Mark ở vũ trụ khácVIMark học AI
Mark ở vũ trụ khác · Another Universe Anatomy of a Business Failure · Streaming · Startups

Quibi

≤10 min

Quibi Holdings · 6 April 2020 to 1 December 2020 · Los Angeles

Whata phone-only short-episode app, shut down after about 6 months
Money raised$1.75 billion before launch day
The real questionwould real users pay for this?
Main sourcesShutdown letter of 21 Oct 2020, Eko v. Quibi court dockets, Roku press release

Source: Quibi's funding announcement (March 2020, via Advanced Television, Tubefilter, secondary source); open letter by Katzenberg and Whitman (21 Oct 2020); shutdown on or about 1 Dec 2020 (via Adweek, Tubefilter, secondary source).

Review in 1 minute

Essence

Money first$1.75 billion raised before the first user ever opened the app.
Phone onlyEpisodes of 10 minutes or less. At launch: no TV viewing, no screenshots.
Nice numbers, little moneyDownloads in the millions, but with 90 days free. Who actually paid is still disputed.
Fixed after complaintsAirPlay, Chromecast, screenshots, TV apps: all arrived after launch day.
Not just COVIDThe shutdown letter itself names "the idea" before "timing".

Source: Quibi funding announcement March 2020 (via Advanced Television 5 Mar 2020, secondary source); CTA's CES 2020 article (ces.tech); Engadget 6 Mar 2020 (price, 90-day trial, secondary source); MobileSyrup 27 May and 24 Jul 2020, Engadget/IPS 10 Jun 2020 (features added later, secondary source); open letter 21 Oct 2020.

Part 1

The question

Two people who had run DreamWorks Animation and eBay, $1.75 billion, a roster of Hollywood stars. Why did it close less than 7 months after launch? The familiar answer is "COVID". The answer in the record is: big money was spent before anyone knew whether real users would pay.

Source: total funding per Quibi's announcement, March 2020 (via Advanced Television, secondary source); launch 6 Apr 2020, shutdown announced 21 Oct 2020 (open letter). The founders' backgrounds are widely published public information.

2018 to Mar 2020The big betRaise $1.75 billion, commission shows, buy ads, all before any viewer.
→
6 Apr 2020Meeting real usersLaunch in the middle of lockdowns. Downloads came; payers are unclear.
→
21 Oct 2020StopShutdown letter: the idea was not strong enough, or the timing, probably both.
Part 2

What happened

Quibi was a private company with no public financial statements. Dates come from Quibi's own announcements, court dockets and Roku's press release; user numbers come from the press and Sensor Tower and are marked as secondary sources.

Up, then down. 2018: first round of about $1 billion. March 2020: total funding $1.75 billion. 6 April 2020: launch. 11 May 2020: Katzenberg blames everything on COVID. July 2020: Sensor Tower estimates at most 72,000 of the early sign-ups stayed to pay. 21 October 2020: shutdown letter. About 1 December 2020: streaming ends. 8 January 2021: Roku buys rights to more than 75 shows. 2018round 1 ~$1B Mar 2020total $1.75B 6 Apr · launch6 Aprlaunch 11 May "everything is coronavirus" Jul 2020≤72,000 stayed 21 Oct · shutdown letter1 Decstreaming ends 8 Jan 2021 Roku
The highest point was launch day. The curve is illustrative, not measured data: Quibi never published its paying-subscriber count.Source: 2018 round and $1.75 billion total per Advanced Television 5 Mar 2020 and Variety (secondary source, citing Quibi's announcement); launch and 90-day trial per Engadget 6 Mar 2020 (secondary source); 11 May 2020 quote per NYT via TechCrunch 12 May 2020 (secondary source); 72,000 per Sensor Tower estimate via TechCrunch 8 Jul 2020 (secondary source); open letter 21 Oct 2020; end of streaming per Tubefilter 23 Oct 2020 (secondary source); Roku press release 8 Jan 2021.
2018 · First roundAbout $1 billion, with the major Hollywood studios investing.
Jun 2019 · A promise to spend big on contentKatzenberg said the first year would see $1 billion spent on content, with production costing "up to" $100,000 per minute.
Jan 2020 · The idea unveiled at CESQuibi stands for "quick bites". Episodes of 10 minutes or less. Turnstyle technology lets you rotate between portrait and landscape without losing the shot. Whitman: "We are not shrinking TV content for a phone."
Mar 2020 · Another $750 millionTotal $1.75 billion, about a month before launch. That same month Quibi and Eko sued each other over Turnstyle.
6 Apr 2020 · Launch amid lockdowns$4.99/month with ads, $7.99 without, a 90-day free trial, about 50 shows. Phone only.
First week · About 1.7 million downloadsA figure given by Whitman. A week later, Quibi promised TV viewing "after viewers complain" (Bloomberg).
11 May 2020 · "Everything is coronavirus"Katzenberg: 3.5 million downloads, 1.3 million active users, "not close to what we wanted".
May to Jul 2020 · Adding what was missingAirPlay in late May, Chromecast in June, screenshots in July.
Jul 2020 · The 90 free days run outSensor Tower estimates at most 72,000 of roughly 910,000 people who signed up in the first 3 days stayed to pay. Quibi: wrong "by an order of magnitude".
21 Oct 2020 · Shutdown letterWind down, return cash to shareholders, look for buyers for the assets. Streaming ends on or about 1 Dec 2020.
8 Jan 2021 · Roku buys the libraryExclusive global distribution rights to more than 75 shows. Price not disclosed.
Sep 2021 · The Eko case closesBoth sides drop all claims. Turnstyle technology goes to Eko.

