Across the last 90 days StatusBird recorded 42 major or critical incidents, and the number of incidents a service had tells you almost nothing about how long it was unusable. Plaid had 5 incidents and 1642 min of downtime; ActiveCampaign had the same 5 incidents and 214 min. Same count, roughly eight times the damage.
StatusBird checks each of the 84 services it tracks every 2 minutes, and every figure below comes from those readings over the 90 days ending October 06, 2026, plus the monthly outage index. The full measurement rules are on the methodology page.
Does a higher incident count mean more downtime?
No, and the monthly index makes that plain. Compare May and July: four incidents versus fourteen, and the combined downtime is within a rounding margin of each other.
| Month | Major/critical incidents | Combined downtime | Services affected | Most affected |
|---|---|---|---|---|
| September 2026 | 11 | 12.6 hours | 7 | ShipStation |
| August 2026 | 16 | 40.4 hours | 9 | Plaid |
| July 2026 | 14 | 22.1 hours | 8 | Shippo |
| June 2026 | 8 | 37.0 hours | 8 | Squarespace |
| May 2026 | 4 | 21.9 hours | 4 | Square |
| April 2026 | 20 | 122.1 hours | 11 | AfterShip |
May 2026 logged 4 incidents and 21.9 hours down. July 2026 logged 14 incidents and 22.1 hours down. June had 8 incidents and 37.0 hours, while September had 11 incidents and only 12.6 hours. If you track "number of outages" as your vendor health metric, you are tracking the wrong column. Minutes are the column that maps to lost orders.
How is incident length actually distributed?
Lopsidedly. Here is every incident StatusBird recorded in the last 30 days, 16 in total, sorted by length.
| Service | Started (UTC) | Length | Worst status |
|---|---|---|---|
| Cloudflare | October 5, 2026 16:34 | 1236 minutes | major |
| Shippo | October 1, 2026 04:42 | 512 minutes | major |
| ShipStation | September 25, 2026 07:57 | 254 minutes | major |
| Zapier | September 9, 2026 17:53 | 110 minutes | major |
| BigCommerce | September 16, 2026 12:29 | 108 minutes | major |
| Plaid | September 26, 2026 03:41 | 94 minutes | major |
| BigCommerce | September 16, 2026 09:17 | 72 minutes | major |
| OpenAI | September 25, 2026 22:59 | 56 minutes | critical |
| OpenAI | October 1, 2026 18:48 | 52 minutes | major |
| Zapier | October 5, 2026 18:56 | 50 minutes | major |
| Shippo | September 24, 2026 19:25 | 30 minutes | major |
| Triple Whale | October 2, 2026 21:50 | 20 minutes | major |
| ActiveCampaign | September 28, 2026 15:07 | 14 minutes | major |
| Cloudflare | October 2, 2026 21:48 | 12 minutes | major |
| Shippo | September 17, 2026 04:49 | 12 minutes | major |
| Shippo | September 24, 2026 08:52 | 8 minutes | major |
Most of these incidents cleared inside an hour. The shortest, a Shippo blip on September 24, lasted 8 minutes, which is four consecutive failed checks at a 2-minute interval. Then there is a long tail: one Cloudflare event at 1236 minutes and one Shippo event at 512 minutes, each longer than every other incident in the window put together in terms of what it does to your day.
The same shape holds over 90 days. The longest single incident on record in the window is Plaid starting August 2, 2026 at 14:52 UTC and recovering August 3 at 12:54 UTC, 1322 minutes. Plaid's average incident length is 328 min across just 5 incidents. Cloudflare's average is 346 min across 4. Shippo's average is 145 min across 13.
Conclusion to quote: services do not fail in a normal distribution. They either blink for under an hour or they fall over for most of a working day, and the second kind is what you need a plan for.
Which categories are the incidents clustering in?
