Slack Outage History: 1,459 Status Entries, and None Called an Outage Since November 2025

Slack outage history, as of November 4, 2026: Slack’s status site lists 1,459 incidents, outages and notices since September 2013, 78 of them labeled Outage. The last Outage label is November 10, 2025. A failure on May 27, 2026 hit users globally for about 7 h 24 min, yet that month still shows 100% uptime.

The pages ranking for this query stop well short of that. Slack’s own calendar shows 12 months, StatusGator starts at January 2025, and Slack’s documented history API returned 28 entries when we called it, reaching back to May 14, 2026. We walked all 4,844 of Slack’s public day pages instead, then added the monthly uptime figures Slack prints on its calendar, four versions of its SLA wording, and the causes it states. Everything below comes from Slack’s site, its engineering blog, its SEC filings or archived copies of those pages.

The last ten entries

Length is the window Slack states in its note. An asterisk marks first post to last live note, where no clock times are given. Every entry since November 11, 2025 is labeled Incident or Notice. Dates link to Slack’s day page.

The pace has slowed: 66 entries in 2026 through Nov 4, 2026, against 89 in the same months of 2025, 102 in 2024 and 161 in 2018. July to October 2026 had 11 entries in four months. We cannot tell from the record alone whether that means fewer failures or fewer posts, but the ID numbers in Slack’s feed offer one check: IDs 1553 to 1582 cover the period, and only 2 of those 30 numbers are missing from the public list, so a wave of unpublished entries is not showing up.

Which part of Slack breaks: every service tag, by year

Entries carrying each tag, as tagged on Slack’s site today. One entry can carry several tags. *2026 runs to Nov 4, 2026. Tags begin in August 2017; 324 of the 372 untagged entries predate them. Slack has remapped tags: an April 2021 copy of its API shows Calls, Connections, Posts/Files and Link Previews, which now read Huddles, Connectivity and Files, or are gone. That is why 2017 entries about Slack calls carry Huddles. Workflows and Canvases tags first appear in 2023.

Connectivity is the most common tag at 329 entries, followed by Messaging at 296 and Apps/Integrations/APIs at 235. The mix has moved: Connectivity peaked at 78 entries in 2018 and sits at 11 so far in 2026, while Workflows has 49 entries since its tag first appears in 2023. Of the 1,099 tagged non-maintenance entries, 470 (43%) touch no Connectivity, Messaging or Login/SSO tag at all. Those are feature faults such as search, file previews or integrations, the class Slack’s old SLA excluded from downtime.

Slack incidents per month and year

Non-maintenance entries by day of first post. Bars alternate color by year. Red dots mark months with at least one entry labeled Outage; the last dot is November 2025.

*2026 is partial. 12 Maintenance entries, the last on Mar 19, 2015, are excluded here and kept in the ledger. Medians are not like-for-like across eras: most lengths before 2018 and in 2026 are first post to last post, while most from 2019 to 2025 are windows Slack states.

Outage is a label Slack chooses, and it hasn’t used it since November 2025

Slack’s API documentation lists three types: incident, notice and outage. 78 of 1,459 entries have ever carried Outage. There were 7 in 2025 and 0 in 2026 so far. The 359-day gap since Nov 10, 2025 is the longest in the record; the previous longest was 337 days, from February 2022 to January 2023.

The gap does not line up with quiet. On January 8, 2026, some users had “a total inability to send and receive messages” for about 3 h 25 min after a routine backend configuration job. On May 27, 2026, an infrastructure change caused what Slack calls “a cascading scheduling failure” with “widespread service impact across all regions and clients”, and its note says “This issue affected Slack users globally.” Both are labeled Incident. Compare that with November 10, 2025, a 15 min connection dip, labeled Outage.

That matters because the label seems to drive the one reliability number Slack prints. The status calendar shows an uptime percentage on each month. Across the 7 calendar copies we could read, covering 38 months, 11 months show less than 100%, and every one of them has at least one Outage entry. All 11 full months from December 2025 to October 2026 show 100%, including May 2026. The latest months below 100% are November 2025 at 99.96% and October 2025 at 99.94%. The lowest in our copies is February 2025 at 98.98%, the month of the 9 h 28 min database event, a figure well short of 99.99%.

Latest 11 months below 100% in the calendar copies we hold. Implied downtime is (100 minus printed uptime) times minutes in the month, our arithmetic. 1 month shows 100% despite an Outage entry, including March 2024, an entry limited to data residency customers. Slack has not published how the figure is calculated today; its 2015 SLA said downtime depended on minutes and “the percentage of customers that were affected”, which fits printed downtime running well below stated windows.

