Skip to main content
株式会社オブライト
株式会社オブライト

When the Cloud Goes Down

SME BCP for Microsoft 365, Banking and Accounting Outages, with Costs

Published 2026/11/1715 min read

SME BCP for Microsoft 365, Google Workspace, bank and accounting SaaS outages: fallbacks, status pages, costs, 99.9% uptime, contact rules, checklists.

Preparing for a cloud outage comes down to four things, written on one or two sheets of paper before anything breaks: which tasks you cannot afford to lose, where you check the outage status, what you do instead for each task, and how staff communicate. Even when run by large vendors, cloud services such as Microsoft 365, Google Workspace, online banking and accounting software are not immune to major outages. Large outages at a major email service and at US bank online-banking services were reported recently. A day will come when your work stops through no fault of your own. This article is for owners and office managers at small businesses without IT staff, and covers how to prepare, including rough costs.

Why small businesses are hit harder by cloud outages

The cloud saves you from running your own servers and cuts maintenance work. The flip side is that you cannot fix it yourself when it stops. With your own server you can call a contractor and push for recovery; with a cloud outage you wait for the provider. Small businesses are especially exposed for these reasons.

- Work is concentrated in one service: many companies run email, calendar, file sharing, chat and web meetings on Microsoft 365 or Google Workspace. One outage can remove both communication and documents at once
- No fallback is defined: without spare lines or procedures like large firms, "just wait" becomes the only option
- No in-house expert: telling an outage from your own misconfiguration takes time, and needless operations can make things worse
- Money and client contact are involved: late payments, invoices and quote replies damage client trust directly

BCP (business continuity planning) does not mean writing a grand document. It means deciding in advance who does what, with which tool, and how far work continues when something stops. A cloud outage is a legitimate scenario alongside earthquakes and fires. For protecting against data loss, see cloud backup and disaster recovery. This article focuses on the case where data is safe but unreachable.

What recent large outages tell us

Recent reports included a large Microsoft email outage attributed to a software defect introduced by an earlier fix, and delays in online banking at several US banks after a technical failure at a major bank. We leave the details to the companies' statements and the news coverage; three points hold in general.

- Large vendors go down too: a provider's size and reputation do not guarantee it will not stop
- The cause is invisible from outside: users cannot choose the cause or the recovery time, and can only see what the provider publishes
- Recovery estimates change midway: after a "restoring soon" notice, it can still take hours. Plan so that work continues while you wait, not on the assumption that you wait

Step 1: List the tasks you cannot afford to lose

Start with a list of tasks, not systems. Instead of "email stopping is bad", be specific: "if email stops, we cannot reply to quotes or confirm orders". For each task, write what happens if it stops for half a day and for a full day.

- Communication: internal and external email, chat, phone, web meetings
- Files: sharing and viewing quotes, contracts, drawings, customer lists
- Money: transfers and deposit checks in online banking, card payments, invoicing
- Accounting and payroll: cloud accounting, payroll, timekeeping, expense claims
- Revenue-critical: booking systems, online shops, order management, inquiry forms
- Operations: inventory, CRM, e-contracts, ticketing

Then sort them into three levels: fine to stop for half a day, hurts sales or trust if stopped for a day, and tied to legal or payment deadlines. Decide fallbacks for the third level first. Payroll days and tax or social-insurance deadlines, for example, are when a banking outage hurts most. Marking such days on the calendar already clarifies your priorities.

Step 2: Fix where you check outage status

When work is disrupted, the first question is whether the problem is on your side or the service's. Bookmark the status pages the providers publish and decide who looks at them.

- Google Workspace: the Google Workspace Status Dashboard shows the status of each service such as Gmail, Drive and Meet, and can be viewed without signing in
- Microsoft 365: "Service health" in the Microsoft 365 admin center shows incidents affecting your tenant and the incident ID. It needs administrator rights, so confirm in advance who holds them
- Banks, accounting software and others: keep a list of each provider's notices page, outage page and official social account

Note that the status page itself may be unreachable. If your internet line is down, no page will open, so check from a smartphone on mobile data. Asking several colleagues to check on their own phones is a simple, effective way to separate causes. If other sites load, a phone on another line shows the same error, and other users of the service seem affected, an outage on the service side is likely.

Step 3: Decide the fallback for each task

A fallback is worthless unless it is ready to use beforehand. Signing up for a new service and handing accounts to everyone after the outage starts is not realistic. The table below lists main tasks, fallbacks and preparation; adapt it to your own services.