Source: Advanced Television 5 Mar 2020, Variety (funding, secondary source); Tubefilter 11 Jun 2019 and Business Insider Oct 2020 (content cost, secondary source); ces.tech (CES 2020); Engadget 6 Mar 2020 (price, trial, 50 shows; Tubefilter says 51, secondary source); TechCrunch (1.7 million, citing Whitman, secondary source); Bloomberg 13 Apr 2020 (TV, secondary source); TechCrunch 12 May 2020 and Tubefilter 12 May 2020 citing NYT (secondary source); MobileSyrup 27 May and 24 Jul 2020, IPS 10 Jun 2020 (secondary source); TechCrunch 8 Jul 2020 (Sensor Tower, secondary source); open letter 21 Oct 2020; Roku press release, Business Wire 8 Jan 2021; court docket 2:20-cv-02250 (terminated 24 Sep 2021).

Part 1

What happened

$1.75 billion6 Apr 202090 days freeshutdown letter
Part 2

Why

four stacked betslocked in the phonemoney first, users laterthe COVID myth
Part 3

Work lesson

riskiest assumptioncheap tests with real usersstaged funding
Part 3

Why: five layers of causes

Strategy: four bets stacked on each other

stacked assumptionsopen letter, 21 Oct 2020
Four assumptions had to be true together: people want Hollywood-quality shows, want episodes of 10 minutes or less, want to watch only on the phone, and will pay for a separate service. The shutdown letter itself questions the sum of these bets: the idea was not strong enough to justify a standalone streaming service. Hollywood showscostly, polished + ≤10 minutes"quick bites" + phone onlywatch on the go + pay separately$4.99 or $7.99 "the idea itself wasn't strong enough to justify a standalone streaming service"
Each bet sounds reasonable, but all four must hold at once. The chance that all four are right is lower than any one of them. That is why each bet should be tested cheaply with real people before money is poured into their sum.Source: open letter by Katzenberg and Whitman, 21 Oct 2020 (quoted via Tubefilter, CFO.com, Verdict); format per ces.tech (CES 2020); price per Engadget 6 Mar 2020 (secondary source).

Product: viewers locked inside the phone

features added after complaintstech press, Apr to Oct 2020
Launch day 6 April 2020: no TV viewing, no screenshots. 13 April: TV viewing promised after viewers complained. Late May: AirPlay. June: Chromecast. July: screenshots. October: TV apps. 21 October: shutdown letter. 6 Aprlaunch✗ TV✗ screenshots 13 AprTV promised "afterviewers complain" late MayAirPlay JunChromecast Julscreenshots OctTV apps 21 Octshutdown
Every feature arrived after a complaint. Short video lives on being captured, shared and talked about. At launch, Quibi blocked screenshots and ran only on phones, just as people were sitting at home next to a TV.Source: Bloomberg 13 Apr 2020; MobileSyrup 27 May 2020 (AirPlay) and 24 Jul 2020 (screenshots); IPS/Engadget 10 Jun 2020 (Chromecast); iMore 21 Oct 2020 (Apple TV app). All secondary sources.