The SLA: what Slack promised, and what the page says now

This corrects a claim still repeated across the web and on Slack’s own older posts. A Slack blog post still says “Slack is committed to 99.99% uptime, or no more than 4.3 minutes of downtime per month”, and a 2023 engineering post refers to “Slack’s 99.99% availability SLA”. The current pricing page text contains no uptime figure at all. The public document no longer backs them up, though an individual order form may. In our 99.9% uptime calculator and SLA comparison, 99.9% allows 43.8 minutes a month; 99.99% allows about 4.4 minutes, 13.1 minutes a quarter and 52.6 minutes a year.

Slack did pay on the old promise. Its September 2019 10-Q reports “approximately $8.2 million in credits to paid customers” for availability issues in the quarter ended July 31, 2019. The status record for those three months shows 70 entries, the busiest three-month stretch on file (tied with June to August 2019), including 4 labeled Outage with about 22 h 19 min of stated impact. Slack’s final 10-K warned it “may modify or reduce the amount of credits”. The public promise was gone about three years after Slack’s last 10-Q, filed in June 2021.

What the status record implies against a 99.99% budget

Minutes between the first and last clock times Slack states for each Outage-labeled entry, summed by calendar quarter (first post to last live note where no times are given). Light blue: within 13.1 minutes. Pink: up to 60. Red: up to 600. Dark red: above 600. Q1 2019 includes a 102-hour push-notification fault that Slack says hit some Grid customers only.

This is not a measurement of customer uptime. A stated window is the longest stretch in which some users were affected, often a small share, so it can overstate what any one workspace saw, and Slack’s printed figure does run lower. It also misses Incident-labeled failures such as May 2026, and so understates in the other direction. It is one defined, checkable reading of Slack’s own words. On that reading, 13 of the 19 quarters from Q4 2019 to Q2 2024, when the quarterly 99.99% wording was on the page, exceeded a 13.1-minute budget, and 5 had no Outage entry. 2025 totals 1,849 minutes against a yearly budget of 52.6 minutes; 2026 totals zero minutes on this definition, even with a 7 h 24 min global event in the year.

The longest and widest Slack incidents

Longest broad events

Ten longest Outage-labeled entries with a stated window, plus the two 2026 Incident entries above. Excludes the February 2019 Grid notification fault. The October 2025 start is our adjustment: Slack’s note says 12:13 PM PDT on October 19, 13 hours before its first status post at 1:15 AM PDT on October 20. We count first post to the stated 3:51 PM PDT end.

Slack did not name the vendor in October 2025. The timing matches the AWS us-east-1 failure of October 19 and 20, 2025, and Slack has written that most of its core runs in us-east-1. That link is our inference. For what a stretch like this costs a business, use the hourly revenue figures in our cost of downtime report.

Longest of any kind

*First post to last live note. Most of these are narrow bugs that began before Slack posted, so the date posted can be weeks after the start.

The one to read if you have retention rules is August 13, 2026. Channels with custom retention settings could not load older messages. Slack spent 14 updates and about 32 d 1 h restoring messages, edits and shared files in batches, finishing on September 14. A minutes-down figure says little about an event like that.

Widest: every entry with all nine legacy tags

13 entries carry nine or more tags, the last on Feb 22, 2022. Tags measure breadth of features, not share of users: the November 17, 2020 entry has all nine and is titled “for a very low percentage of users”. Since May 2025 many notes carry a Scope line, which says more about reach: 49 entries have one, and 23 of those describe users globally or in all regions.

How long Slack incidents last, and when they get posted

1,388 entries with a usable length: 804 from clock times Slack states, 584 from first post to last live note. 71 have neither. Red, in brackets: entries labeled Outage.

The median is 2 h 10 min, and one in ten runs past 20 h 43 min. Outage-labeled entries are shorter than the rest, with a median of 46 min: 61% are done within an hour and 70% within two. Notes on 19 of the 44 Outage entries since 2018 mention a revert or rollback.

First status post of 1,346 entries that began with a live post, in US Eastern time. Numbers shown where 20 or more. Peak: Tuesdays at 3 PM ET, 27 posts.

52% of first posts land on weekdays from 9 AM to 6 PM Eastern, a window that is 27% of the hours in a week. Only 6% land on a weekend, and Friday has 175 posts against 286 on Tuesday. This chart shows when Slack posts, not when faults begin, and people report more while they work, so we can’t separate more failures from more detection. Posts also trail the stated start: a median of 58 min since 2022, with 82% posted after the start Slack later gives. October 22, 2026 went up 2 h 43 min after the 16:08 UTC start, as one note, after it was over.

What Slack says caused its outages

We read the notes of all 44 entries labeled Outage since January 2018 and coded each by the cause Slack gives. 24 (55%) trace to Slack’s own change: 63% of the 38 entries that name a cause. The same pattern shows up in our coding of Cloudflare post-mortems (27 of 40) and Azure post incident reviews (58%).

