On April 23, 1985, Coca-Cola announced it was retiring the formula it had been selling for 99 years and replacing it with a new one. The decision was not arrogance. Pepsi had been winning public blind-taste face-offs since the mid-1970s, and Coke's share lead in its flagship market had been slipping for fifteen straight years. The new formula was the company's answer, and it came with evidence: nearly 200,000 taste tests, which the new formula won against the old Coke and against Pepsi. The company's own history page still carries the number without embarrassment - a formula "preferred in taste tests of nearly 200,000 consumers."
Seventy-nine days later the company reversed itself. The old formula came back under the name Coca-Cola Classic, and the formula that had just been preferred in nearly 200,000 taste tests became a punchline for a decade. The standard telling of this story is a lesson about arrogance: the company got cocky, ignored its customers, paid the price. I think that reading is exactly backwards. The part that should worry anyone who buys attention is the part nobody repeats: the research was right.
The answer was right
The new formula kept tasting better, and people kept saying so even while the protest raged. By July, Coca-Cola executives were telling reporters that more than 150 million Americans had tried the new drink - double the number who ordinarily drink Coke in a year. The protest's own general could not reliably tell the formulas apart. Gay Mullins, the 57-year-old Seattle man who founded the Old Cola Drinkers of America, sat for blind taste tests more than once, and by later accounts he either failed to tell the two formulas apart or picked the new one outright. The sips did not lie. Americans really did prefer the taste. That is the inconvenient detail the parable drops.
What the tests never asked was the only question in the room. A taster got two unlabeled cups and one instruction: pick the one that tastes better. That is the whole questionnaire. Nothing in that setup asks what the taster will do if the cup they didn't pick disappears from the world. The tests measured preference, and preference is thin. What arrived at Coca-Cola's Atlanta headquarters that spring was attachment, and attachment does not answer questionnaires. It answers events.
The grief the survey missed
The consumer hotline went from 400 calls a day before the change to 1,500 a day by June. People filled their basements - the company's own history records hoards of up to 900 bottles. One letter arrived addressed to "Chief Dodo, The Coca-Cola Company." Another, quoted in the Los Angeles Times that summer, read: "Changing Coke is like God making the grass purple or putting toes on our ears or teeth on our knees." That is not a note about flavor. By the time executives faced the press on July 11 to announce the old formula's return, more than 15,000 people had called to praise the decision, many of them, the Times reported, sounding as if they were in tears. Chairman Roberto Goizueta's summary was, "It's just wonderful to be loved again."
At that same press conference, company president Donald Keough said the thing I want nailed to the wall of every marketing department: "The simple fact is that all of the time and money and skill poured into consumer research on the new Coca-Cola could not measure or reveal the deep and abiding emotional attachment to the original Coca-Cola felt by so many people." He admitted the old formula's hold on people had caught the company by surprise. And when cynics suggested the whole affair had been staged for publicity, his answer was, "The truth is we are not that dumb and we are not that smart."
Here is the lesson I keep from 1985, and it is not "listen to your customer." Coca-Cola listened with more rigor than most companies ever bring to a question, and every answer it got was accurate. An audience will answer anything you ask, honestly, in the terms you set. What it will not hand over is the deep structure: what your name means to people, whether they would fight for it, whether they come back when nobody is watching. Most of the people carrying that attachment do not know they are carrying it, which is exactly why no questionnaire can reach it. It has one instrument, and the instrument is an event. A launch. A change. A disappearance. The audience tells you what it feels in what it does after the event, and only there.
The only honest instrument
This is why I have stopped trusting pre-purchase research, and why I think about paid placement the way I do. A slot on the Bid Spotlight board is not a survey. Nothing is being asked of the people who see your name near the top of the leaderboard; they were already browsing, already hunting for their next thing. The position is an event. It puts you in front of that crowd and hands back the only numbers that have ever been honest: they click or they don't, they come back or they don't, the promo video gets watched or it doesn't.
Every estimate you made before spending money - what the market wants, what the audience will respond to - is a sip test. It can be perfectly designed, perfectly executed, perfectly correct about taste, and still miss the entire bond. The board itself runs on the same law. Ranks do not move on sentiment, intent, or a promising conversation. A position changes hands when somebody actually pays. The leaderboard is a machine for running the one experiment that reads an audience truthfully: put your name up, count what comes back, and ignore what anybody merely says.
If you want to know what your audience will do instead of what it will say, claim a spot on the board and give it something to answer.