Mark ở vũ trụ khácVIMark học AI
Mark ở vũ trụ khác · Another Universe Anatomy of a Business Failure #4 · Product · Market research

New Coke

A B ✓ → 79 days

The Coca-Cola Company · 23 April to 11 July 1985 · United States

Whatdropped the original formula for a new taste
Outcomecustomers revolted; the old taste returned after 79 days
Root causemeasured "which sip tastes better", not "what if the old one is gone"
Main sourceThe Coca-Cola Company's own history pages
Re-read in 1 minute

Essence

Nearly 200,000 tastersIn taste tests the new formula was preferred. Plenty of data.
23 April 1985Coca-Cola dropped the original formula and sold only the new taste.
1,500 calls a dayBy June, the consumer hotline was flooded with complaints.
A U-turn in 79 days11 July 1985: the old taste came back as "Coca-Cola Classic".
The wrong questionMeasured "which sip is better", not "what if the old one disappears".

Source: The Coca-Cola Company, history page "New Coke: The Most Memorable Marketing Blunder Ever?" (launch date, return date, nearly 200,000 tasters, 1,500 calls a day in June 1985).

Part 1

The question of the case

America's biggest soft-drink company tested a new taste on nearly 200,000 people, and the new taste won. Why did the launch become one of the most famous flops ever? Because they measured an easy thing very carefully, then used it to decide something else entirely.

Source: The Coca-Cola Company, New Coke history page (nearly 200,000 tasters; the company itself calls it "the most memorable marketing blunder").

Before 1985Measure tasteBlind tests, one sip at a time. The sweeter new taste wins.
→
23 April 1985Replace itNot one more choice: the original formula is dropped.
→
11 July 1985U-turnCustomers lost something they were attached to. The old taste returns after 79 days.
Part 2

What happened

Dates and figures come from Coca-Cola's own history pages; anything taken from the press is marked "secondary source".

1985. On 23 April, Coca-Cola announced the new formula and stopped selling the original. By June, the hotline was getting about 1,500 calls a day, compared with about 400 before the change. On 11 July, 79 days later, the original formula returned as Coca-Cola Classic. The new taste stayed on sale alongside it. 79 days with only the new taste on shelves 23 Apr · launchoriginal dropped June~1,500 calls/dayabout 400 before 11 Jul · U-turnold taste back: "Classic" The new taste did not vanish: both were sold side by side, and the old one quickly took over. illustrative timeline, not to scale
Not a collapse, but a U-turn. Coca-Cola did not go bankrupt or lose the company. What it nearly lost was the trust of its most loyal customers.Source: The Coca-Cola Company, New Coke history page (23 April, 11 July, 79 days, 1,500 calls a day in June versus about 400 before).
Calls per day to the hotline: about 400 before the change, about 1,500 in June 1985, nearly four times as many. before the change~400 / day June 1985~1,500 / day, nearly 4x
Customers phoned in to save a soft-drink taste. A news wire at the time (UPI, 10 June 1985) counted about 40,000 calls so far, mostly negative.Source: The Coca-Cola Company, New Coke history page (1,500 and about 400 a day); UPI, 10 June 1985 (40,000 calls, secondary source).
Early 1980s · Pressure from a rivalPepsi ran its blind-taste "Pepsi Challenge" campaign. Coca-Cola started looking for a new taste (secondary source).
Before April 1985 · Large-scale taste testsNearly 200,000 people took part. The new formula was preferred.
23 April 1985 · New taste announcedThe original formula is discontinued. There is only one Coca-Cola: the new one.
May–June 1985 · BacklashAbout 1,500 calls a day. A psychiatrist hired to listen to calls said some callers sounded as if they were talking about the death of a family member.
11 July 1985 · U-turnThe original formula returns as "Coca-Cola Classic". ABC interrupted its programming to report it.
1990 · RenamedThe new taste is renamed "Coke II" (secondary source; one source says 1992).
2002 · GoneCoke II is discontinued in North America (secondary source).
2019 · One more timeCoca-Cola sold a limited run of New Coke for "Stranger Things" season 3, set in 1985.

Source: The Coca-Cola Company, New Coke history page (tests, 23 April, backlash, 11 July, ABC) and "1985: New Coke and Stranger Things" (2019); Wikipedia and Mental Floss for the Coke II dates 1990/1992 and 2002 (secondary sources); History.com for the Pepsi Challenge (secondary source).