TaskTypical servicesFallback when it stopsPreparation
Internal communicationMicrosoft 365 (Teams, Outlook), Google Workspace, SlackSwitch to a different vendor: LINE, SMS, phoneKeep everyone's mobile numbers and an emergency group
Email to clientsExchange Online, GmailA separate company-use free mailbox (not personal), phone, faxKeep a key-client contact list on paper and on phones
File sharingOneDrive, SharePoint, Google DriveKeep the latest versions synced to the PC for offline useTurn on offline access; keep copies of quote and contract templates elsewhere
Web meetingsTeams, Google Meet, ZoomA different meeting service, or a conference callKeep two or more meeting services usable, even on free tiers
Transfers and deposit checksOnline bankingBranch, ATM, phone transfer request, another bank accountKnow payment dates and transfer early with margin; hold a second bank account
Accounting and invoicingCloud accounting, invoicing servicesRecord temporarily on paper or a spreadsheet, enter after recoveryPrepare invoice and journal templates in a spreadsheet; export monthly
Timekeeping and payrollCloud timekeeping, payrollPaper time cards or a shared noteExport timekeeping data monthly
Booking and ordersBooking system, online shopTake orders by phone or messaging apps, paper intake sheetsPrint an intake procedure and handwritten forms
PaymentsCard and QR paymentsCash, bank transfer, another payment methodContract a second payment method (one terminal is enough for small shops)

The key idea of this table is to escape to a different company's service. When Microsoft email stops, Teams from the same company may stop too, since the same cause can take down related services. Keep at least one communication channel whose operator and network path are different, such as LINE or the phone. If you use consumer apps like LINE for business, define in advance what information may be shared.

For files, this is about the latest copy at hand, before backups. Even when the cloud is down, a synced copy on the PC lets you write quotes and review documents. A wrong sync setting can leave needed files missing, so once a month, disconnect from the internet and check that key files open. For the mechanisms and costs of protecting the data itself, see Microsoft 365 and Google Workspace backup costs.

Step 4: Rough cost of preparing

Most fallbacks are free or cheap. The figures below are general guides and vary by company size, headcount and contracts. Check each vendor's official pricing or get a quote for real amounts.

PreparationRough costNote
Emergency LINE group, SMS or phone tree0 to a few thousand yen (printing)Keep mobile numbers current; check monthly
Offline contact list and procedures (paper plus phone)0 to a few thousand yenList clients, banks, maintenance firms, providers and contract numbers
Spare meeting or chat service accounts0 to several hundred to low thousands of yen per user per monthFree tiers often suffice; a few paid seats are enough
Mobile connectivity backup (mobile router or phone tethering)Device plus a few thousand yen per monthWorks when your own line is down; check the data cap
Second bank account and second payment methodAccount fees roughly 0 to a few hundred yen; payment terminals have separate setup costs and feesAn escape route when transfers or payments stop
Email and file backup SaaSRoughly several hundred to a few thousand yen per user per monthVaries widely by product and plan; see the article above
Outsourcing procedure writingRoughly tens of thousands to a few hundred thousand yenUnnecessary if you can write it yourself; an outside review helps catch gaps

You do not need everything at once. Start with what is free (contact list, communication rules, procedures, checking offline sync) and spend ten minutes once or twice a month verifying it. Then add paid measures for the tasks where an outage costs most (transfers, orders, bookings). Comparing the loss against the cost of the safeguard is the same logic used in maintenance contracts; see how to read maintenance contracts and SLAs.

A beginner's guide to reading an SLA (uptime)

Cloud contracts and official sites state an SLA (service level agreement) with figures such as "99.9% uptime". The figure makes sense once you calculate how long the service may be down. Taking a month as 30 days (43,200 minutes), the maximum downtime is as follows.

UptimeAllowed downtime per month (30 days)Allowed downtime per year (365 days)
99%about 7 h 12 min (432 min)about 3 days 15 hours
99.5%about 3 h 36 min (216 min)about 1 day 19 hours
99.9% (three nines)about 43 min (43.2 min)about 8 h 46 min
99.95%about 21.6 minabout 4 h 23 min
99.99% (four nines)about 4.3 min (4.32 min)about 52 min

For example, Microsoft 365 and Google Workspace use 99.9% per month as the benchmark in their official terms (check each vendor's official SLA for the exact conditions). That means up to about 43 minutes of downtime a month is within the contract. For tasks that cannot tolerate 43 minutes, such as taking orders or making transfers, do not rely on the SLA figure alone; have a fallback. When reading an SLA, keep these points in mind.

