After 28 Hours of Bluehost Downtime
Bluehost-hosted VPS websites including this blog have just come up after having been down for more than a day. This is huge. Many of their customers—not me, but many—depend on their websites for their livelihood. Real people are being really hurt.
Bluehost’s response has been an interesting case study in crisis management, and how not to do it.
(I wrote this post around 3 pm, but I couldn’t post it, obviously, until the site came back up.)
One of the first principles in a service crisis like this is to do all you can to communicate, to connect, and do as much as possible by all means to retain customers’ trust.
One communication stream failed by way of poor planning. Bluehost’s own blog was down for part of this time, I’m told. This was a serious mistake in their planning. They should have used another hosting provider altogether, so that their blog and their service would (probably) never be down simultaneously.
Meanwhile, their Twitter information flow (http://www.twitter.com/bluehostsupport and http://www.twitter.com/bluehost, see also the .txt record here) never put forth any information at all. Tweets like this one were flowing the entire time the system was down: “No update at the moment. We are still working on getting the rest of the servers back up.”
Customers on Twitter asked over and over again for an ETA on system restoration. More than 265 times in about 24 hours, Bluehost responded with variations of, “We are still working on getting accounts back up. They are coming back up, but we do not have an ETA.” Another 285-plus times their tweets included the phrase, “as soon as…” or “ASAP.”
Each one of those 550 or so instances was an answer (of sorts) to a request for an ETA. That was at least one request every three minutes.
One customer wrote, “?? Our server is still down, I know CPR, will that help? Seriously is there an ETA? Clients need to know.” Everyone was asking for an ETA. Last night someone even said, “Would you please give us an ETA—even if it’s not true?”
There was never, throughout the entire period, any more ETA information offered than what I’ve indicated here.
Especially revealing was one Bluehost tweet that said, “No official ETA yet (engineers are too busy fixing the problem to get back to us) but we’ll post updates ASAP!” It’s as if they didn’t recognize that customers had two needs, not just one: we needed our sites back up, but we were also asking for information.
Nearly 160 times they told customers to go to their Facebook page for more complete updates, but the Facebook page was updated only five times.
Now everyone knows (or should know) that Bluehost might really not have known exactly when it would be resolved. What I have trouble believing is that they never had the slightest clue whatsoever. They probably knew from the beginning, “It will be at least an hour.” Early on they might have known, “It will be at least four hours.” Or something like that.
So when they say, “We don’t know the ETA yet,” what customers are hearing is, “We may not know exactly when it will be fixed, but we know more than we’re telling”—which is a great way to destroy customers’ trust.
Check out the Twitter stream for the last day or so, and you’ll see more people expressing that they were upset by the lack of any ETA information than by actual downtime.
Meanwhile when their blog came back up, there was not a hint of a word of an apology, or at least not as of this writing. This is really bad for trust, too.
The downtime hurts Bluehost’s reputation badly, but they’ve multiplying the damage considerably, through poor communications.
This is not the way to manage a crisis.
I understand that inMotion VPS hosting has a great reputation…