Part 1

What happened

200,000 tasters23 Apr 19851,500 calls/day79 days
Part 2

Why

one-sip testreplace, not addattachmentmyths
Part 3

Work lesson

the decision questiontest in real lifea way back
Part 3

Why: four layers of cause

Measurement: one sip to decide a lifetime

measuring the easy thingCoca-Cola history page
The question that was measured: in a blind test, which sip tastes better? Answer: the new one. The question the decision really needed: if the old taste disappears from the shelves, how will customers feel? That was barely measured directly. WHAT WAS MEASURED "Which sip tastes better?" blind, unlabelled, one sip nearly 200,000 people → new taste wins WHAT NEEDED MEASURING "What if the old one is gone?" labelled, a full can, a full week habit, memories, identity → barely measured The data was right for the first question. The decision belonged to the second.
The research was not wrong; it answered a different question. Coca-Cola itself writes that all the research could not measure customers' deep emotional bond with the original. Many critics (secondary sources) add that a single sip tends to favour the sweeter drink, unlike drinking a whole can.Source: The Coca-Cola Company, New Coke history page (nearly 200,000 people; the bond that could not be measured); on sip tests: History.com and Malcolm Gladwell, Blink (2005) (secondary sources).

Strategy: playing by the rival's rules

fighting on the rival's termssecondary sources
The rival chose blind tasting as the battlefield. Coca-Cola answered with a taste that wins blind tests. But what customers buy every day is not a blind sip, it is a familiar brand. Rival"take the blind test" → Coca-Colamakes a blind-test winner → Customersbuy a labelled brand Winning on the field your rival picked is not winning where your customers actually live.
A target metric borrowed from the competition. The "Pepsi Challenge" pressure is told in many books and articles; market-share figures for those years differ between sources, so this piece uses none.Source: History.com, "Why Coca-Cola's 'New Coke' Flopped"; Thomas Oliver, The Real Coke, the Real Story (1986) (secondary sources).

Organisation: replace, not add, with no way back

replace, not add · no way back plannedCoca-Cola history page
Two ways to launch a new taste. Adding: the old one stays on the shelf, the new one stands beside it, customers choose, and if it is wrong you withdraw the new one. Replacing, the way Coca-Cola chose in 1985: the old one is removed, customers are forced to switch, and if it is wrong you must reverse in public. ADD old + newif wrong, withdraw the new REPLACE · 1985 only the new oneif wrong, reverse in public Once customers lose their choice, "preferred" turns into "taken away".
A one-way decision treated like a two-way one. Had the new taste only stood beside the old one, a misleading taste test would have cost one product line.Source: The Coca-Cola Company, New Coke history page (original discontinued at launch, returned on 11 July 1985); the comparison is Mark's commentary.

People: a small but strong signal

intensity, not just sharesecondary sources · Schindler 1992
In the research, about 10 to 12 percent of tasters said they would be angry or quit Coca-Cola if the old taste disappeared. This small group was treated as a minority. But a minority with strong feelings can sway a whole group, which one-by-one taste tests cannot see. most: liked the new taste or did not much care 10–12% A small group said: "lose the old taste and I'm angry, I'm gone". In focus groups, the strongly opposed swayed others (Schindler, 1992). Count the share and it looks small. Measure the intensity and it looks dangerous.
The warning was there, but it was counted as a small number. The 10–12% figure comes from books by Thomas Oliver and Mark Pendergrast via summaries; each ties it to a different part of the research, so read it only as "a sizeable minority".Source: EBSCO Research Starters and Wikipedia "New Coke" citing Oliver (1986) and Pendergrast (secondary sources); Schindler, R. M. (1992), Marketing Research, pp. 22–27.

Not the taste

A psychiatrist hired to listen to calls said some sounded like people talking about the death of a family member (Coca-Cola history page).

But loss

Keough at the U-turn: the research could not measure the bond with the original (Coca-Cola; History.com).

And identity

A familiar drink is part of daily life, not just a flavour (Mark's commentary).
Part 4

Fact check: what's true and what's myth

13 things often said about New Coke, checked against Coca-Cola's history page and other sources. Tap a line to see why.

