Coca-Cola ran 200,000 taste tests that said New Coke would win — 79 days later a national revolt forced the original formula back

The most rigorous piece of consumer research in marketing history pointed confidently in exactly the wrong direction. In the early 1980s Coca-Cola ran blind taste tests on close to 200,000 people. A new, sweeter formula beat the old Coke. It beat Pepsi. The data was clean, it was enormous, and it was nearly unanimous. On April 23, 1985, the company reformulated the world's best-known consumer product on the strength of it. Seventy-nine days later, on July 11, 1985, it surrendered and brought the original back as "Coca-Cola Classic." The research was not sloppy. The sample was not too small. The numbers were real. They measured the one thing that did not decide the outcome.

The data that flattered the boardroom

By 1985 Coca-Cola had a defensible reason to act. Pepsi had spent years running the "Pepsi Challenge" — public blind sip-tests in which people, on average, preferred the sweeter Pepsi — and Coke's market share lead had been narrowing for two decades. So Coca-Cola did what a disciplined, data-led company is supposed to do: it tested its way to an answer. It developed a smoother, sweeter formula and put it through roughly 200,000 blind taste tests against both the original Coke and Pepsi. The new formula won. Internally, the confidence was total; the company has since described the figure as nearly 200,000 consumers preferring the new taste. On those numbers, reformulating was not a gamble. It looked like the safe, evidence-based move. That is what makes the case dangerous rather than merely embarrassing — the decision was wrong and well-researched at the same time.

What the sip test could not see

A blind taste test measures one variable: which liquid tastes better in a single mouthful, stripped of everything else. That is a real signal, and for a sweeter drink it favors the sweeter formula — the first sip is where sugar wins. But almost nothing about how people actually consume a soft drink resembles that test. Nobody drinks one anonymous ounce. People drink a whole can, over time, attached to a red-and-white object they have known since childhood, tied to memory, ritual, and identity. The 200,000 tests captured the sip. They could not capture the relationship. Worse, the format had a known bias the company under-weighted: in a single-sip comparison the sweeter sample reliably wins, yet that preference can reverse over a full serving. The research wasn't measuring the wrong people. It was measuring the wrong moment — the narrowest slice of the experience, mistaken for the whole.

The market answered in a way the test never asked

When New Coke launched, the response was not a quiet shift in preference share. It was a revolt. The company's consumer hotline, which had been fielding around 400 calls a day, climbed to roughly 1,500 daily by June 1985. Protest groups formed with names like the Old Cola Drinkers of America, which claimed to recruit 100,000 members and threatened a class-action suit to force the original back. People hoarded the old formula by the case; press coverage turned a soft-drink decision into a story about American identity. None of this fury showed up in a sip test, because none of it is about taste. It was about a brand acting like it owned a product its customers felt they owned. When Coca-Cola announced the original's return on July 11, 1985, the news broke into network programming, and the hotline took on the order of 31,600 calls in the days that followed — relief, not preference. The metric that won the boardroom predicted none of it.

The 79 days

The reversal was fast, and the speed is the tell. A genuinely wrong product takes time to fail in the market; a product that violated a relationship failed almost on contact. Within 79 days Coca-Cola executives stood up and gave the original formula back, branded as Coca-Cola Classic, and it quickly outsold both New Coke and Pepsi. The company's own historians now call it one of the most memorable marketing missteps ever — and, with hindsight, note how thoroughly it underestimated the bond consumers felt with their Coca-Cola. The instructive part is not that a giant blundered. It is that the blunder ran straight through 200,000 data points and out the other side. More data did not protect the decision, because more of the wrong measurement is not insurance. It is amplified confidence in the wrong answer.

The lesson: a metric that wins the room is not a metric that predicts the market

New Coke is taught as a branding cautionary tale, but the sharper lesson is about evidence. Coca-Cola did not fail to do research; it failed to question what its research was a proxy for. "Which tastes better in one sip" was treated as a stand-in for "which will people buy and defend in their lives," and the two came apart violently. This is the most seductive failure mode in any design or product decision: a clean, large, reassuring number that measures a narrow thing, presented as if it measured the whole thing. The test that flatters the boardroom — high N, clear winner, easy to act on — is very often not the test that predicts the market, precisely because the easiest thing to measure is rarely the thing that decides the outcome. The discipline is not to gather more data. It is to keep asking whether the data is measuring demand and context, or just the sip.

Why a Design Intelligence company tells this story

We exist to keep this exact gap from opening. Design intelligence means grounding a decision in real demand and lived context — what people actually do with a product, attached to it, over time — rather than in a single metric that happens to win the room. New Coke is the perfect warning because the research was excellent and the conclusion was catastrophic; the failure was entirely in mistaking a narrow signal for the whole picture. We use the intelligence of AI to help leaders pressure-test a design or product decision against the real-world relationship people have with it before the market does — so the number that gets approved in the boardroom is one that also survives contact with how people live. The sip is easy to measure. The relationship is what the market actually votes on, and it is the only thing worth being confident about.

Sources

• ●New Coke: The Most Memorable Marketing Blunder Ever? (The Coca-Cola Company)
• ●Why Coca-Cola's "New Coke" Flopped (HISTORY, updated 2 Apr 2025)
• ●New Coke debuts, one of the biggest product flops in history — April 23, 1985 (HISTORY, This Day in History)
• ●New Coke (Wikipedia)
• ●New Coke: A Classic Branding Case Study on a Major Product Change Failure (The Branding Journal, 24 Feb 2025)
• ●New Coke Turns 40: Facts About Coca-Cola's Massive 1985 Marketing Failure (ReMIND Magazine, 23 Apr 2025)