Skip to content
5WPR
Get in touch
Ragan PR Daily Awards5W is a 2026 PR Agency of the Year FinalistRagan's PR Daily Awards · Read the announcement →

5W RESEARCH · Published November 12, 2026

How Four AI Engines Answered Questions About the Outlook Outage

5WPR tested 24 prompts three times on four AI engines during the November 12, 2026 Microsoft Outlook outage and recorded 288 answers, brand mentions and cited sources.

5WPR ran 24 prompts three times each across Perplexity, ChatGPT, Gemini and Claude on November 12, 2026, recording 288 AI answers during the Microsoft Outlook outage. Microsoft was named in 70.1% of answers to Microsoft-specific prompts and in 52.8% of all answers, but the engines diverged: Perplexity named Microsoft in 79.2% of its answers and a Claude API model in 1.4%.

By Ronn Torossian, founder and chairman of 5WPR. Last updated: November 12, 2026. Incident details reflect reporting from the afternoon of November 12 and may have changed since.

What are the key findings of the audit?

The audit found that what an AI engine said about the Outlook outage depended on which engine was asked. The key results are below.

  • Microsoft was named in 152 of 288 answers (52.8%), in 143 of 204 answers to the 17 Microsoft-specific prompts (70.1%), and in 9 of 84 answers to the 7 general monitoring prompts (10.7%).
  • Perplexity named Microsoft in 57 of 72 answers, ChatGPT in 47, Gemini in 47 and the Claude API model in 1.
  • Apple appeared in 69 answers (24.0%) and Zoho in 54 (18.8%), mostly in ChatGPT and Gemini answers.
  • StatusGator was the most cited publisher, in 44 of 287 answers with retrieval evidence (15.3%), ahead of AP News at 36 (12.5%).
  • Microsoft-owned domains were cited in 57 of 287 answers with retrieval evidence across the full test (19.9%).
  • Gemini was the only engine to include Microsoft's incident identifier, EX1550543, in 20 of its 72 answers (27.8%).

What happened during the Outlook outage?

On November 12, 2026, Microsoft said it was investigating reports of users having trouble accessing Exchange Online mailboxes through multiple connection methods in the Americas. The company pointed administrators to incident EX1550543 in the Microsoft 365 admin center, according to an Associated Press report carried by WOSU.

ABC News reported that Microsoft first pointed to infrastructure under elevated load across multiple regions. It later said the issue could affect Microsoft 365 connectivity for users in any region.

For communications teams, the outage raised a second question. A customer who asks an AI engine what is happening receives an answer the company did not write, and that answer can name competitors, monitoring services or news outlets. This audit recorded what four engines returned.

How did the four AI engines differ on naming Microsoft?

The four engines differed sharply in how often they named Microsoft: Perplexity in 57 of 72 answers, ChatGPT and Gemini in 47 each, and the Claude API model in one.

Table 1. Brand appearance rate by engine, top 12 brands. Source: 5WPR analysis using Profound's tools. Each cell uses that engine's 72 answers as the denominator.
BrandPerplexity (sonar-pro)ChatGPTGeminiClaude (API)
Microsoft Corporation79.2%65.3%65.3%1.4%
Apple Inc.11.1%38.9%43.1%2.8%
Zoho Corporation11.1%19.4%41.7%2.8%
Downdetector37.5%11.1%9.7%0%
Fastmail2.8%11.1%27.8%2.8%
Yahoo Inc.4.2%9.7%26.4%2.8%
UptimeRobot20.8%19.4%2.8%0%
Amazon.com, Inc.0%34.7%6.9%0%
Amazon Web Services23.6%12.5%5.6%0%
Proton AG2.8%8.3%29.2%0%
Google LLC12.5%16.7%6.9%2.8%
Salesforce, Inc.5.6%18.1%13.9%0%

The Claude result needs context. The engine in this audit was Anthropic's claude-sonnet-4 model, called through the API with Anthropic's web search tool, so it does not describe the Claude chat product. All 72 of its answers carried web retrieval evidence, yet it named Microsoft once, and it named Apple, Zoho, Fastmail and Yahoo twice each.

