One outage, 55 languages: how the world told the Microsoft story
Lesson from:
4,556 articles, 2,272 outlets and four very different stories about the same Microsoft 365 failure
4,556 articles from 2,272 outlets in 55 languages: that is how much news one Microsoft 365 outage made in its first day, as counted by LSN News, my news engine, by 17:45 UTC on Friday, 13 November. But the world did not tell one story. In Britain it was about councils, trains and GP surgeries. In America it was “Is Outlook down?”. In Spain, France and Romania it was a global collapse. And the gloomiest headlines were written in Russian.
Contents · 12 sections
TL;DR #
- LSN News counted 4,556 articles from 2,272 outlets in 55 languages on the 12–13 November Microsoft 365 outage, by 17:45 UTC on Friday.
- The first headline arrived 44 minutes after the first mass user reports, from Brazil and Spain, before any US outlet.
- About 70% of outlets wrote their own copy (1,594 independent voices). No network had more than 3% of the story.
- The UK told it as a public-services failure, the US as “Is Outlook down?”, Spain, France and Romania as a global collapse.
- Russian-language coverage had the most negative tone, which may be about word choice rather than slant.
Same servers, same failure, same status page. Very different news. Here is what the numbers show, and what they can’t.
The event, in two lines #
Microsoft dates the start of impact to about 16:17 UTC on 12 November, according to NBC News quoting the company. Users noticed at scale around 17:50 UTC: Outage.com logged its first Outlook reports at 17:52, and Downdetector peaked at roughly 130,000 to 140,000 Outlook reports. Microsoft posted its first public acknowledgement at 18:08 and opened its formal advisory at 18:26.
Mail, Teams, SharePoint, OneDrive and Copilot were hit in more than 70 countries. As of Friday evening Microsoft spoke of “broad recovery and stability”, with no final “resolved” notice I could see. Its preliminary explanation is a change (a network capacity upgrade) followed by servers failing health checks and the rest being overloaded. The full post-incident review is not out yet.
That is the event. The rest of this post is about how it was told.
44 minutes to the first headline #
The first articles LSN News captured arrived at 18:36 UTC: Brazil’s O Tempo (“Outlook caiu?”, “Is Outlook down?”) and Spain’s 20minutos (“Caída de Outlook … a nivel mundial”). That is 44 minutes after the first mass user reports, 28 minutes after Microsoft’s first post and 10 minutes after its advisory.
Two things surprised me. First, Spanish and Brazilian consumer-tech desks beat every American outlet; the first US pieces (CNET, Variety, a Fox station in Seattle) landed at 19:00. Second, almost all of the lag is on the publishers’ side. A journalist needs about half an hour to confirm and write. My own pipeline added roughly five to six minutes, which is how often its main feed refreshes, and it was not slowed by the outage at all.
The hourly ramp: three waves around the planet #
Counting the articles in the story hour by hour gives a clear shape:
- Thursday evening, the Americas wave. 3 articles in the 18:00 hour, 112 at 19:00, then about 200 an hour until 22:00.
- The night trough. Down to 77 in the 03:00 hour, while Australia, Japan, Singapore and India were living through their working day without email.
- Friday, Europe wakes up. 258 at 07:00, over 300 an hour through mid-morning, and a peak of 448 articles in the 12:00 hour.
The biggest wave was not the outage itself. It was the second day: the “it’s still broken” stories, the local stories about who was hit, and the explainers. If you only watched Thursday night’s news, you saw less than a third of it.
Many outlets, many voices #
Big numbers in news can be misleading. One wire story copied onto 300 websites is still one story. So LSN News also counts independent voices: outlets that carry the same headline count as one.
For this outage the answer was 1,594 independent voices out of 2,272 outlets, about 70%. That is high for a story this size. The largest single network of outlets in the story, a group of Australian regional papers, accounted for 67 outlets, under 3% of the total. The most-copied headline, “Is Outlook down? Widespread outage hits Microsoft email and other apps”, appeared on 163 outlets.
In plain words: most newsrooms wrote their own copy. Nobody steered this story. That matters for the rest of this post, because the differences between countries are real editorial choices, not one script translated 55 times.
Britain: a story about public services #
The UK had 917 articles from 206 outlets, about 4.5 articles per outlet, the densest coverage of any country. Read the most-repeated British headlines and a theme jumps out:
- “Councils and public services hit by global Microsoft outage” (118 outlets)
- “Rail firms warn of possible issues with emailed tickets after Microsoft outage” (91)
- “Scottish Government activates resilience operations after Microsoft outage” (40)
Then came GPs, pharmacies and police forces asking the public to phone instead of email. Much of this was local newspaper syndication: one group’s regional titles each ran the local angle. But the frame was consistent. In Britain, the outage was a failure of the state’s plumbing.
