Slashdot Mirror
Web Traffic Snarls Sites on Black Friday
eweekhickins writes "A surge of e-commerce traffic on Thanksgiving night and all day Friday apparently caught several retail giants by surprise, with Lowe's, Macys and Victoria's Secret especially hard hit. In fact, almost a third of leading retailers suffered significant slowdowns on Black Friday, according to statistics released this weekend by Keynote Competitive Research, a firm that tracks Web site performance."
Money
How do you know they didn't have huge servers - just not huge enough? (Ditto for bandwidth, etc...)
Seriously, predicting traffic is pretty much a black art. Even if you build out for what you thought would be enough, you still could get caught flatfooted.
And why don't the brick-and-mortars open up at midnight, with 3 times the normal cash registers open? Because the near riot is good for business when it gets covered in the local news. There's a reason that each store stocks 12 units of the best deal, and most of the other prices are just normal sale prices... that generates an aura of crazed shopping, and a line of 200 people who are willing to stand hours in the cold -- 188 of whom will be buying products at a profit.
Try viewing it from the other side of the fence.
...
From upper management's perspective there are dozens of great ideas - some will boost the brand, some will help fill niches that competitors are festering in, some will cement loyalty, some will increase appreciation by best customers, grow the number of best customers, decrease churn, reactivate customers, grow the customer base, provide additional channels for purchase, shift horizontally, shift vertically, decrease long term overheads, increase flexibility by reducing response time to market dynamics,
That's just off the top of my head in a minute - you can't begin to imagine how many ideas get put forward and some of those ideas are really very good. It isn't until you sit in front of your budget and realise you only have the money for 10% of the ideas that reality sets in.
Which ideas do you cut? Enough IT infrastructure for black friday, a brand refresh to counteract waning customer interest or redundant logistics infrastructure so you don't get caught paying for warehouse space while customers are going to compeditors in frustration at your stock not being in stock?
Every body I've ever talked to is convinced their project is the most important. In my (admittedly, limited) experience it is rarely IT that has the greatest strategic or tactical ROI - most of them end up over budget and under performing!
Yes, but you know how accountants are. They will take the yearly retail PROFITS and figure it is approximately 1/12th of yearly SALES, and "conclude" that retailers operate in the red 11 months or yer and turn black staring "black Friday" - as in all profits happen starting then. The monthly and quarterly picture is quite different, of course. This is the same as saying we work till May 17 (or so) every year to just pay taxes. Imagine not taking home a paycheck until almost mid-year!
The network manager, if they're any good, will, however, think in terms of peak page requests and database transactions per minute, how to minimize cross-server requests and how to tweak site assets to minimize bandwidth requirements. Unfortunately the data that he presents to the people that control the spending accounts will not be in line with what they already presume to know based on the last 5 years of sales data. At this point, either the network manager needs to know how to be a salesman or the VP that he reports to needs to trust his people to know what they were hired to know. If both conditions fail then the site is doomed on Black Friday. If either condition holds true then either the web site will be prepared to the network manager's recommendation or, and this is more likely, the following decision process will be followed. Like I said, simple.
Of course every company has a sort of corporate personality. Many of which will tend to throw more people at a problem rather than invest in equipment for the existing people to use. If you can see this happening in their stores then you can be sure that they're doing it in their web department as well. Unfortunately the best that such a practice can hope to accomplish is speeding up recovery time from equipment failures. Many companies are also completely clueless in terms of technology so they may just be doomed from the start. As an anecdotal example for the latter case, I work for a company whose website uses several cross server assets per page. This works fine from an external browser but causes several firewall password prompts per page from any of our in store terminals.
My God! It's full of eval()'s.
Excellent point. I had trouble loading newegg all weekend...where normally it is really fast. Their site already has a built in option for a high or low bandwidth version of their site.......why couldn't they just force everyone to the low version (I am assuming this would save bandwidth and server load, less queries). Yeah sure, it would be missing some fancy tools, and some flash (darn), but it seems far better than turning customers away. I wonder how many $ a site like Newegg lost for each minute of downtime this weekend.
This is an excellent point. Companies just keep sticking more fancy looking crap on their websites leaving bandwidth unchanged.
My university did this with the registration website. The added flash and java to make it look cool, except when registration time hit, all that load brought the registration server to its knees. When many students weren't able to register for several days after their allotted time, the system was reverted to the plain ol' html interface it was before. When will websites realize that just because your public has the bandwidth, doesn't mean they need to use it?
Yes it's an anecdote! Were you expecting original research in a Slashdot comment?
I was online at around 5.50am(PST) when Sears.com died. I was opening a new page to view the details of one product and that new page would not show up. Sears.com would also not work in a new page. Shortly after 6am I was able to get back into the site, only to see the prices had gone up!