Money and sequence: spend first, listen later

spend before learning2018 to Apr 2020
Order of commitments: 2018, about $1 billion raised; June 2019, $1 billion of first-year content promised; February 2020, ads at the Super Bowl and the Oscars; March 2020, another $750 million raised; only on 6 April 2020 did the first real users open the app. 2018~$1B raised Jun 2019$1B content pledge Feb 2020Super Bowl, Oscars Mar 2020+$750M 6 Apr 2020first realusers money committed market feedback The public record does not show how Quibi tested with users beforehand. It shows the order: money first.
Not too little money, but money too early. Once you have pledged $1 billion of content and bought the most expensive ads in America, it is hard to change course when users say "I want to watch this on my TV".Source: Advanced Television 5 Mar 2020 and Variety (funding); Tubefilter 11 Jun 2019 ($1 billion first-year content); press reports on Super Bowl and Oscars ads (no reliable spend figure, so none is given). All secondary sources.

Measurement: downloads are not customers

vanity metrics · free-trial fogApr to Jul 2020
User numbers from different sources. First week about 1.7 million downloads per Whitman. May: 3.5 million downloads and 1.3 million active users per Katzenberg. July: Quibi says more than 5.6 million downloads; Sensor Tower estimates about 4.5 million. Of about 910,000 people who signed up in the first 3 days, Sensor Tower estimates at most 72,000 stayed to pay, about 8 percent. Quibi rejects this figure. downloads (Quibi's figures) first week~1.7 million 11 May 20203.5 million · 1.3 million active Jul 2020>5.6 million stayed to pay after 90 days (Sensor Tower estimate, disputed by Quibi) ≤72,000~910,000 first-3-day sign-ups about 8% · Disney+ about 11% (7-day trial) · Quibi: wrong "by an order of magnitude"
90 free days made the numbers look good and blurred the most important question. Downloads say "people are curious". Only paying says "people need this". Quibi never published its paying count, so every number here is either an outside estimate or the company's own claim.Source: TechCrunch (1.7 million, citing Whitman); TechCrunch 12 May 2020 and Tubefilter 12 May 2020 citing NYT (3.5 million, 1.3 million); TechCrunch 8 Jul 2020 and TheWrap (Sensor Tower: 910,000, 72,000, 8%, 11%, 4.5 million; Quibi: 5.6 million, "order of magnitude"). All secondary sources.

Context: COVID and a lawsuit

timing · litigationdockets 2:20-cv-02250 and 2:20-cv-02299
Two headwinds outside the product. First, COVID lockdowns erased the waiting and commuting time Quibi targeted. Second, the Eko case: 28 January 2020 Eko demands Quibi stop using Turnstyle; 9 March 2020 Quibi sues first for a declaration of non-infringement; 10 March 2020 Eko sues; July 2020 the court denies a preliminary injunction; September 2021 both sides drop claims and Turnstyle goes to Eko. COVID, from Mar 2020 Quibi aimed at: waits, commutes Reality: at home, next to a TV a real headwind but the shutdown letter names "the idea" before "timing" Eko v. Quibi over Turnstyle 28 Jan 2020 · Eko demand letter 9 Mar · Quibi sues first (2:20-cv-02250) 10 Mar · Eko sues (2:20-cv-02299) Jul 2020 · preliminary injunction denied Sep 2021 · all claims dropped, Turnstyle to Eko the court never ruled on who was right
The headwinds were real, but they do not replace the core question. COVID erased the commuting time Quibi targeted, and Katzenberg once blamed "everything" on it. The Eko case was an allegation never decided by a court; the preliminary injunction was denied because Eko had not shown "irreparable harm", without a ruling on the merits.Source: dockets Quibi Holdings, LLC v. Interlude US, Inc. d/b/a Eko, U.S. District Court, C.D. Cal., No. 2:20-cv-02250 (filed 9 Mar 2020, terminated 24 Sep 2021) and JBF Interlude 2009 Ltd v. Quibi Holdings, LLC, No. 2:20-cv-02299 (via Patexia, Docket Alarm); injunction denial per MediaPost, IBC, The Fashion Law (secondary source); settlement per Variety via Tubefilter (secondary source); Katzenberg quote per NYT via TechCrunch 12 May 2020 (secondary source).

