Nobody at Converse made the pair

Listen to this episode of Social Media Train Wrecks, then use the piece below.

The worst image Converse put out this month is one the company never made. Somebody else built it, out of two posts Converse published eleven days apart, and put them in a single screenshot.

That is the part worth your Friday, and it is the part I got wrong on the show this week.

What actually happened, in order

On September 3, Converse posted a Chuck 70 X image to its own Threads account. The subject is Karina from the K-pop group aespa, the brand's new global ambassador, currently on a world tour. She is standing in a spotlight shaped like the Converse star. One point of that star lands on her long white skirt as a bright triangle. A pair of black high-tops hangs from her hand, arm straight down at her side. The caption read, "Under stage lights or streetlights, the star always finds the frame."

Read cold, on a phone, the white garment and the triangle of light read as a hood and a robe, and the shoes hanging off an extended arm read as a body. Actress Yvette Nicole Brown put it on Threads in four words: "That's a lynching, yes?"

The image sat up for two weeks. Nothing happened.

Then on September 14, a different Converse account in a different country posted a different campaign. Converse Mexico put up the collab with Palmes, a Danish lifestyle brand, on Instagram: a man on a ladder, holding a pair of sneakers overhead.

Four days later the fire started, and it started because fitness creator Kida Maree posted both ads side by side. Her line, quoted in Ryan Joe's writeup at Adweek on September 18: "There is NO REASON why WE should see an advertisement & immediately think of Strange Fruit by Billie Holiday."

Converse, which Nike owns, pulled the Karina image and issued a statement. "We're sorry. We understand why this image is deeply upsetting and recognize that we got this wrong. We removed it from our channels and are working to remove it everywhere it appeared. This should not have happened, and we will do better." By the following Monday the company was repeating that statement to ABC News and declining to say anything else. Civil rights attorney Lee Merritt, in a post quoted in that same piece, aimed at the only sentence that mattered: "Do better means catching this before it airs, not after we're offended." Rep. Troy Carter of Louisiana was blunter. Black pain is not a marketing prop, lynching is not a creative concept, the Klan is not an aesthetic.

Plenty of people think the whole thing is a misfire of reading rather than a misfire of art directing. Culture writer Meecham Whitson Meriweather said outright that everyone is insane and pointed at the caption as proof the image was about stage lighting. I understand the argument. It does not change one thing about what an operator should do Monday, and I will explain why in a minute.

The correction, and it is my own

On Tuesday's episode I said this was the composition of one photo, and I built the whole fix around one asset. Put a cold reader on every finished frame before it publishes, somebody who was not in the room when it got made, give them authority to hold the post button, two minutes a piece, cheapest insurance in marketing.

I still believe in the cold read. But a cold read on each asset, one at a time, clears both of these ads.

Run it. Asset one, September 3: a pop star in a white skirt under a star-shaped spotlight holding shoes. Loaded, yes, and a good reviewer flags it, which is why I would still take that gate over nothing. Asset two, September 14: a guy on a ladder holding shoes over his head. In isolation that photo is nothing. It is a man reaching. There is no version of a per-asset review, done in Mexico, eleven days later, on a completely separate campaign for a Danish collab, that catches a problem which only exists when that photo touches a photo from a different account, in a different hemisphere, from a different brief.

Nobody at Converse made the pair. Kida Maree made the pair. And the pair is the thing that went around the world.

Here is the reusable version, and it applies to a five-person brand exactly the same way. The unit you review is not the unit a stranger consumes. You review an asset. A stranger consumes a grid, a scroll, a screenshot somebody else assembled out of whatever you published near each other. Approval happens one file at a time because that is how files arrive. Damage happens in pairs, because that is how feeds render.

That is also the reason the art-criticism defense does not help you operationally. Maybe the caption does explain the lighting. The caption is not in the screenshot. Nothing you wrote is in the screenshot. Somebody else chose the crop, chose the neighbor, and chose the frame, and you were not consulted on any of the three.

The fifteen days are the real warning

The detail I keep coming back to is the gap. That image was live and quiet for two full weeks. Nobody screenshotted it. Nobody wrote a thread.