The audit counted recorded brand mentions and did not review answer text. It therefore cannot say whether the Claude model answered with product names such as Outlook or Exchange instead of the parent company. Matched-answer review is the step that would separate a difference in what the engines said from a difference in how brand names were counted.

Why does the headline figure change with the prompt mix?

The 52.8% overall rate blends two different groups of prompts. Seventeen prompts asked directly about Microsoft service status, reliability, troubleshooting or news, and Microsoft was named in 143 of those 204 answers (70.1%, calculated from the per-prompt results in Table 4).

The other seven prompts asked about outage tracking and monitoring tools in general. Microsoft was named in only 9 of those 84 answers (10.7%), and Better Stack, UptimeRobot, Datadog and Downdetector led those prompts instead.

Which brands appeared besides Microsoft?

Apple, Zoho and Downdetector were the next most frequently named brands, each driven by different engines. Apple was named in 69 of 288 answers, Zoho in 54 and Downdetector in 42.

Table 2. Full leaderboard of the 25 named entities. Source: 5WPR analysis using Profound's tools. Appearance uses 288 answers, owned-domain citations use the 287 answers with retrieval evidence, and intervals are 95% Wilson intervals.
#BrandNamed inAppearance rate95% intervalOwned-domain citations
1Microsoft Corporation152/28852.8%47% to 58.5%57/287 (19.9%)
2Apple Inc.69/28824.0%19.4% to 29.2%9/287 (3.1%)
3Zoho Corporation54/28818.8%14.7% to 23.7%4/287 (1.4%)
4Downdetector42/28814.6%11% to 19.1%8/287 (2.8%)
5Fastmail32/28811.1%8% to 15.3%3/287 (1.0%)
6Yahoo Inc.31/28810.8%7.7% to 14.9%23/287 (8.0%)
7UptimeRobot31/28810.8%7.7% to 14.9%35/287 (12.2%)
8Amazon.com, Inc.30/28810.4%7.4% to 14.5%6/287 (2.1%)
9Amazon Web Services30/28810.4%7.4% to 14.5%5/287 (1.7%)
10Proton AG29/28810.1%7.1% to 14.1%6/287 (2.1%)
11Google LLC28/2889.7%6.8% to 13.7%1/287 (0.3%)
12Salesforce, Inc.27/2889.4%6.5% to 13.3%3/287 (1.0%)
13Better Stack26/2889.0%6.2% to 12.9%9/287 (3.1%)
14Cisco Systems, Inc.25/2888.7%5.9% to 12.5%4/287 (1.4%)
15Datadog25/2888.7%5.9% to 12.5%6/287 (2.1%)
16Cloudflare, Inc.22/2887.6%5.1% to 11.3%1/287 (0.3%)
17Oracle Corporation20/2886.9%4.5% to 10.5%1/287 (0.3%)
18EX1550543 (Microsoft incident ID)20/2886.9%4.5% to 10.5%Domain not established, excluded
19New Relic19/2886.6%4.3% to 10.1%6/287 (2.1%)
20SolarWinds Corporation19/2886.6%4.3% to 10.1%8/287 (2.8%)
21Meta Platforms, Inc.18/2886.3%4% to 9.7%3/287 (1.0%)
22Zoom Video Communications, Inc.18/2886.3%4% to 9.7%0/287 (0%)
23Slack18/2886.3%4% to 9.7%0/287 (0%)
24Atlassian Corporation18/2886.3%4% to 9.7%7/287 (2.4%)
25StatusGator15/2885.2%3.2% to 8.4%45/287 (15.7%)

Microsoft's interval of 47% to 58.5% sits well above second-place Apple. Apple (19.4% to 29.2%) and Zoho (14.7% to 23.7%) overlap, so the audit does not support a claim that one was more visible than the other. Downdetector, Fastmail, Yahoo and UptimeRobot also overlap.

Microsoft appeared in nine of the 12 answers to "Is Outlook down right now?" and Apple appeared in seven of those same 12. Across the full sample, Gemini named Zoho in 30 answers and Fastmail in 20. These figures count mentions of any kind, including comparisons, references and suggested alternatives, and the audit did not check why Apple was named.