Each source counts the complaint calls differently. Coca-Cola: about 1,500 a day in June 1985. UPI on 10 June 1985: about 40,000 calls so far. History.com: up to 8,000 a day. The often-quoted 400,000 has no source that could be found. Coca-Cola · June 1985~1,500 calls/dayprimary source UPI · 10 June 1985~40,000 calls (cumulative)news at the time History.com"up to 8,000 a day"secondary, conflicting social media"400,000 calls and letters"no source found
Famous cases grow their numbers over time. This piece uses Coca-Cola's 1,500 calls a day as the main figure.Source: The Coca-Cola Company, New Coke history page; UPI 10 June 1985; History.com (secondary source).
SOLIDNew taste launched 23 April 1985, old taste back 11 July 1985

Coca-Cola's history page gives both dates and calls it 79 days.

SOLIDCoca-Cola taste-tested nearly 200,000 people

Coca-Cola says "nearly 200,000 consumers". History.com (secondary source) says 190,000 blind tests: a small difference, the same point.

SOLIDIn June 1985 the hotline got about 1,500 calls a day

Coca-Cola gives 1,500 a day, versus about 400 before the change. UPI (10 June 1985) also gives 1,500 a day.

SOLIDKeough: "we are not that dumb, and we are not that smart"

President Donald Keough's words on 11 July 1985, answering those who said Coca-Cola had "planned the whole thing", quoted on Coca-Cola's history page.

CAUTION"New Coke only lived 79 days"

79 days is when the old taste came back. The new one stayed on sale alongside it, was renamed Coke II in 1990 (one source says 1992) and was discontinued in North America in 2002 (secondary sources).

CAUTION"8,000 calls a day, 400,000 letters"

Sources conflict: Coca-Cola says 1,500 a day, History.com up to 8,000, UPI about 40,000 cumulative calls. No source could be found for 400,000, so this piece drops it.

CAUTION"Coca-Cola's market research was wrong"

It correctly answered "which sip tastes better". The mistake was using that answer for a different decision: dropping the old taste altogether.

CAUTION"Customers revolted because the new taste was bad"

The new taste won blind tests. Most of the reaction was about loss and attachment: the psychiatrist listening to calls compared it to grief, and Keough said the research could not measure that bond.

CAUTION"There was no warning sign"

According to books on the case (secondary sources), about 10–12% of tasters said they would be angry or quit Coca-Cola if the old taste were gone. Schindler (1992) argues the focus groups showed how far this group's feelings could spread.

CAUTION"Coca-Cola lost hundreds of millions of dollars"

No loss figure published by Coca-Cola could be found. The numbers online cite no source, so this piece uses none.

FALSE"Coca-Cola failed on purpose as a publicity stunt"

No document shows that. Keough denied it on the day of the U-turn. Testing nearly 200,000 people and dropping your flagship product is far too big a bet for a stunt. Coca-Cola did benefit after the U-turn, but benefiting is not the same as intending.

FALSE"New Coke tasted bad"

In blind tests, the new taste was preferred (Coca-Cola history page). Taste was not the problem.

FALSE"Coca-Cola never asked its customers"

They asked nearly 200,000 people. They just did not ask the right question.

Source: The Coca-Cola Company, New Coke history page; UPI 10 June 1985; History.com; Wikipedia, Mental Floss, EBSCO (secondary sources); Schindler (1992), Marketing Research.

Part 5

Verdict

Scores are Mark's assessment based on the sources in Part 8, not figures from a study.

Avoidable
4/5
Warnings beforehand
3/5
Speed of correction
5/5
Lessons kept
4/5
Done right
  • Serious, large-scale research
  • Listened and reversed within 79 days
  • Admitted the mistake in public, did not blame customers
Done wrong
  • Measured taste, then decided about attachment
  • Replaced instead of adding, no way back
  • Played down a small, strongly opposed group

A rare failure with a happy ending: nobody cheated; a good measurement was used in the wrong place. And the company dared to turn back early.

More: why call it a "business failure" when the company is fine

Other pieces in this series are about companies that collapsed. New Coke is not: Coca-Cola is still standing. But this is the most common kind of failure at work: a product decision with great numbers that still goes wrong, because the numbers answer the wrong question.

