Starbucks' Tank Day fiasco in South Korea deserves a place in future editions of the best branding books as an example of what can happen when a huge global brand acts in blissful unawareness of cultural and political context. It's been devastating for the coffee chain's reputation in its third biggest market by store count, and it's unlikely to be forgiven soon.
As the dust settles, along with the fragments of smashed Starbucks cups and tumblers, apologies have been reiterated and a potential scapegoat has emerged. Unsurprisingly, it turns out that you shouldn't rely on an AI to write your marketing slogans, but Starbucks' Tank Day disaster can't be blamed entirely on a chatbot.
For those who missed the chaos, Starbucks South Korea, which is operated by the Shinsegae Group, launched a promotion involving large 'Tank Series' tumblers. 'Tank Day', it called it, using the name in English.
But there was a fatal oversight in the timing. The campaign launched on 18 May, the anniversary of the 1980 Gwangju Uprising, or Democratization Movement, during which military dictator Chun Doo-hwan sent tanks and armored carriers into the streets to suppress protests. Between 200 and 2,300 people are estimated to have been killed.
Making things worse, the campaign's slogan, which translates as something like “thwack on the desk”, reminded people of a phrase that became notorious in connection to the death of a student activist in 1987. Authorities claimed that Park Jong-chul died after an officer “hit the desk with a thwack”.
Starbucks' cancelled the campaign within hours as people protested outside stores, smashed Starbucks tumblers and mugs and deleted their loyalty apps en masse. Since then, government ministries have pledged to drop Starbucks from public events, delivery worker unions said they would refuse orders, and the country's president, Lee Jae Myung, has described those responsible for the campaign as “low-class peddlers”.
Shinsegae has fired Sohn Jeong-hyun, the former chief executive of Starbucks Korea, and the group's chairman has issued an apology on national TV.
An internal investigation was also carried out. According to Yoncap News , that found no evidence of any intent to reference the historic massacre. It also found that the controversial slogan was proposed by an AI chatbot.
Some have been quick to pin the blame on the AI, as if this explains everything, but that's like blaming Photoshop for a botched retouching or an Excel spreadsheet for a budget miscalculation. The fault lies with the people who used it.
By now, most people who work in branding and marketing will probably have turned to an AI chatbot as a potential source of ideas. They'll know that initial suggestions can range from the trite and cliched to the outright bizarre. Using AI on autopilot isn't an option, but it's been reported that some managers who approved the Starbucks campaign didn't even open email attachments with the marketing material.
Marketing turnarounds are getting faster and faster, and some brands seem to treat social media campaigns with a casualness that would have been unthinkable with traditional media. Campaigns get signed off faster than ever at a time when they can be exponentially replicated in an instant in a so-called post-truth world in which people judge what they see based on how it makes them feel more than what the intent might have been.
Fewer people are doing more, which is a problem that AI was supposed to solve, but there are old lessons to be remembered. Groupthink, described by the social psychologist Irving Janis in the '70s, has been blamed for debacles as varied as the US government's escalation of the Vietnam War and Pepsi's cringey Kendall Jenner ad.
A group's drive to reach a consensus can lead to poor, even irrational, decision-making. That risk is greater when at least one member of the group is an AI. Who wants to be the one to call out the technology that the CEO believes is the solution to everything? But today more than ever, branding requires multiple layers of careful analysis that questions every idea, whether it comes from a human or a machine.
