Home › Topics › Is It Us or Is It Them? Monitoring SaaS Reachability › Communicating Vendor Outages

Communicating a Third-Party Outage Without Taking or Passing the Blame


You have done the isolation work. The hosted file service is down from every site, from the phone hotspot, and from your outside check point. And the vendor's status page has finally turned yellow. The verdict is "them." Now you face a problem that is, if anything, harder than the technical one. Sixty people who cannot open their documents want to hear from IT. And nothing you say will bring the service back.


There are two easy ways to get this wrong. The first is to take the blame: "we're working on it," "we'll have it fixed soon." That language implies control you do not have. It sets you up to miss a promise you never should have made. The second is to pass it: "not our problem, it's the vendor." That is technically true and lands on users as "IT does not care." Both cost you trust, and trust is the only thing that makes the next outage easier. The message you want lives between them. It is honest about who owns the fault, fully engaged with what your users are going through, and specific about what happens next. I have sent both of the wrong ones. Neither went well.


This guide gives you that message. By the end you will have the four-part structure for the first announcement and an update rhythm that holds even when nothing changes. You will have phrases that neither own nor dismiss, plus workarounds worth offering. You will have a way to track the vendor incident so you have a record. You will also have the follow-up conversation with the service's business owner, and three templates you can paste and fill in under pressure.


In this guide

  • Why a vendor outage is a communication problem first
  • The first message: four parts, no speculation
  • Update cadence during a vendor incident
  • Language that neither owns nor dismisses
  • Giving users workarounds
  • Tracking the vendor incident for the record
  • After the outage: the follow-up questions
  • Templates: initial, update, all-clear
  • Where to go next

Why a vendor outage is a communication problem first


When your own server fails, your job is to fix it and communication is a side task. When a vendor's service fails, the fixing is happening in a building you will never enter, done by people you will never meet. Your levers are exactly three: information, workarounds, and expectations. That makes communication the main job, not the side task, and it deserves the same care you would give a firewall change.


It also helps to accept something about how users see it. To them, the CRM (the customer relationship management system) is "the thing IT gave us." They do not distinguish between a service you run and one you rent, and they should not have to. Their pain is identical either way, and IT is the face of the service whether or not IT can touch it. So the goal of every message is not to establish innocence; it is to be the calm, informed party in the room. Innocence takes care of itself once the facts are on the table. For the general craft of incident communication — audiences, channels, tone under stress — Incident Communication covers the fundamentals. This guide is about the special case where the fix is not yours to make.


The first message: four parts, no speculation


The first message should go out as soon as you have a verdict. With a good isolation playbook, that is about ten minutes after the first report. It has four parts, in this order, and nothing else.


  • 1. What is affected. In the user's terms, not yours: which service by the name they call it, which functions (web, desktop app, mobile, sync), which sites or groups. "The hosted file service is unavailable for all sites: web and the sync client both fail. Files already synced to your laptop still open." Precision here prevents half of the follow-up questions.
  • 2. What we know. Facts only, with times. Since when, confirmed how, and where the problem lives. "Failing since 09:20. We have confirmed it from three of our sites and from outside our network. The problem is on the vendor's side; the vendor has acknowledged an incident." If the vendor has not acknowledged yet, say that too: "the vendor's status page has not updated yet; our ticket is open."
  • 3. What we are doing. The concrete actions, however small. "We have opened a priority ticket. We are monitoring the service from every site so we will know the moment it recovers. The workaround below is available now."
  • 4. When the next update comes. A clock time, not "soon" and not "when we know more." "Next update at 10:00, or sooner if anything changes." This single line is what lets people stop asking.

What is missing from that list is deliberate: no cause, no estimate, no apology on the vendor's behalf, no reassurance you cannot back. You do not know why the service is down; the vendor may not either. You do not know when it will return; anyone who tells you is guessing. Speculating feels helpful and is the fastest way to be wrong in public. If the vendor gives an estimate later, you will relay it as theirs.


