What is the collective noun for a number of cloud outages? A thunderstorm? A puddle of…?
A wave of major cloud downtime in the last few months has left companies that offer such online solutions scrambling to inform and appease their customers, the government, their stockholders, and other stakeholders. Cloud-based services need to take heed the subtleties of communication and the heightened need for trust that cloud computing and cloud-based hosting require. The social data during downtime demonstrates this quite dramatically.
What is the collective noun for a number of cloud outages? A thunderstorm? A puddle of…?
A wave of major cloud downtime in the last few months has left companies that offer such online solutions scrambling to inform and appease their customers, the government, their stockholders, and other stakeholders. Cloud-based services need to take heed the subtleties of communication and the heightened need for trust that cloud computing and cloud-based hosting require. The social data during downtime demonstrates this quite dramatically.
If you go ‘all in’ to the cloud and entrust your business, livelihood, video game high-scores, cherished photographs or most tedious of spreadsheets, you have made the decision that hard copies and in-house server costs are not for you. Instead, you prefer the reduced hardware clutter lifestyle and on-the-go accessibility. But what if the cloud collapses for that hour you need it, or worse, for a number of days?
Some of the most notable of these ‘disasters’ include the Amazon outage, the Sony hack, and the Skype outage of recent months. By their nature, cloud solutions are not immune to outages, hacks, etc. Cloud providers must understand how to effectively communicate with and navigate customers through a cloud emergency to provide service recovery to customers in the near term and help protect the brand in the long term. It is nuanced, and it is the need to appease and satiate a customer dependency and manage expectations that is acute.
There are a number of critical periods in a cloud outage phase:
1. The onset of outage itself
2. The company disclosure of the outage (as opposed to user-declared)
3. The crisis management and accompanying activity thereafter.
Get it right and the ship is steadied once more, impact to users is nominal, volumes of conversation remain low and their sentiment remains positive. Our insights team shared this great piece of research with me:
Amazon
Day 1 – Outage
Day 2 (7000 mentions) – Amazon declares issue
Impact: Chatter and negative sentiment increase 10x
Day 10 (2500 mentions) –Amazon issues 5,700 word document detailing how the outage occurred, an apology and credit to all customers
Impact: Overall chatter dropped off by more than 75% the next day
Sony Playstation Network
Day 1 –Outage
Day 8 (20,000 mentions) – Sony announces the first hack
Impact: Chatter reaches its highest point, but drops by 25% the day after
Day 13 (4000 mentions) – Sony holds a press conference announcing a new security position and a “welcome back” promotion
Impact: Overall chatter dropped within 2 days
Skype
Day 1 – Outage
Day 1 (8500 mentions) – Skype releases tweets, blog posts and a press release saying they are aware of the outage and are fixing it
Impact: Chatter dropped by 50% the next day, before the issue was even resolved
Lessons of a collapsed cloud?
It’s not necessarily about WHAT you say but WHEN you say it. Amazon released a very verbose, technical, and off-putting document explaining the outage, which was poorly received, but total and negative chatter about the brand dropped off anyway. The consumers wanted closure, and they got it. The same was true with Sony. Negative commentary dropped off after they held a press conference to explain what happened. The key is to follow Skype’s crisis management model—tell them what is going on right away, keep them updated, and reach out to them through various media outlets. If you get it right, it could be argued that rapid and clear lines of communication are a key strength of the cloud when adversity strikes.
Thanks to Eric Nau and Kara Clark for their assistance.