Which engine surfaced Microsoft's incident identifier?

Only Gemini surfaced Microsoft's incident identifier, EX1550543, in 20 of its 72 answers (27.8%). ChatGPT, Perplexity and the Claude API model did not include it. Microsoft used that identifier to direct administrators to the incident record in the Microsoft 365 admin center.

Where did the AI answers get their information?

The engines drew on status monitors, news outlets, reference sites and community forums, and StatusGator was the most cited individual publisher. It appeared in 44 of the 287 answers with retrieval evidence (15.3%).

Table 3. Most-cited publishers. Source: 5WPR analysis using Profound's tools. Each domain counts once per answer, and an answer could cite several domains.
PublisherAnswers citing itShare of 287 answers
statusgator.com4415.3%
apnews.com3612.5%
variety.com3512.2%
en.wikipedia.org3411.8%
mashable.com3411.8%
reddit.com3211.1%
learn.microsoft.com3110.8%
uptimerobot.com3110.8%
windowscentral.com2910.1%
tomsguide.com289.8%
cnet.com279.4%
isdown.app258.7%

Microsoft-owned domains were cited in 57 of 287 answers with retrieval evidence across the full test (19.9%). Microsoft Learn appeared in 31 answers and microsoft.com in about 7% of answers. The audit did not compute an outage-only citation rate, because the 287 answers include the seven general monitoring prompts.

Some counts differ slightly between the publisher and brand views. StatusGator shows 44 answers as a publisher and 45 as a brand's owned domain, and UptimeRobot shows 31 and 35, probably because the brand view counts additional subdomains.

What did each prompt return?

Microsoft led 17 of the 24 prompts, and the seven general monitoring prompts were led by monitoring and observability brands. Table 4 lists the leading brand for every prompt, with 12 answers per prompt.

Table 4. Leading brand and runners-up by prompt. Source: 5WPR analysis using Profound's tools. Each prompt has 12 answers (three trials on four engines).
PromptLeading brandLeader rateRunners-up
Is Outlook down right now?Microsoft75%Apple 58.3%, Zoho 33.3%, Fastmail 33.3%
Is Microsoft 365 down today?Microsoft75%Apple 50%, Meta 33.3%, Amazon.com 25%
Is Microsoft Teams down?Microsoft66.7%Downdetector 25%, Salesforce 25%, Cisco 25%
Why is Outlook not working today?Microsoft75%Apple 58.3%, Zoho 41.7%, Yahoo 41.7%
Is there a Microsoft outage today?Microsoft75%Apple 33.3%, Amazon.com 33.3%, AWS 33.3%
Microsoft Outlook outage November 12 2026Microsoft66.7%EX1550543 25%, Apple 16.7%, Zoho 16.7%
When will Outlook be back up?Microsoft83.3%Apple 50%, Amazon.com 33.3%, Downdetector 25%
Is Exchange Online down?Microsoft75%Zoho 50%, Fastmail 50%, Proton 41.7%
Is Microsoft 365 reliable?Microsoft75%Apple 75%, Zoho 75%, Google 33.3%
How many outages has Microsoft had in 2026?Microsoft75%Cloudflare 50%, AWS 41.7%, Oracle 41.7%
Has Microsoft responded to the Outlook outage?Microsoft75%Apple 41.7%, Zoho 25%, Fastmail 25%
What caused the Microsoft Outlook outage?Microsoft66.7%Apple 33.3%, Zoho 25%, Downdetector 25%
What should I do when Outlook is down?Microsoft58.3%Apple 50%, Zoho 41.7%, Fastmail 41.7%
How do I check Microsoft 365 service status?Microsoft75%Cisco 33.3%, Salesforce 25%, Datadog 25%
How do I email my team if Outlook is down?Google50%Microsoft 41.7%, Slack 41.7%, Apple 33.3%
Microsoft outage news todayMicrosoft75%Amazon.com 33.3%, AWS 33.3%, Apple 25%
Which companies are affected by the Microsoft outage?Microsoft58.3%AWS 25%, Cloudflare 16.7%
best outage tracking toolsBetter Stack58.3%UptimeRobot 50%, Datadog 41.7%, Downdetector 33.3%
most reliable cloud service status tracker websiteDowndetector41.7%StatusGator 33.3%, UptimeRobot 16.7%, Atlassian 16.7%
What are the best free tools to check if major cloud software services are experiencing widespread downtime?Downdetector58.3%UptimeRobot 50%, Microsoft 33.3%, AWS 33.3%
Which business uptime monitoring platforms provide real-time alerts when enterprise office software and cloud email servers go down?Datadog50%SolarWinds 41.7%, UptimeRobot 33.3%, Better Stack 25%
check server outage beginnersUptimeRobot50%Downdetector 25%, AWS 25%, Better Stack 25%
where to subscribe to live cloud incident tracking servicesBetter Stack41.7%Atlassian 41.7%, Datadog 16.7%, New Relic 16.7%
What are the top synthetic monitoring tools to keep online services running smoothly during peak holiday retail events?Datadog50%New Relic 50%, SolarWinds 41.7%, Zoho 25%