The three latest Outage entries, all in November 2025, blame traffic spikes, a run with no other example in the set. Slack’s engineering blog gives the deeper versions:

How complete is Slack’s record?

• Every day page exists. All 4,844 dates from August 1, 2013 to Nov 4, 2026 returned a page: 1,178 with entries and 3,666 saying “Nothing to report”. The first entry is September 13, 2013. Only 2 months since 2014 are empty, both in 2014.
• The feed and the pages agree. All 28 entries in the API response match a day page on title, service tags and number of notes.
• ID numbers run ahead of entries. The highest ID is 1582, and the site shows 1,471 entries, 111 fewer (7%). Slack’s own documentation shows ID 546 on September 7, 2018, when 505 entries existed. IDs could be drafts, cancelled or deleted posts; we can’t tell which. One note, for May 31, 2026, says an entry “will be removed from the status page” because it affected one customer. As of our pull it was still listed.
• Old copies match. Internet Archive holds two earlier responses from the same API: 59 entries from Nov 2, 2020 to Apr 26, 2021, and 52 from Sep 30, 2024 to Mar 20, 2025. All 111 are still on the site with the same title and the same type label. So we found no deletions and no relabeling in those samples, and the ID gaps in them (3 and 6) already existed when the copies were taken.
• Known events are all there. The five post-mortem dates above and the 2019 credit quarter all have entries. The June 30 and September 30, 2021 events are labeled Incident, not Outage.
• Notes get edited. A January 2023 note says “We’ve edited the times in this summary”. A September 2021 note says the title and services were updated. The January 23, 2024 entry’s last note describes a February 21 issue. The calendar’s figure for May 2023 moved from 99.993% to 99.992% between copies, so closed months can change.
• Write-ups arrive late, and fewer give times. 781 entries end in a retrospective summary, 456 of them more than 6 hours after the previous note and the slowest 128 days later. Any tool that treats the last update as the end of the outage will overstate length. Clock-time windows are also getting rarer: 93 of 103 entries in 2025 state one, against 8 of 66 in 2026, even though Slack’s help page still promises an issue summary within 24 hours.

Why other Slack outage trackers disagree

StatusGator says 105 outages since January 2025; Slack’s own list has 169 entries in that span, and we can’t reconcile the two from outside. Durations differ more. For Oct 22, 2026, IncidentHub shows about 1 minute and OnlineOrNot under a minute. Slack posted one note for that event, and its text says 18 minutes. Downdetector shows user reports, a separate signal worth watching. The @SlackStatus account on X describes itself as no longer active, so don’t rely on it.

What this means for MSPs

Backup communications

Size the plan to Slack’s own record: 61% of Outage-labeled events ended within an hour, but 9 h 28 min in February 2025 and 7 h 24 min in May 2026 cover most of a working day. Write down a channel that needs neither Slack login nor Slack SSO, because Login/SSO appears on 129 entries. Slack’s account of May 2020 describes moving to “the incident Zoom” and sending a company-wide email. Clients need the same: a named bridge, an email list exported this quarter, and a phone tree for the people who approve changes. Move on-call paging off Slack-only delivery too; a tool that alerts into a channel fails when you most need it, and Slack’s own dashboards failed in January 2021.

What to tell clients

Say what you know and when you will update, without waiting for a status post that has trailed the stated start by a median 58 min. Don’t quote the status page’s uptime percentage as a measure of their experience: it showed 100% for May 2026. Don’t call an event “not an outage” because Slack’s label says Incident. GitHub, by comparison, publishes a monthly availability report itemizing its incidents; Slack’s closest public equivalent is this status list. If message history matters to a client, the August 2026 retention event is the example to raise when you review retention settings and exports.

Status feed monitoring

Poll https://slack-status.com/api/v2.0.0/current, which needs no authentication and lists all active incidents, or use the email, RSS or Atom feeds. Three practical rules from this data: filter on the services array rather than type, because a rule waiting for outage has not fired in 359 days. Treat an empty feed as weak evidence, because 8 of 66 entries in 2026 appeared only as a single note. And keep your own copy, because /history holds only about six months of entries in all three responses we checked, despite documentation that calls it a list of all past issues. The Salesforce Main Services Agreement says the services “may not be accessed for purposes of monitoring their availability, performance or functionality”, so read it before building a synthetic login probe on a client’s workspace; the status API is the route Slack offers.

Claiming SLA credits

Start with the contract. If a client’s order form has no uptime term, the public page gives commercially reasonable efforts and no credit. If an older or negotiated term applies, the March 2023 wording shows what to expect: a claim emailed with the workspace URL, within 15 calendar days of quarter end, checked against Slack’s records, and capped at 30 days of paid service. Log each event’s start, end and affected users on your side when it happens, because the 43% of tagged entries that are feature-only faults would have been excluded under that wording anyway. Then decide how your own client SLA treats Slack: an MSP that promises four nines on top of a vendor with no public number is carrying that risk itself.

