The 15 Worst CPG Product Fails of All Time

Every brand team has one launch it never brings up at the reunion. This is a ranking of the 15 worst CPG product fails of all time: the New Cokes and Crystal Pepsis, the $325 million cigarette, the purple ketchup that actually sold, and, more usefully, the one data signal each product ignored on the way to the discount bin.

None of these were dumb ideas from dumb people. They were well-funded teams with real research. What they have in common is quieter and more useful: in every case a number was available in time to change the decision, and in every case it either arrived too late or lost the argument to somebody's gut.

The 15 worst CPG product fails of all time

1. New Coke (1985)

Coca-Cola ran roughly 200,000 taste tests and most people preferred the new formula. None of the tests asked how a shopper would feel when the old one disappeared. New Coke lasted 79 days before the original returned as Coca-Cola Classic. The signal it missed: a sip test measures a preference, not what a shopper does on the second and third purchase.

2. RJR Premier (1988)

The smokeless cigarette. Smokers said it tasted like burning plastic, and non-smokers did not want one either. Reports put the loss at $325 million or more, with some estimates running past a billion dollars all in. The signal: if your core user hates it and your target user does not exist, the addressable market is zero.

3. Crystal Pepsi (1992)

Clear cola, sold on the idea that clear meant pure. It did roughly $474 million in its first year, about one percent of all US soft-drink sales, then production stopped inside a year once first-time trial ran out. The signal: trial is not velocity, and the first twelve weeks flatter everyone.

4. Frito-Lay WOW! Chips (1998)

Fat-free chips made with olestra, a record $400 million debut, and a federally suggested warning label about digestive effects we will leave to the imagination. Sales halved within two years. The signal: the product experience is the retention curve, and no advertising budget outruns it.

5. Gerber Singles (1974)

Adult meals in jars that looked exactly like baby food, sold under the name Singles. Pulled within months. The signal: nobody buys a product that narrates their life back at them in the checkout line.

6. Colgate Kitchen Entrees

A toothpaste brand's frozen dinners. The famous Colgate lasagna is so poorly documented that Colgate now denies it existed and the Museum of Failure exhibits only a replica box. What is verifiable is a 1960s test of frozen chicken and crabmeat entrees that was quietly buried. The signal: brand equity has a lane, and mint is not a dinner note.

7. Pepsi A.M. (1989)

Cola positioned for breakfast. The people who drank cola at breakfast were already doing it, and naming it A.M. only made everyone else feel judged. Dead within a year. The signal: an occasion is not a category.

8. Coors Rocky Mountain Sparkling Water (1990)

A brewery selling water in beer-adjacent packaging. Beer drinkers saw watery beer, and everyone else saw beer. Gone by the mid-1990s. The signal: if the shelf cannot tell what you are in two seconds, neither can the shopper.

9. Life Savers Soda (1980s)

It tested well in flavor panels. On the shelf, a candy brand on a soda made shoppers picture drinking liquefied candy, and it sold like it was cursed. The signal: a concept test measures the concept, and the shelf measures the purchase.

10. Ben-Gay Aspirin

From the brand whose entire identity is a warming, tingling burn came a pill you swallow. The lore outlived the product because the idea is a clean crime against brand logic. The signal: strong awareness cuts both ways when the association is that specific.

11. Cosmopolitan Yogurt (1999)

The magazine extended itself into the dairy case. Eighteen months later there was no yogurt. The signal: a licensing deal is not a reason to exist in a category, and shoppers can feel the difference on the shelf.

12. Heinz EZ Squirt Purple Ketchup (2000)

The twist is that it worked, 25 million bottles and a record-high share of the ketchup category. Then every kid in America had tried purple ketchup exactly once, and the line was gone by 2006. The signal: a novelty spike has a decay curve already scheduled the day it launches.

13. Orbitz (1997)

A drink with edible spheres suspended in it, marketed as texturally enhanced. Shoppers described it as a lava lamp you drink. Dead within a year, though sealed bottles now sell as collectibles, making it the rare fail that appreciated in value. The signal: distinctive only counts if people also want to consume it.

14. Coca-Cola BlaK (2006)

A coffee-cola in a premium bottle at about $1.89 for eight ounces, with twice the caffeine of a regular Coke. Discontinued by 2008. A decade later, coffee sodas and cold brew are everywhere. The signal: early and wrong look identical unless you can afford to wait, and most launches cannot.

15. Maxwell House Ready-to-Drink Coffee (1990)

Coffee in a carton, ready to drink, except cold coffee was not yet a habit, so you were meant to heat it, and the carton could not go in the microwave. The signal: run the entire usage occasion once, start to finish, before a national rollout.

The damage at a glance

What every one of these had in common

Line them up and the failures stop looking like fifteen different mistakes. They look like one mistake, repeated. A signal was there, and someone either did not see it or did not believe it. The signal is almost always one of three numbers.

The first is repeat purchase rate. Trial tells you a shopper was curious once. Repeat tells you whether the product earned a standing place in the cart. Crystal Pepsi and purple ketchup both had enormous trial and almost no repeat, and both looked like hits for exactly one buying cycle. The second is the split between base and incremental volume: a spike financed by a discount or a novelty is not the same as demand you get to keep. The third is sales velocity, units per store per week, which strips out how many doors you happen to be in and shows how hard each one is actually working.

It shows up the same way in every real report. A launch opens at 4.1 units per store per week in its first four weeks, riding an end cap and a feature. By week eight the display is gone and velocity is 1.6. By week twelve it is 0.7, under the 1.0 that keeps an item in the set. Shipments still look fine, because the distributor bought the launch, not the shopper. The repeat rate, sitting near nine percent against a category norm closer to thirty, said all of this in week five. Nobody was looking at week five.

How to catch it in your own first twelve weeks

You do not need a museum of dead brands to learn the lesson. You need the same three numbers on your own launch, weekly, before the reorder decision instead of after it. Most teams see them a quarter late, in a recap deck, once the markdown is already scheduled and the shelf review is already lost.

Every fail here had a number that showed up in time and lost to somebody's gut. Scout puts the launch signals that decide it, repeat rate, base versus incremental, per-store velocity, in front of you weekly, so next time the number wins the argument. Watch your next launch's first-twelve-weeks curve.