{{ searchResult.published_at | date:'d MMMM yyyy' }}

Loading ...
Loading ...

Enter a search term such as “mobile analytics” or browse our content using the filters above.

No_results

That’s not only a poor Scrabble score but we also couldn’t find any results matching “”.
Check your spelling or try broadening your search.

Logo_distressed

Sorry about this, there is a problem with our search at the moment.
Please try again later.

While many online retailers will be counting the profits from increased sales on ‘Mega Monday’ there is a sobering lesson on how not to run a discount promotion from Ebuyer.com.

The electronics retailer joined the discount trend by offering a range of sale items for just £1, including laptops and PCs. However, the website couldn't handle the extra traffic that this offer generated. 

Ebuyer.com was down for several hours, before eventually coming back online at around 2pm. By then, angry consumers had started flooding Facebook and Twitter with complaints and abuse.

Unfortunately the increased traffic sent by the £1 bargains caused the site’s servers to crash, even before the sale went live at 11am.

Ebuyer managing director Armando Sanchez posted a response admitting there has been a "degree of disappointment over this sale," but didn't go as far as to apologise.

Fabio Torlini, VP of Cloud at web hosting company Rackspace, said the outage will have done significant damage to Ebuyer's brand:

For me it's like if Harrods had a massive sale and had people queuing outside but then didn't open the shop. It's likely that people will look elsewhere and they'll probably feel a bit bitter about it, it's a massive faux pas.

Research shows that 38% of consumers will abandon a website if it takes more than 10 seconds to load, and nearly half of those surveyed (45%) said that if they believe a website is too slow they will turn to the retailer's competitors instead.

Torlini said websites can easily avoid outages by testing how much traffic they can cope with before launching a campaign. Cloud technology also allows sites to be more flexible and bring in more resources if they are hit by peaks in traffic.

This isn’t the first example of a retailer struggling to cope with increased traffic. In 2008 Debenhams’ site crashed for two days during a sale and causing customers to turn to its rivals instead. Also, US e-commerce sites, including Victoria's Secret, experienced some downtime thanks to the extra Black Friday traffic. 

The results of a website outage can be devastating for e-tailers, but it is avoidable. In the case of busy Christmas shopping days, or sales and promotions, retailers should be able to pre-empt the extra traffic that will hit their websites. 

For example, retailers can do some load testing to see how much traffic the site can handle before it falls over, then take steps to deal with this well in advance of any anticipated traffic spikes. 

Failing to do this means sales are lost to competitors, and annoying hundreds (or thousands) of customers. Ebuyer may well be wondering whether this £1 sale was worth the effort.

David Moth

Published 29 November, 2011 by David Moth @ Econsultancy

David Moth is Editor and Head of Social at Econsultancy. You can follow him on Twitter or connect via Google+ and LinkedIn

1690 more posts from this author

Comments (7)

Comment
No-profile-pic
Save or Cancel
Avatar-blank-50x50

Pete Duncanson

Is it me or is this just an advert for Rackspace? The stock answer of "put it in the cloud" is not all that simple to do with a mature site, easy on a green field site but not something that is already in place and stable.

Personally I think this was caused by the usual oversights/clashes of over reaching management and under funded/valued IT.

about 5 years ago

David Moth

David Moth, Editor & Head of Social at EconsultancyStaff

Pete, the idea was simply to get comment from somebody who has experience in web hosting to add some context to the issue, not to advertise Rackspace.

about 5 years ago

Avatar-blank-50x50

Jenni

I don't think Rackspace are really in a position to be giving an opinion. They host a website which immediately fell over (and was down for longer than Ebuyer) when it was featured on Dragon's Den a few months ago...despite Rackspace apparently being informed well in advance and supposedly taking measures to handle the volume of traffic to the site.

about 5 years ago

Avatar-blank-50x50

Pete Duncanson

Our take on the cock up and the possible technical and political causes, even got a reply from eBuyer too.

http://offroadcode.com/blog/2011/11/28/the-ebuyer-%C2%A31-sale-cock-up/

about 5 years ago

Stephen Thair

Stephen Thair, Director at Seriti Consulting

.
.

Frustrating that in this day and age this is still happening!

Ok, lets assume that ebuyer.com did do what they've been saying on the web and that

[quote]"Unfortunately even when planning ahead and expecting increased traffic (which we were) the levels of visitors to the site were even higher than those we had anticipated and so we experienced problems with the site." [unquote]

So taking that at face value they did their load testing, assessed their capacity and decided that "yes, given marketing projections we should be able to cope".

Ok, so what went wrong then?

(1) The marketing projections were wrong and that got far more traffic than anticipated. Not IT's fault but lots of luck passing the buck back to Marketing on that one.

