than the deck. A brief held in a shared brief and review workspace that shows a last edited timestamp per section clears that bar. A brief attached to an email from February does not, and no amount of insisting will make it.

The second lever is small and effective. When somebody agrees something in a chat thread, reply in the thread with a link to the field you just updated. It takes forty seconds, it teaches the group where decisions live, and it converts the chat from a competing record into a pointer at the real one. Do it for six weeks and people start asking for the link before they ask for the answer, which is roughly what the template is trying to achieve in the first place.

Where this stops

Making the brief authoritative works for decisions. It works badly for everything else, and teams that push it too far end up with a brief nobody can read.

Discussion belongs in the thread. Rationale belongs in the record, briefly. The long argument about whether the retrofit segment is worth chasing does not belong anywhere in the brief, and trying to capture it produces a twelve page document whose decision sections are buried. The discipline is to write the decision and one sentence of context, and to leave the transcript where it happened. Where the closing comment on a review thread carries the reasoning that changed a field, that closure needs to be captured deliberately, which is a review workflow problem and a genuinely hard one.

The honest limit is that none of this survives a leadership team that decides things in rooms without writing them down. If the show director agrees a change verbally with the group MD and nobody types it anywhere, the brief will be confidently wrong and every downstream artefact will inherit the error with perfect fidelity. Source of truth is a claim about where the record lives. It is not a claim that a record exists.

This week, take three scope changes from your current cycle and trace each to the moment of agreement. Write down where each one actually happened. If none of the three happened in the brief, you have your answer about which document your team treats as real.