Which means a pre-publish gate, even a perfect one, gets graded on the wrong day. The post cleared review, cleared publication, cleared two weeks of actual audience exposure, and then failed anyway because something new showed up next to it. Everything you have already published is still an open position. It can be re-read at any time, against material that did not exist when you published it, including material you are about to make.

So the cold read is not a gate. It is a rhythm.

Translate it to a food and beverage feed

Your grid renders three across. Which means posts you made a week and a half apart end up physically touching, and neither of you nor your calendar had any say in it.

The pairs that actually hurt a food brand are dull, which is exactly why nobody catches them. A smiling product hero landing next to a customer-service reply about a bad batch. A hiring post next to a heartfelt thank-you to a ten-year employee, which together read as a closure. A founder post about tightening costs, then a price change three squares later, and now the tightening looks like the reason. A close-up of somebody's hands working next to a post about your new food-safety certification. A fryer shot next to the lighter menu you launched in January.

Not one of those posts is a mistake. Every one of them passes review. The damage lives in the seam, and the seam has no author.

Three things to do this week, all free

The polarity on this one is negative, so these are mitigation and response, not opportunity. Nobody reading this is in the Converse fire. The point is to find your own seams before a stranger does.

1. The pasteable post. Do the grid read on yourself, then publish what you found. This is the rare case where the audit is the content.

I read our own feed this week the way somebody who has never heard of us reads it. Nine posts, three across, on a phone, out loud.

[The pair you found, described plainly. These two posts, [X] days apart, next to each other, said [the third thing they said together].]

Neither post was wrong. Together they said something I did not mean. So we changed [the specific, boring thing you changed. The order. The spacing. Who looks at it.]

If you run an account, go read your own last nine. Not the posts. The pairs.

Three rules on it. Name something you actually found, because a hypothetical reads as a humblebrag. Do not name another brand, because the second you reference Converse this stops being about your feed and starts being a dunk. And name the change, not a value. Nobody believes a policy, everybody believes a Tuesday.

2. The still. Shoot the grid, not the post. A second person holds your phone at arm's length, your own profile grid on the screen, three across, shot at the desk where the posts actually get made. Eye level, natural light, the clutter left in. Do not build a designed carousel of your grid, and do not crop out the room. The picture has to look like somebody checking their work, because that is the whole claim. If your phone screen does not photograph clean, shoot it slightly off axis to kill the glare and let the reflection stay.

3. The video, fifteen to thirty seconds, five shots.

• Shot one, two seconds. A hand passing a phone across a desk to somebody else. No dialogue.
• Shot two, four seconds. That person's face, scrolling, saying the brand name out loud and what the account seems to be about.
• Shot three, eight to twelve seconds. Over the shoulder on the grid, that person naming one pair and what the two posts say together. One take. Their words, not yours.
• Shot four, four seconds. Close on the two thumbnails side by side, thumb still in frame.
• Shot five, three seconds. You, saying what you are changing, in one sentence, with the number of days between the two posts in it.

Three rules here. The person holding the phone cannot work on the account, and if you have to reach outside the company to find them, that is the exercise working. Nobody says the word optics. And do not script shot three, because the only useful thing in the whole video is a stranger's unrehearsed read, and a stranger reading your line is just you with a different face.

The free internal move that makes all of it repeatable: screenshot your own grid, nine at a time, once a week, and paste the screenshots into one running document. Then read only the pairs and the triples. Never the singles. Mark every pair where the two together say a third thing, and keep the marked ones as a list. Inside a month you will have your own catalog of the seams your brand produces, which is a more useful review checklist than anything you could buy, and it cost you four minutes a week.

The part you keep

Converse had the budget, the ambassador, the legal team and the approval chain, and the thing that hurt them was assembled for free by somebody on Threads out of material Converse handed over willingly. You cannot approve your way out of that. You can only go back and re-read what you already published, on a schedule, the way a stranger assembles it.

Two minutes before you publish is a good habit. Four minutes a week after you publish is the one that would have caught this.

This piece extends this week's episode of Social Media Train Wrecks, where I prescribed a per-asset cold read before I went and looked at the second ad. The show is the briefing, the article is the extension, and the post is what you do with it. Subscribe on Apple Podcasts, and every episode lives on the show page.

If you want somebody outside your company to do the cold read on your feed, find me on LinkedIn.