(2) Not estimating enough contingency in their load calculations. Ok, so how much of a "fudge factor" did they add to Marketing's estimates and hence their capacity requirements? 20% extra? 50% extra? 200% extra? Clearly it wasn't enough because demand blew through available capacity BUT as an Web Operations Manager I know this is a real 2-edged sword. The bean-counters are generally unwilling to shell out for 200-300% of server capacity "just in case"... and the stronger you (as WebOps) push that case the more likely you get labelled as "Chicken Little" and if the sky DOESN'T fall as in the eBuyer scenario then you've just spent a lot of money "for nothing". It doesn't matter that the business wastes far more on advertising every year, or on business and liability insurance that "doesn't get used". Those are "acceptable" to the business in a way that "unused servers" are not. It's a double-standard, but we live with it every day.

(3) Failing to have a "Plan B". For me, this is the real criticism here of eBuyer and ANY e-commerce site that has variable demand. Most modern load-balancers (Cisco, F5, Netscaler etc) have the ability to throttle connections. Now your limit and USE IT - it's far better to have your site still up and X many people getting in and Y many people getting a HTTP 503 Service Unavailable message than have the traffic overwhelm the site and NO-ONE getting a good user experience.

Better still is to re-direct the users to a simple "waiting page" on a cloud server. A simple HTML page saying "sorry, we are really busy but you are currently "in a queue" and we estimate it will be "x minutes" before you get in to your great bargains (blah blah blah)". Then refresh the page with a random javascript timer and hopefully when they hit the site next time they'll get one of the open connections.

But the key is you HAVE TO HAVE A PLAN to manage connections in case of a "flash mob" on your site!

It might be:

(1) connection throttling as I have outlined above

(2) simplifying the site to reduce page weight etc and hence increase capacity

(3) offload as much as possible to a CDN or use an advanced dynamic caching server like http://aicache.com/ to reduce the number of connections to your origin servers

(4) if you are running in a virtualised environment you may be able to shuffle around resources between resource pools to ensure you have the computing power where you need it (assuming in this scenario you haven't blown through your entire capacity).

(5) "burst out to the cloud" - easier said than done, particularly when the bottleneck might be your database layer and not your web/app layer.

But the key here is to never assume that:

(a) your Marketing guys are any good at projecting traffic (they aren't),
(b) never assume your contingency is enough (it probably isn't),
(c) never assume that your load testing scenarios were realistic and hence the site will react the way you expect (because unless you do load-testing for a living, they probably aren't)

and hence having a plan to manage what happens when the # connections & requests grows beyond expectations is crucial.

Do it well and you'll be a WebOps god AND your company will make lost of money.

If you are interested in this sort of stuff please come along to our London Web Performance User Group (http://www.meetup.com/London-Web-Performance-Group/) and learn from people like www.seatwave.com on how to manage when your traffic can go up 44x TIMES in 15 seconds when TV advertising drives rabid fans to your website to buy tickets for their favourite bands!

cheers,
Steve

about 5 years ago

Avatar-blank-50x50

Tim

For me, the biggest issue here was the way they handled the problem. Yes, they messed up on the technical side, but whilst they were doing this, their marketing team kept posting spin on their social media, kept telling people how great the promotion was. All the while, their customer base (new and existing) were getting angry... angry because they couldn't get on the site, angry because they couldn't see the sale offers, angry because if they did get through and make a purchase they got an email shortly afterwards telling them they couldn't have the products they bought.

The marketing campaign leading up to the sale was a huge success. The technical implementation and management fell short of the requirements... the customer communication and management of the problem was, for me, the bit they failed on in the most appalling way.

To add insult to, well, other insults... they didn't even have the decency to come out with an apology. The post-promotion messages were all spin.

People forget about technical hitches quite quickly... they seldom forget about being treated badly. In my view, *that's* the price eBuyer will pay most dearly.

about 5 years ago

Stephen Thair

Stephen Thair, Director at Seriti Consulting

@Tim - good point.

My secret weapon for getting budget for WebOps is to make the point that "if you are spending millions on a cross-channel media campaign" (which large brands can easily do) every single pound you spend acquiring customers and driving them to a website THAT ISN'T THERE is a waste of money".

And hence maybe you should give me a little bit of that money to make sure the site *IS* there :-) It's an insurance sell, basically.

Not to mention the brand reputation impact over the longer term!

Ditto with bounce rates etc - spending money on improving performance will reduce the bounce rate which will increase conversions, but the real "cost multiplier" is that it increases the ROI on your marketing spend (and lets be honest, they always have the best budgets...)

cheers,
Steve

about 5 years ago

Comment
No-profile-pic
Save or Cancel
Daily_pulse_signup_wide

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Daily Pulse newsletter. Each weekday, you ll receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.