The "lucky" part is the quick U-turn. Coca-Cola did not argue with its customers. It accepted that the research had missed something, and brought the old taste back while it could still be fixed.

Source for "More": The Coca-Cola Company, New Coke history page (79 days, Keough's words); the rest is Mark's commentary.

Part 6

Work lesson: measure what customers need, not only what is easy to measure

For any change to something people already rely on: an interface, a price list, a process, a report template, a dish on the menu.

Four steps before replacing something customers are used to: write the decision question before choosing the measure, test in real life rather than one sip, measure intensity not just share, and add before you replace, with a turn-back threshold ready. 1 · decision questionwrite it firstthen choose the measure 2 · test in real lifea full can, a full weeklabels and habits included 3 · measure intensitywho would leave?not just "what %" 4 · add, then replaceturn-back thresholdwritten on day one "Customers prefer it" is not "customers accept losing the old one".
Each step patches one thing that broke in 1985. The wrong question, the one-sip test, a strongly opposed minority played down, and a replace-only decision with no way back.Source: the failures come from Coca-Cola's history page and the sources in Part 4; the steps are Mark's suggestions.
Decision question first, measure secondWrite one line: "We are about to decide ___. To decide, we need to know ___." A measure that does not fill the second blank is not enough.
Test in real lifeA sip is not a can; a demo is not a week of use. Let a small group use the new version for real, labelled, for a few days.
Measure intensity, not just shareAlso ask: "If we removed the old one, what would you do?" 10% who would leave matter more than 60% who "slightly prefer" the new one.
Add first, replace laterKeep the old version running alongside for a while. Write the turn-back threshold, such as complaints per day, before you launch.
Check your own team (5 minutes)

Take the biggest change your team is about to make. Answer three questions, one line each:

1. Which number makes the whole team confident, and what question does it answer?
2. Is that really the question behind the decision?
3. If we are wrong, how and when do we turn back?

If question 2 is "not quite" or question 3 is blank, that is your team's "New Coke".

The truth is we are not that dumb, and we are not that smart.Donald Keough, President of Coca-Cola, 11 July 1985, as quoted on The Coca-Cola Company's history page.
Part 7

Read next

The Coca-Cola Company, "New Coke: The Most Memorable Marketing Blunder Ever?"A company retelling its own biggest blunder. Rare and worth reading.
Robert M. Schindler (1992), "The real lesson of New Coke"A short paper on how focus groups saw what one-by-one taste tests could not.
Thomas Oliver, The Real Coke, the Real Story (1986)An inside account (secondary source), where many of the often-quoted numbers come from.
Anatomy of a Business Failure #05: Enron (2001)When nobody can explain where the money comes from, and why "just a few bad people" is not enough.
Part 8

Primary sources

This case has no court or regulator file, because there was no fraud. The best primary source is Coca-Cola's own material; press and books are marked as secondary sources.

  1. The Coca-Cola Company, "New Coke: The Most Memorable Marketing Blunder Ever?", company history page. coca-colacompany.com
  2. The Coca-Cola Company, "1985: New Coke and Stranger Things" (2019). coca-colacompany.com
  3. The Coca-Cola Company, "Veteran Employees Remember Infamous 1985 Launch of New Coke". coca-colacompany.com
  4. Schindler, R. M. (1992), "The real lesson of New Coke: the value of focus groups for predicting the effects of social influence", Marketing Research, December 1992, pp. 22–27.
  5. UPI, "New Coke generates 40,000 calls, mostly negative", 10 June 1985 (secondary source). upi.com
  6. History.com, Christopher Klein, "Why Coca-Cola's 'New Coke' Flopped" (secondary source). history.com
  7. Thomas Oliver, The Real Coke, the Real Story (1986); Mark Pendergrast, For God, Country and Coca-Cola; via EBSCO Research Starters "Coca-Cola Introduces a New Formula" (secondary sources). ebsco.com
  8. Wikipedia, "New Coke" and "Coke II"; Mental Floss, "What Was Coke II?" (secondary sources, for the 1990/1992 and 2002 dates). mentalfloss.com
  9. Malcolm Gladwell, Blink (2005), on sip tests (secondary source).

Note: the pages above were checked through a search tool on 13 October 2026; reopen the original links to proof the wording before publishing.