One practical detail: send the message through a channel that is actually working. If the outage is the email service, an email announcement is a message to nobody. Have a second channel agreed in advance. That could be the chat tool, a text-message group, the intranet home page, or a phone tree for the essential people. Know which one you will use before you need it.


Update cadence during a vendor incident


Cadence means a fixed rhythm of updates that you keep whether or not anything has changed. It matters more in a vendor outage than in one of your own, because you have so little to report. You cannot describe the fix in progress, so the rhythm itself is the reassurance. A sensible default is every thirty minutes for a Tier 1 service (one the business stops without). For Tier 2 (painful, but with a workaround for a day), use every hour. Adjust the cadence to match the vendor's own update rhythm when they have one. The tiers come from the dependency inventory, which is where the decision belongs: made once, in calm, not during the outage.


"No change" is a valid update and must be sent. "10:00 update: the file service is still unavailable. The vendor reports the cause is identified and a fix is in progress with no estimate yet. Workaround unchanged; next update at 10:30." That tells people exactly what they need: that you are still there, still watching, and still on schedule. Silence, by contrast, is read as "IT forgot" or "IT gave up." It restarts the flood of individual questions the first message was meant to stop.


At Northgate Retail the first time the hosted stock-lookup system failed, IT said nothing for fifty minutes, because there was nothing to say. The help desk logged sixty-one calls, most of them asking the same question. Two store managers drove to head office to ask it in person. The second time, the message went out at minute eleven: it is the vendor. Here is the evidence. Here is the workaround. Next update at ten. The help desk logged four calls. That outage was longer than the first one, and the review described it as "well handled", which was entirely about the message.


Every update restates all four parts briefly, because people join the incident late and will not scroll back. When the vendor's status moves from "investigating" to "identified" to "monitoring," relay each step in plain words, attributed: "the vendor says." When the vendor gives an estimate, pass it on with the attribution intact and a caveat. "The vendor estimates recovery by 11:00. We will confirm from our own checks before calling it fixed." Then honor the caveat. The all-clear goes out only when your own checks from every site agree. It names what may still need attention, such as restarting a sync client.


Language that neither owns nor dismisses


Most of the trust you win or lose in a vendor outage comes down to a few phrases. The pattern is consistent: attribute the fault accurately, attach yourself to the effort, and speak to the user's situation rather than your own.


Instead ofSayWhy
"We're fixing it.""The vendor is working on it. We are tracking their progress and checking from our side every minute."You are not fixing it; claiming so means owning a timeline you do not control.
"It's not our problem; contact the vendor.""The problem is on the vendor's side, so we cannot repair it directly. Here is what we can do in the meantime."Accurate about ownership, but keeps IT in the room and moves straight to help.
"It should be back shortly.""The vendor has not given an estimate. Next update at 10:30 regardless."An estimate you invented becomes a broken promise in thirty minutes.
"Sorry for any inconvenience.""We know this is landing in the middle of month-end close. If you are blocked on something specific, reply and we will find a path."Generic apology reads as a form letter; specific acknowledgment reads as attention.
"The vendor messed up again.""This is the vendor's second incident this quarter; we will raise it in the review with the service owner."Blame is satisfying and useless; a recorded pattern is useful.
"The IdP's token endpoint is returning 503s.""The sign-in service every app uses is failing, which is why several apps are affected at once."Users need the shape of the problem, not the protocol.
"Everything is back.""Our checks from every site show the service working since 11:04. If your sync client still shows an error, restart it."Specific, evidence-based, and honest about the loose ends.

Two smaller habits. Use "we" for what IT is doing and the service's everyday name for the vendor. So the message reads as a team update rather than a legal disclaimer. And answer the question people are actually thinking: "why can't we just switch to something else?" Do it before someone asks in the all-hands channel. "Switching a core service is a project, not a workaround. It is one of the questions we will bring to the review."