- What counts as downtime differs by contract: often only a full service failure counts, not slowness or partial loss of features, and each service is generally calculated separately
- An SLA is not a guarantee of no downtime: usually it only gives compensation (service credits) such as a partial fee reduction or extended term when the benchmark is missed. It is not damages
- Compensation is not automatic: many services require you to apply within a deadline. Keep the date, time and incident ID
- Consumer or free plans may have no SLA: check that your plan is covered
- Fit with your usage: even at 99.9% monthly, a single outage during peak business hours hurts more than the number suggests

An SLA is material for deciding how much downtime to design your work around. A realistic goal is to keep working even if a service is down for about an hour once a month.

Template: internal communication rules for an outage

What causes chaos in an outage is not the operation but the communication. If everyone hunts for information alone and the same question flies around many times, response slows down. Put rules like these on one sheet of A4 and hand it out. Feel free to use them as they are.

- Appoint one coordinator: whoever first notices a possible outage does not retry on their own judgment, but contacts the coordinator (office manager, owner) from a smartphone. Name a deputy for when the coordinator is away
- Three checks first: (1) is it the same on a line other than yours (phone mobile data), (2) does the provider's status page show anything, (3) do others have the same symptom
- Once judged an outage, notify everyone by a different channel: in the LINE group or by SMS, write "Service X is down. Until recovery, do Y. Next update at HH:MM." Always include the time of the next update
- Contact clients individually, only those likely to be delayed: by phone or another mailbox, say "replies will be delayed due to a system outage." Do not try to notify everyone and cause confusion
- Record work on paper or an offline spreadsheet: keep deposits, orders and inquiries in a form you can enter after recovery
- Do not: re-enter passwords repeatedly (it can lock the account), change settings during an outage, or spread unverified information on social media
- After recovery, first check operation and unsent mail: then enter the records and spend five minutes noting improvements for next time
- Keep a log: start and recovery times, affected tasks, actions and time taken. It also helps when applying for SLA compensation

Checklist to complete before an outage

- Listed the tasks you cannot afford to lose, with the effect of a half-day and full-day stop
- Bookmarked the providers' status pages (Google Workspace Status Dashboard, Microsoft 365 service health, etc.)
- At least two people hold Microsoft 365 admin center rights, so checks work when one is absent
- Everyone's mobile numbers and an emergency LINE group or SMS tree exist
- Contacts for clients, banks and maintenance firms are kept on paper and on phones
- Verified with the internet off that key files open offline
- Know payment dates for payroll, purchases and taxes, and transfer early
- Confirmed how to transfer via branch, ATM or phone (or hold a second account)
- At least one alternative meeting service and one alternative payment method are usable
- Put communication rules on one sheet and handed them out
- Printed and stored paper procedures and handwritten intake forms
- Ran a drill once or twice a year, assuming email is down for half a day
- Checked the uptime in the SLA of each service and how to claim compensation

It does not have to be fully ticked. With more than half done, confusion during an outage drops considerably compared with deciding nothing. Clear the rest one item per month.

FAQ

If Microsoft 365 or Google Workspace goes down, should we switch to another service right away?

No. Outages are usually resolved within hours, so switching tends to be an overreaction. What you need is not a switch but fallbacks that let work continue meanwhile: another communication channel, offline files and paper procedures. Consider switching only when problems recur or no improvement is shown.

Can the provider compensate us for the loss from an outage?

Generally, compensation is capped at what the SLA defines when the benchmark is missed, such as a fee reduction or extended term (service credits), not damages for your loss. You often must apply within a deadline. The actual terms depend on your contract and the official SLA, so check each service's terms.

Is it acceptable to use consumer apps like LINE as a backup channel?

It works for emergency contact, provided you set internal rules in advance, such as not sending client information, personal data or contract details. Because accounts are outside company control, handle information carefully. Limiting it to outage notices and procedures is the safe choice.

What if online banking is down on payroll day?

Check beforehand the options each bank offers, such as branch, ATM or phone transfer requests; process payments a few days ahead for margin; and hold a second bank account. Operate early when payroll or tax deadlines are near. Ask your bank for the specific procedures.

Can a company with no IT staff make a BCP?

Yes. The four steps in this article (list tasks, fix status checks, set fallbacks, set communication rules) can be done with paper and a smartphone. Start by choosing three critical tasks and setting a fallback for each. If unsure, ask your IT maintenance contractor or an adviser to review the procedures once.

How often should we run a drill?

Once or twice a year, spending 10 to 30 minutes on a scenario such as email being unavailable for half a day, to see whether people follow the communication rules. Check the contact list about every six months for staff changes and new phone numbers.

株式会社オブライト

Oflight Inc. is a Tokyo-based IT company providing software development, digital signage, web development, network systems, and IoT/AR/VR technology services.

Feel free to contact us

Contact Us