The idea

"Wasn't strong enough to justify a standalone streaming service": the first reason in the shutdown letter.

Timing

The second reason in the letter. The letter does not say "only".

Both

"We suspect it's been a combination of the two", per the letter as quoted by Verdict and Yahoo (wording: use with care).

Source: open letter 21 Oct 2020, quoted via Tubefilter, CFO.com, Verdict, Yahoo/AOL (secondary sources quoting the original).

Part 4

Fact check: what is true and what is myth

13 things often said about Quibi, checked against the shutdown letter, court dockets, Roku's release and the press (marked as secondary). Click each row for the reasoning.

Two explanations from the same person. On 11 May 2020 Katzenberg told the NYT: I attribute everything that has gone wrong to coronavirus, everything. On 21 October 2020 the shutdown letter signed by Katzenberg and Whitman gave two reasons: the idea was not strong enough for a standalone service, or the timing, and suspected both. 11 May 2020 · to the NYT "I attribute everything that has gone wrong to coronavirus." COVID: 100% 21 Oct 2020 · shutdown letter 1. the idea wasn't strong enough 2. or the timing · "likely both" no more "everything"
The same person, two explanations, five months apart. The bars illustrate emphasis, not a measured share.Source: NYT 11 May 2020 via TechCrunch 12 May 2020 and Hollywood Reporter (secondary source); open letter 21 Oct 2020 via Tubefilter, CFO.com, Verdict (secondary sources quoting the original).
SOLIDQuibi raised $1.75 billion before launch

Quibi announced a $750 million second round in early March 2020, for a total of $1.75 billion, about a month before launch. Many outlets agree on the figure. One mismatch: NextTV cites Whitman saying the $750 million included $400 million closed at the end of 2019 (Advanced Television 5 Mar 2020, Tubefilter 4 Mar 2020, NextTV; secondary sources).

SOLIDQuibi lasted about 6 months

Launch 6 Apr 2020, shutdown letter 21 Oct 2020, streaming ended on or about 1 Dec 2020 (open letter; Tubefilter 23 Oct 2020, secondary source).

SOLIDKatzenberg once blamed "everything" on COVID

To the NYT on 11 May 2020: "I attribute everything that has gone wrong to coronavirus. Everything. But we own it." (via TechCrunch, Hollywood Reporter; secondary sources). In the same interview he admitted the user numbers were "not close to what we wanted".

SOLIDAt launch you could not take screenshots

Screenshots arrived only in July 2020, described by Quibi as "a sort of public beta" (MobileSyrup 24 Jul 2020, secondary source).

CAUTION"Only 72,000 people paid"

This is a Sensor Tower estimate, an upper bound, and covers only the roughly 910,000 people who signed up in the first 3 days, not total subscribers. Quibi rejected it as wrong "by an order of magnitude", yet never published its own figure (TechCrunch 8 Jul 2020, secondary source).

CAUTION"You couldn't watch Quibi on TV"

True at launch. AirPlay came in late May, Chromecast in June, TV apps in October 2020, just as the company was about to close (MobileSyrup, IPS, iMore; secondary sources).

CAUTION"Quibi stole Eko's Turnstyle technology"

This was Eko's allegation, never decided on the merits. In July 2020 the court denied a preliminary injunction because Eko had not shown "irreparable harm". The case ended on 24 Sep 2021 with a settlement: all claims dropped, Turnstyle handed to Eko, no admission of wrongdoing; reportedly no money changed hands (dockets 2:20-cv-02250 and 2:20-cv-02299; Variety via Tubefilter, Cord Cutters News; secondary sources for the terms).