America: “Is Outlook down?” #
The US had the most outlets, 673, but the frame was different. The single most repeated headline in the whole story was a service-journalism question, “Is Outlook down?”, carried by 163 outlets, many of them local TV stations. “Is Microsoft still down? Thousands report issues with apps” followed on 49.
That is consumer journalism: is it me, is it them, when will it be back. It is useful, and it is a very different picture from the British one. Only about 20% of US headlines called the outage “global” or “worldwide”.
Spain, France, Romania: the whole world went dark #
In Spain 77% of headlines put “global” or “mundial” right in the title. In France it was 72%, in Romania 65%, in Greece 56%:
- “Microsoft Outlook sufre una caída a nivel mundial” (Spanish, 9 outlets)
- “Une panne mondiale majeure touche la messagerie Outlook de Microsoft” (French, 10)
- “Pană de proporții la Microsoft: Outlook, afectat în întreaga lume” (Romanian)
- “Παγκόσμιο μπλακ άουτ στη Microsoft” (“Global blackout at Microsoft”, Greek)
My reading, and it is a reading, not a measurement: for an audience far from Redmond and far from the US working day, the most newsworthy fact is scale. “It’s global” is the hook.
Asia-Pacific had its own frame again. In Australia, Japan, Singapore and the Philippines the stories were about the state: the spy agency and federal police losing email, courts delayed, and Australia’s Home Affairs Minister Tony Burke activating a priority-restoration agreement, according to ABC News.
Tone: Russian coverage was the gloomiest #
LSN News gives each article a tone score, the kind GDELT computes: negative numbers mean gloomier wording. Everyone was negative. But not equally. Among languages with at least 20 articles:
- Russian: −7.1 (71 articles)
- Ukrainian: −6.3; German: −5.9; Polish: −5.2
- French: −4.0; Spanish: −3.5; English: −3.1; Danish: −2.6
(Finnish scored −9.6, but on only 21 articles, so I would not read much into it.)
What this does not show is hostility. Tone scores count words. The usual Russian word for an outage, сбой, means a failure or breakdown, and headlines like “Масштабный сбой затронул Outlook, Teams, Copilot” (“Large-scale failure hit Outlook, Teams, Copilot”) score badly on word lists without being slanted. I have not read enough of the articles to say more. Treat the tone gap as a question, not a finding.
What this data can and can’t tell you #
Some honest limits:
- Counts are a snapshot. The story kept growing: by 18:30 UTC Friday it held 4,662 articles from 2,301 outlets. The numbers above are frozen at 17:45.
- Country framing comes from headlines, matched with rough multilingual word patterns. Italian titles, for example, say “in tutto il mondo”, which my pattern only partly catches.
- LSN News sees what it collects. It reads about 13,000 sources. That is a lot, but it is not every newsroom on Earth, and paywalled or slow sites arrive late.
- Independent voices are measured by identical headlines and known outlet networks. Two outlets rewriting the same agency copy with different headlines count as two voices.
Frequently asked questions #
How many news articles covered the Microsoft outage? #
LSN News counted 4,556 articles from 2,272 outlets in 55 languages by 17:45 UTC on 13 November, and the story was still growing (4,662 articles by 18:30).
How fast did the news pick up the outage? #
The first articles LSN News captured were published at 18:30 UTC and arrived at 18:36, 44 minutes after the first mass user reports at 17:52. Most of that delay is the time newsrooms need to write.
Was the coverage copied from one source? #
No. About 70% of outlets wrote their own copy, and the largest network of outlets accounted for under 3% of the story.
Was Russian coverage hostile? #
The data does not show that. Russian-language articles had the most negative tone score, but tone counts words, and the ordinary Russian word for an outage means “failure”. I have not read enough articles to say more.
The lesson #
In my library, Walter Lippmann’s Public Opinion from 1922 has a line I keep coming back to: every newspaper “is the result of a whole series of selections as to what items shall be printed, in what position they shall be printed, how much space each shall occupy, what emphasis each shall have.”
This outage is a clean example, because the facts were the same everywhere. What differed was the selection. If you read British news you learned the state is fragile. If you read American news you learned to check Downdetector. If you read Spanish news you learned the whole world is wired to one company. All three are true.
That is the reason I built LSN News: one click on the story page shows you all of them side by side, in 55 languages, and lets you decide which picture is yours.