An apology is not a creative process.
In the past few days, Converse pulled a campaign image, apologized publicly and, according to an internal memo reported by Bloomberg, started looking at how creative work gets through its own review. The CEO’s line, in substance: these images should not have been used, and leadership is taking a hard look at the critical lens applied to the work before it goes out.
That sentence is the story for communications leaders. Not the outrage cycle. Not the pile-on. The admission that an asset left the building after legal, marketing and leadership had all touched it.
If you run corporate communications, brand safety or an executive’s reputation, you know the pattern. A campaign goes out. The internet sees an association the room never saw. The brand pulls it, apologizes, promises to do better. Then the same architecture, same reviewers, same calendar pressure, same trust in “legal signed off”, waits for the next asset.
This piece takes that architecture apart, with Converse as the clearest public example of the week. The lesson is not “avoid sensitive campaigns”. The lesson is: pre-publish judgment is a communications control, not a production courtesy.
What actually happened
In mid-September 2026, an image from the Converse Chuck 70 X campaign with Karina, of the group aespa, drew intense criticism: for many viewers, the lighting and framing created visual associations nobody had intended. Converse removed the image from its channels and apologized, saying it understood why the image was deeply upsetting, recognized it had got this wrong, and was working to remove it everywhere it had appeared.
On September 20, Bloomberg reported an internal memo from CEO Aaron Cain to employees: “The leadership team and I are taking a hard look at our internal review process and the critical lens we apply to our work.” The brand has since faced pressure for accountability beyond the apology and the takedown: who created and approved the campaign, and which safeguards will change.
The exact internal workflow is not public. What is public, and what matters for every other company, is the question asked from outside, again and again: how did this get through every gate?
That question is the job of communications.
Why this is not a monitoring problem
Most crisis stacks are built to listen after publication: social listening, media monitoring, escalation playbooks, night desks. Those tools matter once a story is already moving. We described the three layers in Crisis communication software: monitoring vs response vs pre-publish review.
They do not catch the frame that should never have shipped.
By the time monitoring lights up, you are already in response mode: delete, apologize, brief the executives, reassure partners, absorb the secondary coverage. The reputational cost is paid. The process admission arrives after the asset has done its damage.
Pre-publish review sits one layer earlier. It asks a different question: what can a hostile audience, or simply one trained to spot patterns, see in this image, this headline or this caption that our production room no longer sees?
That is not the same skill as legal clearance. Legal can confirm a claim is substantiated and an image right is in order. Reputational judgment asks whether the composition creates associations the brand cannot defend.
The Converse story shows exactly the gap we named in that guide: monitoring will not catch the sentence, or the silhouette, you publish.
The three places where creative review breaks
1. Room blindness
The people who briefed, shot, retouched and approved an asset have already decided what it means. They see the concept. The public sees the pixels.
Lighting, framing, color, gesture, negative space: all of it creates associations that were never in the brief. A room that has stared at the same file for three weeks will not invent the reading a stranger invents in three seconds.
The fix: add at least one reviewer who was not in the concept meetings, and give them only the final asset, with no backstory.
2. “Legal approved” is not “brand safe”
Legal asks: can we say this? Can we show this person? Are we exposed on IP or defamation?
Reputation asks: if someone screenshots it, crops it and shares it without our caption, what does it look like? Who is hurt by that reading? Can leadership defend it on camera tomorrow?
Those are different gates. Merging them into one sign-off is how “it passed legal” becomes the first line of a crisis memo.
The fix: make reputation review a separate control with a named owner, not a checkbox inside the legal workflow.
3. Speed, and celebrity campaigns, compress the gate
Celebrity and culture campaigns move under pressure: drop dates, embargo windows, influencer schedules, paid media already bought. The review that would have caught the association shrinks to a Slack thread and a “looks good to me”.
That is exactly the moment the gate should slow down, not speed up. High-blast-radius assets deserve more adversarial review, not less.
The fix: define a “high blast radius” tier (celebrity, paid social, out-of-home, anything adjacent to the CEO or to a live crisis) that cannot skip adversarial review, however loud the calendar gets.
The pre-publish checklist
Run this on the next campaign asset before it leaves staging:
- The hostile crop test. Take the hero frame. Crop the most ambiguous area. Ask three people outside the campaign: what does this look like, with no caption?
- The association scan. List the historical, cultural or graphic associations a bad-faith reader could invoke. You are not endorsing those readings. You are stress-testing your ability to survive them.
- The channel mismatch. Does the asset change meaning in dark mode, in a mobile crop, in muted video? Many failures are format failures, not concept failures.
- Named owners. Who owns creative? Who owns legal? Who owns reputation? If two of those are the same person on a deadline, you do not have three gates.
- Kill authority. Who can stop publication without convening a committee? If nobody, you have process theater, not a control.
- Post-mortems without mythology. When something goes wrong, record which gate failed: concept, production, legal, reputation or an executive override. So that “do better” names a precise step.
None of this requires a new agency. It requires treating publication as a controlled release, not the end of a creative project.
What “doing better” looks like
Every brand in this position says it will do better. Communications leaders have to translate that into operational changes:
- Separate reputation review from legal review on high-blast-radius assets.
- Add adversarial eyes that never sat in the concept room.
- Log decisions. Who approved, when, against which checklist. After a failure you need a record, not folklore.
- Train the gate, not only the crisis cell. Most companies rehearse holding statements. Far fewer rehearse the moment someone says “stop the line” in front of an asset.
- Hold partners and co-brands to the same standard. Shared logos, shared blast radius. A partner approval does not dilute responsibility, it multiplies it.
If your only upgrade after a public failure is a longer apology template, you upgraded the statement, not the process.
What Cadence does, and does not do, at this gate
Cadence works on the moment before publication, and on the words: the headline, the caption, the press release, the CEO post, the press reply. It reads them the way a journalist, a lawyer, an employee and your customers will, checks them against everything the company has already said, and proposes a safer version. A human always keeps the decision to publish.
It does not judge an image. The hostile crop test, the association scan, the kill authority: that is your human gate, and it is precisely the one Converse says it is re-examining. What Cadence adds is that the text around the visual does not contradict a past statement, and does not say what leadership will not be able to defend tomorrow.
When a CEO memo talks about “the critical lens we apply to our work”, that is the gate it is talking about. It gets built before the next drop, not after.
What to do next
If your team can monitor a crisis but cannot reliably stop a bad asset, the next cycle is already scheduled.
This week: run the six-point checklist on your next high-blast-radius campaign. For the words around it, paste the post or the press release: twenty seconds, no account. And Popeye can train the team on the “stop the line” moment, before the day it matters.
Sources: Bloomberg, September 20, 2026, on Aaron Cain’s memo; Snopes, on the takedown and Converse’s apology.