Tip: write the first message before you need it. A pre-written draft with blanks for the service, the time, and the next-update time takes thirty seconds to fill in. It reads far better than anything composed while three people stand at your desk. The templates at the end of this guide are meant to be exactly that draft.


Giving users workarounds


A workaround is the most concrete form of help you can offer during a vendor outage. Users remember it long after they forget the outage itself. Prepare the list in calm times, per service, from your dependency inventory. The middle of an incident is a poor moment to discover which client has an offline mode.


Service downWorkarounds to offerWarn against
Email serviceDesktop clients usually keep cached mail readable; drafts can be written and will send on recovery. Use chat or phone for anything urgent.Repeatedly re-entering passwords; it often triggers lockouts that outlast the outage.
Hosted file serviceFiles already synced to laptops still open; edits will sync later. Work locally and save often.Sending large files around by email or pasting them into unapproved sharing tools.
CRMThe mobile app sometimes uses a different path and may still work. Keep notes locally and enter them afterward.Building a spreadsheet copy that will need reconciling for a week.
Chat or meeting toolPhone bridge for scheduled meetings; email for announcements; the second channel agreed in advance.Assuming everyone saw a message posted in the tool that is down.
Identity providerSessions already signed in often keep working; do not sign out. Services that use local accounts remain available.Signing out "to fix it." A signed-in session is the only one you have.
Ticketing toolA designated mailbox or chat channel for reports; log them into the tool afterward.Losing the reports entirely because nobody wrote them down.

For the small number of people whose work truly cannot wait, a phone hotspot or a spare connection sometimes helps. That is when the problem is regional or on the road between you and the vendor. It never helps when the vendor itself is down. Know which case you are in before offering it, or you will spend the outage tethering laptops that still cannot reach anything.


Tracking the vendor incident for the record


While the incident runs, keep a log. It is the raw material for the review and the evidence for any service-credit claim. A service credit is the refund a vendor owes when it misses its availability commitment. The log also gives you a memory you will not otherwise have. Someone may ask in two months, "how long was the file service down in that outage?" The log has five kinds of entries, all with times. Include the vendor's updates, quoted, and your own measurements, from your checks and from manual tests. Record user impact, as best you can count it. Log every message you sent and every workaround or change you made. It fits comfortably in a ticket's notes or a plain text file.


Incident: hosted file service unavailable        Vendor ticket: 4471982
09:20  our office check fails (web + sync); branch and outside checks fail at 09:21
09:24  first user report (2 tickets, main office)
09:31  verdict: vendor side (fails from hotspot and outside point; DNS/IdP/other services OK)
09:33  vendor ticket opened, priority high, evidence attached
09:35  first message sent (chat + intranet); next update promised 10:00
09:52  vendor status page: "investigating degraded performance, file operations"
10:00  update sent: no change; ~40 users affected across 3 sites
10:18  vendor: "identified, fix in progress, no ETA"
10:30  update sent: relayed vendor status; workaround reminder
11:04  office check green; branch green 11:05; outside green 11:04
11:10  manual test: web OK, sync client needed restart on 2 of 3 test laptops
11:12  all-clear sent, with sync-client restart note
11:40  vendor status page: "resolved"; post-incident report requested
Impact: ~1h44m from our first failure to our first green; vendor declared 09:52-11:40

Notice the two different durations at the bottom. Your measured outage and the vendor's declared outage will rarely match. And the gap is not a complaint; it is the most useful fact in the review. The vendor's post-incident report, when it arrives, goes into the same record.


