The question sounds accusatory, and it is almost never meant that way. It gets asked after something has already gone out, once a claim inside it turns out to be wrong or contested, and whoever has to respond needs to know what the team actually knew at the time. Most marketing teams can answer it eventually. Very few can answer it in an hour, and the distance between those two things is where the cost lives.
Approval and Agreement Are Not the Same Thing
Plenty of work ships with everybody on side and nothing written down. A reaction emoji on a thread, a nod in a standup, a reply that says looks good to me with no indication of what “it” was at that moment. All of that is agreement. None of it is an approval record, and the difference only becomes visible when somebody outside the team asks a question the team cannot reconstruct.
A usable record answers four things: who signed off, on which specific version, on what date, and against what evidence. Drop any one of the four and the rest stops being useful. Knowing that legal approved something is worth very little if nobody can say which draft legal was looking at.
Version is the field that goes missing most often. Approval usually happens on a near-final draft, then somebody tightens a headline, swaps a statistic for a fresher one, or trims a qualifier to make a caption fit. Each of those edits is small and reasonable. Collectively they mean the thing that shipped is not the thing that was cleared, and no one involved would describe that as a process failure. Six months later, when the wording is being read back to somebody, that distinction is the entire argument.
The Industries That Had to Solve This First
Heavily regulated sectors worked this out because they had no choice. In pharmaceutical marketing, promotional material moves through medical, legal and regulatory reviewers before it goes anywhere, and the point of that MLR process pharma teams run is not simply catching errors. It is producing a durable account of who accepted which claim, on what evidence, at what point in time, so the question can be answered years afterwards by people who were not in the room.
That structure looks heavy from outside, and for a mailing list campaign it probably is. The underlying idea travels well, though: separate reviewers looking at defined risks, a fixed version under review, and a record that survives the people who created it.
When the Record Has to Leave the Building
In some sectors the approval file does not stay internal. Under United States rules, once a prescription drug is approved, promotional material has to be filed with the regulator at the point it is first disseminated, using a specific transmittal form. Since June 2021 those submissions have had to be made electronically, following final guidance the agency issued in June 2019 that replaced an earlier 2015 draft. What the team assembled internally is also, in part, what reaches the regulator.
Most marketing teams will never file anything with anybody. The structural lesson still holds: an approval record built only for internal comfort tends to be organised around reassurance, while one built on the assumption that an outsider may read it gets organised around evidence. The second kind is more work upfront and dramatically less work later.
Approvals Happen Where the Tools Are Not
The practical obstacle is rarely that teams reject the idea of record keeping. It is that approval happens wherever the conversation is happening, and the conversation has moved. Comments live in a design tool, feedback arrives in a chat channel, the final yes comes over a direct message, and the asset management system records only that a file was uploaded. Collaborative platforms that keep tracking versionsinside a shared workspace close part of this gap, because the edit history and the review sit in the same place rather than in three disconnected systems.
That still leaves the conversation itself. A retention policy written for documents rarely covers the messaging platform where the decision was actually made, and the platform’s own default is often deletion after a fixed window. The approval survives; the reasoning behind it quietly does not. Anyone auditing the trail afterwards finds a decision with no visible basis, which reads worse than a decision that was never formally recorded at all.
The Evidence Has to Exist Before Publication, Not After
There is a detail in advertising regulation that reframes the whole exercise. The UK code requires that marketers hold documentary evidence for objective claims before distributing or submitting a communication for publication. Not when challenged. Before. Regulators also assess how a claim is likely to be understood rather than what the marketer intended it to mean, which means the file needs to back up claimsas an audience would read them, not as the team meant them.
That turns substantiation into part of the approval package rather than something assembled defensively afterwards. If the evidence has to exist first anyway, attaching it to the sign-off costs almost nothing, and it removes the scramble where three people try to remember which report a percentage came from.
Start With the Campaign You Would Least Like to Explain
There is no need to build a review structure for everything. Sort output by consequence instead: material making measurable claims, anything touching a regulated category, anything under a named executive’s byline, anything that would be expensive to retract. Those get a recorded approval with a version reference and an evidence file. Social posts about office renovations do not. Drawing that line explicitly matters more than where exactly it falls, because an unwritten rule about which work needs sign-off gets interpreted differently by every person who joins the team.
The test for whether it is working is unglamorous. Pick something published six months ago, ask who approved it, and see how long the answer takes. If it takes a week, the process exists on paper only. If it takes ten minutes, the question stops being uncomfortable and becomes what it always should have been, which is a request for information rather than the opening line of an investigation.
