Cloud outages and your business: what Google's London outage teaches small firms

On Tuesday 20 October 2026, a power event hit Google's London cloud region, and a lot of familiar UK names had a bad morning. If you run a small business, you probably do not buy cloud servers directly. But your booking tool, payment page, website host or stock system may well sit on one. This post explains what happened, what it tells us, and a short list of things to do this week so the next outage costs you less.

What happened

According to Google's status updates as reported by news outlets, the incident began at 06:08 UTC on 20 October and was described as a "power event" in the London region, known as europe-west2. Google said most services were back to normal by 12:10 GMT, and promised a full incident report in the coming days.

The independent monitoring firm Parametrix measured it slightly differently. It detected the disruption at 06:03 UTC and recorded recovery at 11:39 UTC, a total of 5 hours 36 minutes. It reported that one zone, europe-west2-b, was offline for just over five hours, and that the region's two other zones saw a slight rise in errors. Google listed 38 affected products across compute, storage, databases, data processing and networking.

Reported effects included disruption at Revolut (which said customers' money and data were safe), problems with Lloyds and HSBC services, and spikes in problem reports for Rightmove, DPD and Royal Mail. Google has not confirmed which customers were affected, and reports linking individual companies to the outage were based on timing and public reports. Treat the list of names as indicative, not official.

The useful point for a small firm is Google's own advice during the incident: customers were told to move workloads to another zone or region. That is a sensible instruction if you run your own cloud setup. It is no use at all if you are a customer of someone who does, which is the position most small businesses are in.

What the reaction was

On Hacker News, the outage threads were small, with only a couple of comments each, so there was no strong mood to read there. In the wider coverage, the recurring theme was concern about how many services depend on a handful of large providers. University of Surrey professor Alan Woodward was quoted as saying that major cloud providers remain vulnerable to localised infrastructure failures, and that reliance on them is growing with the AI boom. The tone was measured rather than angry: this was treated as a reminder to plan, not a scandal.

Why a small business should care

You do not need to be on Google Cloud to be affected. Industry commentary on the outage noted that a regional failure can spread through applications, databases, sign-in systems and payment tools. In practice, that means a supplier you pay a monthly fee to can go down because a supplier of theirs went down. You may never be told which cloud sits underneath.

A few typical examples of where this bites:

• An online ordering or booking page that stops accepting orders while your shop is open.
• A card payment or checkout step that fails, so customers leave.
• A stock or EPOS system you cannot reach, so staff cannot see what is available.
• A staff rota or job management app that will not load at the start of the day.

None of this means cloud services are a bad idea. For most small firms they are more reliable and safer than a server in a cupboard. It means you should know what you depend on and have a plan for a few hours without it.

Five things to do this week

1. List your critical online tools. Write down every online service that, if it stopped for half a day, would cost you sales or stop staff working. Website, payments, bookings, email, accounting, rota, stock, phone system. Most small businesses find between five and ten.

2. Find each supplier's status page. Bookmark it, and note how to reach their support. During an incident, a status page tells you whether the fault is yours or theirs, which saves an hour of restarting routers.

3. Ask your key suppliers two plain questions. Where are you hosted, and what happens if that location fails? A good supplier can answer in a sentence, such as whether the service runs across more than one zone or region and how it recovers. If the answer is vague, that is information too.

4. Write a one page "system is down" plan. For each critical tool, note the manual fallback. Paper order pads or a spare card reader for the till. A phone number customers can ring. A saved list of today's bookings. A message you can post on social media in two minutes saying you are open and taking orders by phone. Keep a copy somewhere that does not depend on the same tools.

5. Check your backups, including where they live. The NCSC's small organisations guidance says to copy all the data your business needs to operate, including your website, emails, invoicing, documents and customer information. It suggests using online storage and an external device together for extra safety, keeping external devices disconnected when not in use, and learning how to restore a backup and checking it contains all your important data. A backup that sits on the same provider as the thing it protects will not help if that provider is down.

What this does not cover

A fallback plan is not the same as full resilience. Building a system that keeps running across several regions, with tested failover, is real engineering work with real cost, and it is not something every small business needs. Google's advice to move workloads elsewhere assumes you can do so quickly, and that depends on how the system was built in the first place. For many firms the sensible middle ground is choosing suppliers who take this seriously, and having a manual way to keep trading for a few hours.

It is also worth remembering that this event lasted hours, not days. The aim is not to eliminate downtime. It is to make sure a morning like 20 October is an inconvenience, not a lost day.

If you build or commission software

If a developer is building or running a system for you, it is reasonable to ask which cloud region it runs in, whether it spans more than one zone, how often backups are taken and when the restore was last tested. Those are not unreasonable questions, and the answers belong in your support agreement. If you want help reviewing how your website, ordering or booking system is hosted, our cloud and security team can look at it with you and tell you plainly whether it needs changing.

The short version

Cloud outages are going to happen occasionally, and you cannot prevent them. You can know what you depend on, know who to contact, keep a manual fallback ready and make sure your data is backed up somewhere independent. An hour spent on that list this week is cheap compared with a morning of lost orders.