01 Jun In-Class Case Study: Coca-Cola’s “New Coke” Debacle (1985)

Data-Driven Decision Making — Measuring the Wrong Thing

Case Overview Links to an external site.

In the early 1980s, Coca-Cola was facing increasing competition from Pepsi, whose “Pepsi Challenge” taste tests seemed to show that consumers preferred Pepsi’s sweeter flavor. In response, Coca-Cola initiated one of the largest market research efforts in its history — nearly 200,000 blind taste tests — to evaluate a reformulated version of its flagship product. The results showed that the new, sweeter recipe was preferred over the original — and even beat Pepsi in some tests. Based on this overwhelming data, Coca-Cola executives decided to abandon the 99-year-old original formula and launch a new version called “New Coke.”

Despite the seemingly strong quantitative evidence, New Coke flopped almost immediately. Loyal drinkers reacted with shock and anger, writing letters, making phone calls, organizing protest groups, and demanding the return of the original formula. Only 79 days after the launch, Coca-Cola reversed course and brought back the original recipe as “Coca-Cola Classic.”

The Core Failure: Metric Validity

What happened? Coca-Cola collected lots of data — but they measured the wrong thing. The blind taste tests told the company which drink tasted better in isolated conditions, but they didn’t capture the emotional and cultural value consumers had for the original Coke. In other words, the metric (taste preference in a blind, decontextualized test) lacked construct validity for the business decision at hand — which was not just about taste, but about brand identity and emotional attachment.

Why It Matters

High sample size ≠ good inference: 200,000 taste tests couldn’t make the results useful if the wrong variable was being measured.

Quantitative vs. qualitative: Numbers alone failed to capture how customers felt about the original Coke — and that emotional bond was far stronger than any slight taste preference.

Real-world behavior vs. artificial tests: In blind tests, people only evaluated taste. In the real world, people care about meaning, nostalgia, and identity.

Class Discussion / Assignment

Identify the research design flaw.

1. Why did the blind taste tests mislead Coca-Cola’s decision-makers? What key variable was not measured?

2. How could Coca-Cola have improved the validity of their metrics? Suggest at least two alternative or additional research methods that would have captured emotional and brand attachment.

3. What data would you collect today to avoid a similar mistake? Propose a mixed-methods research plan (quant + qual) appropriate for this decision.

4. What lessons does this case teach about data-informed vs. data-driven decisions? Reflect on how contextual insights should complement numerical data.