Search every Slack status entry

Frequently asked questions

When was the last Slack outage?

The latest entry on Slack’s status site as of November 4, 2026 is October 22, 2026: Slack Real-Time Connection Disruption, which Slack says ran from 16:08 to 16:26 UTC. The last entry Slack labeled Outage is November 10, 2025. The largest event since then is May 27, 2026, about 7 h 24 min and global by Slack’s own note, labeled Incident.

How often does Slack go down?

Slack posted 66 status entries from January 1 to November 4, 2026, about 6.5 a month, against 89 in the same part of 2025 and a high of 161 in 2018. Most are partial: 5.3% of the 1,459 entries since 2013 carry the Outage label, 78 in all.

What was the worst Slack outage?

By stated length and reach, February 26, 2025: from 6:45 AM to 4:13 PM PST (9 h 28 min) a large percentage of users had trouble with messages, workflows, channels and login, after a database maintenance action met a caching defect. May 27, 2026 (7 h 24 min, global) and October 20, 2025 (14 h 36 min, an upstream vendor) are the next broad events.

Does Slack have a 99.99% uptime SLA?

Not on its public SLA page today. The page, effective July 11, 2024, says Salesforce uses commercially reasonable efforts to keep Slack available, with no percentage and no credit. The March 2023 version promised 99.99% a quarter with credits of 10 times the amount paid for the outage period. A signed order form can still say more, so check the contract.

Is there a Slack status history for iPhone or iOS?

21 of the 1,459 entries name iOS, iPhone or iPad in the title, and 31 mention them anywhere. The latest is October 9, 2026: Slack for iOS closing unexpectedly when opening channels, resolved by an app update after about 3 d 1 h. Search iOS in the ledger below. 24 entry titles name Android.

Why does Downdetector show a Slack outage when the status page is green?

They measure different things. Downdetector counts user problem reports. Slack posts an entry after its own review, a median 58 min after the start time it later states (entries since 2022), and 12% of 2026 entries were first posted as a single after-the-fact note. The record has no location field, and none of the 1,459 entries mention California, so a problem in one city or state may sit with an ISP rather than Slack.

Methodology

References

1. Slack System Status, History calendar and day pages. Accessed November 2026.
2. Slack, Slack Status API documentation. Accessed November 2026.
3. Slack status history API response. Accessed November 2026.
4. Slack Service Level Agreement, effective July 11, 2024. Accessed November 2026.
5. Slack Service Level Agreement, January 7, 2015 (archived by Slack). Accessed November 2026.
6. Slack Service Level Agreement, August 15, 2019 (Internet Archive). Accessed November 2026.
7. Slack Service Level Agreement, March 20, 2023 (Internet Archive capture of April 24, 2024). Accessed November 2026.
8. Slack status calendar, 2024-03-04 (Internet Archive). Accessed November 2026.
9. Slack status calendar, 2024-03-24 (Internet Archive). Accessed November 2026.
10. Slack status calendar, 2024-04-19 (Internet Archive). Accessed November 2026.
11. Slack status calendar, 2025-10-18 (Internet Archive). Accessed November 2026.
12. Slack status calendar, 2026-03-13 (Internet Archive). Accessed November 2026.
13. Slack status calendar, 2026-08-24 (Internet Archive). Accessed November 2026.
14. Slack status history API, April 28, 2021 (Internet Archive) and March 25, 2025 (Internet Archive). Accessed November 2026.
15. Slack pricing. Accessed November 2026.
16. Slack Supplemental Terms and Salesforce Main Services Agreement. Accessed November 2026.
17. Slack Technologies, Form 10-Q for the quarter ended July 31, 2019. Accessed November 2026.
18. Slack Technologies, Form 10-K for fiscal 2021. Accessed November 2026.
19. Engineering at Slack, “A Terrible, Horrible, No-Good, Very Bad Day at Slack”. Accessed November 2026.
20. Engineering at Slack, “All Hands on Deck”. Accessed November 2026.
21. Engineering at Slack, “Slack’s Outage on January 4th 2021”. Accessed November 2026.
22. Engineering at Slack, “The Case of the Recursive Resolvers”. Accessed November 2026.
23. Engineering at Slack, “Slack’s Incident on 2-22-22”. Accessed November 2026.
24. Engineering at Slack, “Slack’s Migration to a Cellular Architecture”. Accessed November 2026.
25. Slack, “Slack is custom-built for enterprise companies”. Accessed November 2026.
26. Slack Help, “Get updates on Slack’s uptime alerts”. Accessed November 2026.
27. StatusGator, Slack outage history, IncidentHub and OnlineOrNot (comparison only). Accessed November 2026.