CAUTION"Every minute of content cost $100,000"

Katzenberg said production cost "up to" $100,000 per minute, a ceiling, not an average (Business Insider, Tubefilter; secondary sources). First-year content spend also differs between outlets: $1 billion or $1.1 billion, planned or spent.

CAUTION"Quibi spent big without ever testing with users"

The public record shows the order: funding, a $1 billion content pledge, Super Bowl ads, and only then real users; and the features users asked for all came after launch. But there are no public documents on how Quibi tested internally, so "never tested" cannot be claimed.

CAUTION"Quibi returned $350 million to investors"

The shutdown letter says it would "return cash to our shareholders". The figure of about $350 million comes from Bloomberg and the WSJ, not from the company (secondary sources).

FALSE"Quibi failed only because of COVID"

The shutdown letter itself gives two reasons, the idea first, and suspects both. What users complained about (no TV, no screenshots) were product choices made before the pandemic. COVID was a real headwind, but not the only reason.

FALSE"Roku bought Quibi"

Roku bought exclusive global distribution rights to more than 75 shows and documentaries, not the company. Turnstyle was not part of the deal and later went to Eko (Roku press release, Business Wire 8 Jan 2021; Cord Cutters News, secondary source).

FALSE"Quibi proved short video is a bad idea"

The shutdown letter doubts "a standalone streaming service" that charges money, not short video in general. Free short video on social platforms kept growing after 2020. What failed was a specific bundle of bets, not the format.

Source: open letter 21 Oct 2020; court dockets 2:20-cv-02250 and 2:20-cv-02299; Roku press release 8 Jan 2021; the articles named in each row (secondary sources).

Part 5

Verdict

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

Avoidable
4/5
Warning signs existed
4/5
Transparency afterwards
3/5
Clean shutdown
4/5
Done right when failing
  • Stopped early while cash remained, instead of burning to the last dollar
  • The shutdown letter names "the idea" as the first reason
  • Sold the library to Roku, so the shows still have viewers
Done wrong
  • Committed big money before any real users
  • Fixed the product after complaints, after launch
  • Never published its paying count, even while disputing others' numbers

Quibi did not die for lack of money or talent. It died because big money ran ahead of the users' answer.

More: why "it was COVID" is the most comfortable explanation

An outside cause nobody controls means nobody is responsible. That is why, after any failure, it tends to be the first explanation spoken. With Quibi, the founder himself said so in May 2020.

The more useful question is: without COVID, what would still be wrong? The app would still run only on phones, still block screenshots, still ask people to pay for 10-minute episodes while free short video was everywhere. By October 2020 the shutdown letter also put "the idea" before "timing".

Source for "More": NYT 11 May 2020 via TechCrunch (secondary source); open letter 21 Oct 2020; the rest is Mark's commentary.

Part 6

Work lesson: test with real users before spending big

For any big bet: a new product, a new course, a large campaign, hiring a new team, buying expensive software.

Four steps before spending big: write down the riskiest assumption; test it cheaply with real users; release money in stages, each stage needing evidence; write stop criteria in advance and measure paying and returning, not downloads. 1 · riskiest assumption"they will pay for…"one written sentence 2 · cheap, real users10 people, 1 weeknot colleagues 3 · money in stageseach stage, one proofnever all at once 4 · stop criteriameasure paying, returningnot downloads "People will love it" is not evidence that people love it.
Each step patches one of Quibi's failures. Four stacked bets, money ahead of users, features fixed after complaints, and free downloads hiding the question "will they pay?".Source: the failures come from the open letter of 21 Oct 2020 and the sources in Part 3; the steps are Mark's suggestions.
List assumptions, mark the riskiest in redWhich one, if wrong, sinks the whole project? Test that first, not the easiest one.
Real people, real behaviourOne person who pays, returns and tells a friend is worth more than ten who praise it in a meeting.
Money in stagesSplit the budget into 3 to 4 stages. The next one opens only when the previous one has written evidence.
Stop criteria written in advance"If after 8 weeks we have fewer than X paying users, we stop." Write it while calm, not when the money is nearly gone.
Check your own team (5 questions, 10 minutes)