Three categories carry nearly all the downtime minutes, and they are not the ones store owners usually worry about first.
| Service | Category | 90-day uptime | Incidents | Downtime | Avg incident |
|---|---|---|---|---|---|
| Shippo | shipping | 98.53% | 13 | 1894 min | 145 min |
| Plaid | payments | 98.73% | 5 | 1642 min | 328 min |
| Cloudflare | infrastructure | 98.93% | 4 | 1384 min | 346 min |
| ShipStation | shipping | 99.8% | 1 | 254 min | 254 min |
| ActiveCampaign | marketing | 99.83% | 5 | 214 min | 42 min |
| BigCommerce | ecommerce | 99.86% | 2 | 180 min | 90 min |
| Zapier | automation | 99.87% | 3 | 168 min | 56 min |
| OpenAI | ai | 99.42% | 2 | 108 min | 54 min |
Meanwhile every ad platform StatusBird tracks, Google Ads, Meta Ads, Microsoft Advertising, Pinterest Ads, Snapchat Ads and TikTok Ads, sat at 100.0% with 0 major/critical incidents for the full 90 days. So did all five marketplaces, all five review platforms and all four helpdesk tools. The risk is concentrated in label and rate APIs, bank-linking, and the CDN layer, not in the tools you log into every morning.
Note the position of the big payment names too: Stripe, PayPal, Square, Adyen, Braintree, Klarna and Authorize.net all finished the window at 100.0%. The payments downtime in the table is entirely Plaid, which sits behind bank-linked checkouts rather than card authorisation.
How do the last 30 days compare with the 90-day record?
The last 30 days carried 16 incidents, more than the 11 that the index recorded for the whole of September 2026. The difference is the first six days of October: a Shippo incident of 512 minutes on October 1, a Triple Whale incident of 20 minutes on October 2, an OpenAI incident of 52 minutes on October 1, and the Cloudflare event of 1236 minutes that started October 5 at 16:34 UTC and recovered October 6 at 13:10 UTC.
Repeat offenders stay repeat offenders. Shippo appears four times in the 30-day list and holds 13 of the 42 incidents recorded in 90 days. Cloudflare appears twice in 30 days and carries 1384 min over 90. Both are also flagged in the current risk model: Cloudflare at 90.0% chance of a non-operational reading in the next 24 hours, Shippo at 15.0%, BigCommerce at 79.6% after its two September 16 incidents.
What do you do with this pattern?
- Grade your vendors by minutes, not by incident count. Pull the downtime column for each service you depend on from its reliability page, for example Shippo at 1894 min or Plaid at 1642 min, and rank from there.
- Write a fallback only for the long-tail services. Anything with an average incident above two hours, meaning Plaid at 328 min and Cloudflare at 346 min, needs a documented alternative: a second label provider configured in your shipping app, a card-only checkout path that does not require bank linking.
- Split your alert channels by expected length. Route Cloudflare, Shippo and Plaid alerts to SMS so you see them in the first minutes, and send short-fuse services like ActiveCampaign, average 42 min, to Slack or Teams where they can wait.
- Stop treating a 12-minute blip as an incident. Four of the 16 incidents in the last 30 days were 20 minutes or shorter. Set your escalation to trigger a customer-facing notice only past a threshold you choose, for example 30 minutes, so you are not posting banners for events like the Cloudflare incident of 12 minutes on October 2.
If you want these patterns to reach you before a customer emails about it, StatusBird watches these services every 2 minutes and sends SMS, email, Slack, Teams or Discord alerts when one stops responding. You can set up alerts here.
Frequently asked questions
Which service caused the most downtime in the last 90 days?
Shippo, with 1894 min of downtime across 13 major or critical incidents and 98.53% uptime. Plaid is second at 1642 min from only 5 incidents, and Cloudflare third at 1384 min from 4.
Is a month with more incidents always a worse month?
No. May 2026 had 4 incidents and 21.9 hours of combined downtime, while July 2026 had 14 incidents and 22.1 hours. September 2026 had 11 incidents but only 12.6 hours.
How long is a typical outage?
Most are short. In the last 30 days the lengths ranged from 8 minutes for Shippo on September 24 to 1236 minutes for Cloudflare starting October 5, with the majority clearing inside an hour. Over 90 days the longest single incident was Plaid at 1322 minutes on August 2.
Which service is most likely to have a problem today?
As of October 06, 2026 the risk model puts Cloudflare highest at a 90.0% chance of a non-operational reading in the next 24 hours, followed by BigCommerce at 79.6% and Vercel at 49.4%.
Do payment processors fail more than other categories?
Not the card processors. Stripe, PayPal, Square, Adyen, Braintree, Klarna and Authorize.net all recorded 100.0% uptime with 0 major or critical incidents over the 90 days. The payments downtime in the window came from Plaid, 1642 min, and Affirm, 22 min.