What should communications teams do during an outage?

Communications teams should publish one timestamped update that serves both customers and administrators, because AI answers draw on whichever sources state the details most clearly. These steps follow from the source mix above. They are 5WPR's suggestions based on the data, and the audit did not test whether they change what engines say.

  1. State which services are affected and which are not.
  2. Tell users what they can do right now, such as which workaround or alternate access path to use.
  3. Give the time of the next update, so a monitoring service or reporter can show when the information was last confirmed.
  4. Name the incident identifier and where administrators can find the technical record.
  5. Send reporters and status monitors the same timestamped details that appear on the company page, so each source can be checked against the others as the incident develops.

How was the audit run?

On November 12, 2026, 5WPR ran 24 prompts three times each on four engines, producing 288 attempted and 288 returned answers, of which 287 carried retrieval evidence. Seventeen prompts covered Microsoft service status, reliability, troubleshooting and news. Seven covered broader outage tracking and monitoring tools.

Table 5. Engines and systems measured. Source: 5WPR analysis using Profound's tools.
EngineModel or systemRetrievalReturnedWith retrieval evidence
Perplexitysonar-proNative Perplexity web search7272 (100%)
ChatGPTOpenAI gpt-5.6-terraOpenAI web search tool7272 (100%)
GeminiGoogle gemini-3.8-flashGoogle Search grounding7271 (98.6%)
ClaudeAnthropic claude-sonnet-4 via APIAnthropic web search tool7272 (100%)

Brand mentions use all 288 answers as the denominator, and citation rates use the 287 answers with retrieval evidence. Product lines were rolled up to parent brands using a registry fixed before scoring. Forty further names in the answers could not be resolved to a brand and carry no metrics. Publisher identity could not be established for 18 retrieval receipts, which are disclosed instead of attributed.

What are the limits of these findings?

The audit is a one-day snapshot of these prompts and engine configurations, and it describes the answers returned rather than what customers saw. It did not poll users, grade answer accuracy, read answer text for product names, or measure how often the public asked each question.

Each engine uses a different model, retrieval tool and answer format, so a gap between two engines can reflect system design as much as brand visibility. Where intervals overlap, the audit does not claim one brand was more visible than another. One of the 25 entities, the incident identifier, has no official domain and is excluded from citation metrics. Related studies are listed in the 5WPR research library.

Research disclosure: 5WPR conducted this audit using Profound's tools. Everything-PR News LLC, which publishes a companion news report on the audit, is owned by a shared interest group that includes Ronn Torossian, who founded and chairs 5WPR.

Get in touch

Let's build your next chapter.

Tell us what you're working on. A senior strategist will respond within one business day.

Email
info@5wpr.com
Phone
212.999.5585
Offices
Miami · HQ100 SE 2nd Street, Floor 38
Miami, FL 33131
Tampa110 South 12th Street
Tampa, FL 33602
New York469 7th Avenue, Floor 8
New York, NY 10018