1. Which assumption does your team's biggest project rest on? Write it as one sentence starting with "Users will…".

2. How many real users (not colleagues, not the boss) have touched it?

3. How much of the budget was spent before you heard from them?

4. Is the metric you report "curiosity" (views, downloads, free sign-ups) or "need" (paying, returning, referring)?

5. Have the project's stop criteria been written down, and who is allowed to say "stop"?

Quibi is not succeeding, likely for one of two reasons: because the idea itself wasn't strong enough to justify a standalone streaming service or because of our timing.Katzenberg and Whitman, open letter, 21 Oct 2020 (as quoted by Tubefilter, CFO.com, Verdict).
Part 7

Read next

Quibi's open letter, 21 Oct 2020Short and direct. Read it to see a company name the reasons for its own failure.
Eric Ries, The Lean Startup (2011)The build, measure, learn loop and the minimum viable product: exactly what Quibi skipped.
Rob Fitzpatrick, The Mom Test (2013)How to ask users questions that get the truth, not polite praise.
Anatomy of a Business Failure #04: New Coke (1985)Another company that did test with real people, but measured the wrong thing.
Part 8

Primary sources

Quibi was a private company: no public financial statements and no bankruptcy filing. Priority goes to the company's own documents, court dockets and the buyer's press release; user and money figures come from the press and are marked as secondary.

  1. Jeffrey Katzenberg and Meg Whitman, "An Open Letter from Quibi", Medium, 21 Oct 2020. medium.com. Read via quotations in Tubefilter (21 Oct 2020), CFO.com, Verdict, Yahoo/AOL.
  2. Court dockets: Quibi Holdings, LLC v. Interlude US, Inc. d/b/a Eko, U.S. District Court, C.D. Cal., No. 2:20-cv-02250 (filed 9 Mar 2020, terminated 24 Sep 2021); JBF Interlude 2009 Ltd v. Quibi Holdings, LLC, No. 2:20-cv-02299 (filed 10 Mar 2020); Judge Christina A. Snyder. Patexia, Docket Alarm.
  3. Roku, "Roku Acquires Quibi's Global Content Distribution Rights", Business Wire, 8 Jan 2021. businesswire.com
  4. CTA (CES), "Quibi introduces mobile-first storytelling at CES 2020", Jan 2020. ces.tech
  5. (secondary source) Advanced Television, "Quibi raises $750m second round", 5 Mar 2020; Tubefilter, 4 Mar 2020. advanced-television.com
  6. (secondary source) Engadget, "Quibi is launching on April 6th with 50 shows", 6 Mar 2020; Tubefilter, 6 Mar 2020 (51 titles). engadget.com
  7. (secondary source) Bloomberg, "Quibi to Let People Watch on Their TVs After Viewers Complain", 13 Apr 2020. bloomberg.com
  8. (secondary source) TechCrunch, 12 May 2020, citing the NYT interview of 11 May 2020; Tubefilter, 12 May 2020. techcrunch.com
  9. (secondary source) TechCrunch, "Report says Quibi lost 92% of its earliest users after free trials expired", 8 Jul 2020 (Sensor Tower figures and Quibi's response). techcrunch.com
  10. (secondary source) MobileSyrup 27 May 2020 (AirPlay) and 24 Jul 2020 (screenshots); IPS/Engadget 10 Jun 2020 (Chromecast); iMore 21 Oct 2020 (Apple TV app).
  11. (secondary source) Tubefilter, 11 Jun 2019 ($1 billion first-year content); Business Insider, Oct 2020 ("up to $100,000 per minute").
  12. (secondary source) MediaPost, IBC, The Fashion Law, Hollywood Reporter 2020 (injunction denied, 3 of 9 claims dismissed); Variety via Tubefilter and Cord Cutters News, Sep 2021 (settlement).
  13. (secondary source) Bloomberg and WSJ via Business Insider, Oct 2020 (about $350 million expected to be returned); Tubefilter, 23 Oct 2020 (streaming ends on or about 1 Dec 2020).

Note: the research environment blocked direct access to Medium and court websites, so the letter and court orders were read via several outlets' quotations; only sentences quoted identically by at least two sources are used. Super Bowl ad spend, headcount (outlets disagree) and an official paying-subscriber count have no reliable source and are left out.