New Coke 1985: Why Coca-Cola Lost a Fight It Won in Taste Tests

The room

The psychology: loss aversion

Loss aversion is one of the best-documented patterns in behavioral economics. People judge outcomes as gains or losses from a reference point, usually what they already have, and losses weigh more. The endowment effect is its everyday form: once something is ours, we demand more to give it up than we would pay to get it. In the work of Daniel Kahneman, Jack Knetsch and Richard Thaler, the object that revealed this was nothing more valuable than a coffee mug.

Coca-Cola’s taste tests measured a gain: would you like this new flavor more? They did not measure a loss: how would you feel if the drink you already have, the one many people tied to memories, places and family habits, were taken away? Those are different questions, and loss aversion predicts different answers. A drinker could prefer the new sip by a small margin and still feel the removal of the old one as a large loss, because the reference point was not neutral. It was the original Coke, which millions of people felt they already owned.

Brand identity amplifies the effect. For many drinkers Coca-Cola was not just a flavor but part of who they were. Mother Jones later described the protest as a revolt against the idea of change itself. Removing the old formula turned a product decision into a personal loss, and people responded the way loss aversion says they will: with far more energy than the small gain on offer could explain.

Why smart people fall for it

Three forces turned a winning taste test into a 79-day retreat.

1. 1

   Losses loom larger than gains

   A slightly better taste was a small gain. Losing the original was felt as a large loss, and the two did not cancel out.

2. 2

   The sip test measures something else

   A blind sip strips out the brand and reportedly favors sweeter drinks, so it misses what the name adds to the experience of drinking a whole can.

3. 3

   Identity turns products into possessions

   When a brand is part of how people see themselves, changing it feels like losing something they own, not like receiving a product update.

Was New Coke really a marketing mistake?

The textbook version treats New Coke as a pure failure. The fuller record is more mixed. The new formula did win blind tests, and Mother Jones, revisiting the case in 2019, pointed to surveys showing it held up in the field, including a narrow 52 to 48 edge in the South. Early sales rose. The problem was never that people could not drink it. It was that the company had removed the alternative.

There is also the outcome. Within months Coca-Cola had two colas on the shelves, the original had retaken the sugar-cola lead from Pepsi by early 1986, and the brand had received a level of attention no advertising budget could buy. That is why some observers suspected a stunt. The company denied it, and the record supports the denial: in June, Coca-Cola USA’s president told TIME the company would stick with what it had done, and the reversal came only after weeks of mounting complaints.

Knowledge at Wharton’s review of Constance Hays’s book on the company argues that New Coke was not the clueless blunder textbooks describe, while noting that bottlers, who knew local tastes, were not consulted before the new drink was imposed from above. That is a narrower but more useful verdict. The research was competent at what it measured. The decision went wrong where it relied on that research to answer a question it never asked.

So the fair answer is that New Coke was a costly mistake in execution and a revealing one in psychology, but not a disaster for the company. Removing the original was the error. Offering a new taste was not.

What did later research show about taste and brands?

Some of the strongest evidence came almost twenty years later from a brain-imaging lab. Samuel McClure and colleagues gave volunteers Coke and Pepsi in anonymous and labeled conditions. Without labels, preferences followed one pattern. When people knew they were drinking Coke, the label alone shifted what they said they liked and engaged regions linked to memory. Pepsi’s label had no comparable effect. In plain terms, the brand was part of the taste.

That result explains both halves of the 1985 story. The Pepsi Challenge was real: strip away the labels and many people prefer the sweeter cola. But Coca-Cola’s advantage had never been only in the liquid. It was in the associations the name carried, and those associations were attached to the original formula. Replacing the formula while keeping the name asked people to transfer a lifetime of associations to a new drink overnight.

The endowment research adds the second piece. In the Kahneman, Knetsch and Thaler experiments, simply being handed a mug was enough to make people demand much more to give it up than others would pay for it. A relationship with a drink built over decades is a far stronger endowment than a mug. The angry minority in Coca-Cola’s own research was not noise. It was an early reading of how large that endowment was.

Other data from 1985 fit this reading. Company surveys showed most people who tried the new drink would buy it again, while an independent survey found most triers said the original tasted better. Both can be true at once. Preference for a taste can be mild while attachment to a brand is intense, and the stronger feeling drives behavior when something is taken away.

How would a modern team decide?

A modern team would start by separating two questions: do people prefer the new product, and how will they react if the old one disappears? The first is a taste or usability test. The second needs different tools, such as asking existing customers to imagine losing the product, measuring how willing they are to switch, and watching how intensely they respond, not just whether they agree.

It would also treat the angry minority as a segment to size, not an outlier to correct. In 1985 roughly one tester in ten reacted against the idea of change. A modern team would ask how many customers that represents, how vocal they are and how much attention they could generate. With social media, a tenth of a loyal base can set the story within hours.

Third, it would prefer reversible moves. Launching alongside the original, or testing the replacement in one region first, would have exposed the reaction at lower cost. Coca-Cola ended up running that experiment by accident: once the original returned as Classic, both drinks were on the shelves, and customers chose with their purchases.

Finally, a modern team would plan its retreat in advance. Coca-Cola’s reversal worked partly because it was fast. Seventy-nine days is short for a company of that size, and the public framing, that the consumer was in charge, turned an embarrassment into a show of respect for loyal drinkers. Deciding before launch which signals would trigger a reversal makes it easier to act before the damage compounds.

The manager's playbook

1. Ask what customers lose, not just gainPair every preference test with a removal question: how would you feel if the current version were no longer available?
2. Size the angry minorityTreat strong negative reactions in research as a segment to measure, and estimate how many real customers and how much attention they represent.
3. Keep the original available at firstWhere you can, launch the new version alongside the old one and let usage, not tests, decide when to retire it.
4. Test in one market before allRoll out in a limited region or channel so that a backlash stays contained and teaches you something.
5. Set reversal triggers in advanceAgree before launch which complaint, sales or sentiment thresholds will prompt a rethink, so the decision is not made under pressure.

Questions people ask