38 post-mortems · 15 dead · 23 ghostsUpdated October 8, 2026
Every brand dies twice: once when the money runs out, and again when people forget it.
Spectre Brands writes the second obituary. Each post-mortem rebuilds a company from its SEC filings, court records, contemporary press and archived web pages, with charts you can check, the commercials as they aired, and original drawings of what the brand looked like.
Two ways a brand ends
Dead 15 brands
The company is gone: liquidated, dissolved or shut down, and nothing operates under the name today.
Ghost 23 brands
The original business failed or shrank to a shadow of itself, but the name is still in use: as a licensed brand, a shell, a single surviving outlet, a limited re-release, or a much smaller operation under new owners.
The index
38 post-mortems

Dead1996 – 2001Overexpansion
Webvan
Raised more than $1.2 billion, built warehouses for millions of orders, and stopped delivering on a Monday in July 2001.
$827M cumulative net loss, 1998 to Mar 2001

Pets.com
1998 – 2000

Groupon
2008 – present

AOL
1985 – present

Kozmo.com
1998 – 2001

eToys
1997 – 2001
theglobe.com
1995 – 2001

Ghost1921 – 2015Technology shift
RadioShack
Once a store within a short drive of almost every American. It kept the stores long after the reason to visit them had moved online.
5,289 company stores at the 2001 peak

Circuit City
1949 – 2009

CompUSA
1984 – 2008

hhgregg
1955 – 2017

Sharper Image
1977 – 2008

Ghost1957 – 2018Debt and buyouts
Toys “R” Us
The store that invented the toy superstore was still selling $11.5 billion of toys a year when it failed. Nearly every dollar it earned from operations went to interest on its own buyout.
$457M interest in its last full year, on $460M of operating earnings

KB Toys
1922 – 2009
Kiddie City
1957 – 1993

FAO Schwarz
1862 – 2015
Zany Brainy
1991 – 2003

Child World
1962 – 1992

Ghost1992 – 1994A fad that faded
Crystal Pepsi
Pepsi took the color out of cola, put about $40 million of advertising behind it and projected $1 billion in sales. Within a year the product was being reworked. The name now comes back for a few weeks at a time.
$40M estimated launch advertising, for a drink reworked within a year

Zima
1992 – 2008
New Coke
1985 – 2002
OK Soda
1994 – 1995

Orbitz
1996 – 1999

Dead1998 – 2001Outcompeted
Sega Dreamcast
The first console with a modem in the box had the biggest launch day the industry had seen. Sixteen months later Sega stopped making it and never built a console again.
510 days from U.S. launch to the exit

Atari
1972 – 1984

3DO
1991 – 2003

Ouya
2012 – 2019

Coleco
1932 – 1990

Ghost1985 – 2010Technology shift
Blockbuster
The biggest name in home video had 9,094 stores in 2004, a billion dollars of new debt and a rival it had declined to buy. One franchised store in Oregon still rents films under the sign.
1 store left, of 9,094 at the 2004 peak

Quibi
2018 – 2020
Hollywood Video
1988 – 2010

Redbox
2002 – 2024

FilmStruck
2016 – 2018

Ghost1925 – 2022Strategy and management
Howard Johnson’s
Once the largest restaurant chain in America, with 28 flavors and an orange roof on every highway. It was sold twice in six years, split in two, and left to its franchisees.
1,040 to 0 restaurants, 1979 to 2022

Burger Chef
1958 – 1982

Sambo’s
1957 – 1984

Chi-Chi’s
1975 – 2004

Steak and Ale
1966 – 2008
Nothing matches that combination.
How the post-mortems are made
No figure appears without a source, and nothing is presented as fact that we could not check. The short version is below; the full rules are on the About page.
- Every figure links to a numbered source: SEC filings and court records first, then contemporary press, then the Internet Archive.
- Numbers we calculate are labeled derived. Estimates are labeled with where they came from. Claims we could not confirm are labeled unverified, or left out.
- Old websites are shown as screenshots of Internet Archive captures, each linked to the original capture.
- Logos, ads and footage appear in historical and editorial context. Drawings are original illustrations, not official assets.
- Cause-of-death weights and what-if scenarios are editorial judgment and are marked as such.