After the outage: the follow-up questions


A vendor outage is not over when the service returns. It is over when the organization has decided whether it is willing to live through the same thing again. That decision belongs to the service's business owner, not to IT, but IT brings the facts. Book fifteen to thirty minutes within a week, while memory is fresh. Come with the incident log, the timeline of your messages, and the vendor's report if it has arrived. (Memory fades faster than the vendor's report arrives.) The questions to settle:


  • Is a service credit owed? Check the agreement for an availability commitment and a claim process; many require the customer to ask within a window, with evidence. Your log and your check history are the evidence. Even a small credit is worth claiming, because the claim itself puts the incident on the vendor's record of your account.
  • Does this service need a fallback? Not every service does. For the ones that stopped work entirely, ask what a fallback would look like and roughly what it would cost. That could be an offline capability, a second provider for one function, or an alternate channel. The owner decides whether the risk is worth the money; your job is to make the trade-off visible.
  • Was the tier right? If a "Tier 2" service produced a Tier 1 reaction, promote it in the inventory. Give it the checks and the communication plan that go with the new tier.
  • Did we know before users did? If the first ticket beat the first alert, the checks need work. If the alert came first but the message went out late, the process needs work.
  • What does the vendor's report change? A single unlucky failure and a pattern of the same failure call for different responses. If the report is vague, or never arrives, that is information about the vendor too.
  • What do we change now? One or two concrete actions with owners: an inventory update, a new check, a workaround document, a contract question for the next renewal. Record them, or the review evaporates.

Keep the tone of that meeting the same as the tone of the messages. The vendor failed, the business felt it, and the question is what to do about the risk. Nobody at the table is on trial, least of all you.


Templates: initial, update, all-clear


Copy these into wherever you keep runbooks (the written procedures you follow under pressure), fill in the brackets, and adjust the voice to match your organization. They are short on purpose; a message people actually read beats a thorough one they skim. Short also fits on a phone screen, which is where most of them will be read.


INITIAL
Subject: [Service name] unavailable - vendor incident - next update [time]

[Service name] is currently unavailable for [all sites / list]. [Functions affected: web, desktop app, mobile, sync].
Since [time], confirmed from our sites and from outside our network. The problem is on the vendor's side;
[the vendor has acknowledged an incident / our ticket is open and their status page has not updated yet].
We have opened a priority ticket and are monitoring from every site so we will know the moment it recovers.
In the meantime: [workaround]. Please avoid [thing that makes it worse].
Next update at [time], or sooner if anything changes.

UPDATE
Subject: [Service name] - update [time] - still unavailable

Still unavailable for [scope]. The vendor reports: "[quoted status]" [with / without] an estimate.
[If estimate: The vendor estimates recovery by [time]; we will confirm from our own checks before calling it fixed.]
Workaround unchanged: [workaround]. [Number] people have reported being blocked; reply if you are one of them
and need a specific path.
Next update at [time].

ALL-CLEAR
Subject: [Service name] - restored [time]

Our checks from every site show [service name] working since [time]. [Known loose ends: restart the sync client
if it still shows an error / queued emails will send on their own / re-enter anything saved locally.]
The vendor has declared the incident resolved; we have requested their report and will review it with [owner].
Thank you for your patience, and for the reports - they helped us confirm the scope quickly.

Send the all-clear only after the reset checks pass from every vantage point, and never on the strength of the vendor's green dot alone. An all-clear you have to retract costs more trust than a late one.


Where to go next


You now have the whole shape of a vendor outage from the communication side. It starts with a first message with four parts and no guesses, plus a cadence that holds through "no change." It uses language that keeps you honest and present, and workarounds prepared in advance. It includes a log that becomes evidence and a review that turns the outage into a decision. Each piece rests on the rest of Is It Us or Is It Them? Monitoring SaaS Reachability. The inventory names the services and their owners. The checks give you the timestamps. The status-page guide tells you how to read the vendor's words. And the playbook produces the verdict you are communicating. Nothing you say will bring the service back. Everything you say decides whether anyone believes the next message, and that is worth ten minutes of care.


Put this into practice with SysPeep Monitor

The message is yours to write, but the timeline behind it should not be. SysPeep Monitor records when each check failed and recovered from the central service and from every remote probe. It alerts you by email and other channels before the first ticket arrives. It keeps the alert history you will quote in the vendor ticket, the all-clear, and the review. The free trial installs on any Windows machine in minutes.

Download Free TrialLearn About SysPeep