Black Friday: Proof That Ecommerce Uptime Truly Affects The Bottom Line

I’m not much of an artist, but let me paint a scene for you.

It’s the day after Thanksgiving (the infamous Black Friday) and you’re trying to buy someone the most incredible Christmas gift ever: Kathryn Bigelow’s 1991 award winning film, Point Break: Pure Adrenaline Edition on DVD. Rather than dealing with the craziness of going out and fighting moms over this year’s all-the-rage gift, you decide to do all of your shopping online this year.

So you do a search on ecommerce sites to find the best price to order it online, but there’s a problem. Due to the high demand for such an amazing movie, the first site is out of stock. The second site only has the original edition, which is completely missing the exceptional bonus material included on the Pure Adrenaline Edition. You won’t settle for some b-side DVD, so you decide to check a third site.

As you click the third result, the page doesn’t load after three seconds. You refresh but much like the surfers in Point Break, you only get adrenaline when the site loads fast. Bored and growing discouraged, you move on to the fourth site.

Two of our account reps — Michael Smith and Tommy Denniston — both attended eTail in Boston recently and ended up snagging me a book of the top 500 retail web sites in the United States. The book included some awesome stats like 2010 revenue, figures on their performance, the percentage of website availability and more. With the help of our Business Analyst Raj Vysetty, Mike and Tom’s data crunching and some true grit, we were able to conclude some pretty alarming stats that helped us draw some incredible conclusions.

Here are some key highlights from 2010 that illustrate exactly how important website uptime is for your ecommerce business. (Here’s an infographic if you’re more of a visual learner.)

Our sample size was 412 of the 500 companies as that is what our source had site availability data for. Data is for 2010.

Lost Time

The median length of downtime for those 412 sites was 840 minutes.
On average, each of these sites experienced 3291 minutes of downtime last year.
The total amount of downtime they experienced all together was 1,343,643 minutes — a combined 2.5 years of downtime!

These numbers all reflect time. Now let’s look at what really matters to these ecommerce sites.

Lost Revenue

On average, each of these 412 companies lost $800,099 in revenue due to downtime.

The total amount of revenue lost due to downtime of all 412 companies was $329,640,928!

To put that in perspective, for that amount of revenue, I could buy 25,356,994 copies of Point Break: Pure Adrenaline Edition (retailed at $12.99) on DVD. This would ensure nearly everyone in the great state of Texas or continental Australia could own their own copy.

The Holidays Are Approaching

The holiday season will be here in a few months and with October knocking on our doors, now is the time for ecommerce sites to make sure their sites will not only perform fast, but stay up. While I can’t measure the individual causes of downtime, DynECT Managed DNS can definitely decrease the length and chances of that happening.

When Johnny Utah was saying goodbye to Bodhi near the end of Point Break, he yelled, “Vaya con Dios.” We’d hate to see sites say the same thing to their revenue. Let’s talk and figure out how much downtime affects you so we can help you keep that potentially lost revenue.

Dshep01

I couldn’t see the infographic (or link) for “Here’s an infographic if you’re more of a visual learner.”

http://twitter.com/Ryohara Ryan O’Hara

I posted it above it above for those that couldn’t view it.

Ron

I don’t see the graphic, either

http://twitter.com/Ryohara Ryan O’Hara

I posted it above it above for those that couldn’t view it.

random

Unsubstantiated statistics. Not something that’s going to get me to jump into business without some backup.

Source of data?
How do you determine lost revenue without knowing the products sold?
Average is high compared to mean, are there maybe some sites in the mix that are outliers (went down and never came back)?
Downtime versus maintenance time?
Root cause of outage?

Its not as simple as orders per hour = orders per hour missed. Lots of customers go away for a while then come back because of brand loyalty or uniqueness. If I prefer Amazon or want a T-shirt from Threadless it is not very likely I won’t be back to purchase later. Any stats on the spike that followed the outages?

Not that I doubt the huge outages, just that I am one of those who get irked by random use of stats in corporate blogs.

ps. “Here’s an infographic if you’re more of a visual learner.” is in there twice but only 1 has the link.

http://twitter.com/Ryohara Ryan O’Hara

Hey Gang,

My
apologies for the delay. Thanks for reading my blog. The source of the data is
from Internet Retailer Top 500 Guide. What we did was take a look at the
statistics they had for website availability (which Tom comments on where that
data came from). Tom Denniston will explain
how they determined that percentage in another comment (which he just posted on
the comments). Their data gave us a percentage of website availability,
and naturally I subtracted this by 1 to determine the % of
unavailability.

What I personally did was convert the percentage into minutes, to get the total minutes of website unavailability.

The Internet Retailer Top 500 Guide also mentioned the 2010 Sales
data for these companies. We took the
annual revenue, and divided that by the
number of minutes in a year, to determine how much revenue these companies make
per minute. Once that was determined, we
multiplied the number of minutes of downtime and the revenue per minute for
each company, and that was how we determined losses. The sample size was only
412 sites because there was no data provided for 88 of the top 500.

We looked at this data as something that is not only interesting
since it’s very rare to have information like this determined, but also worth
sharing with our readers. Two of the weaknesses of our research is not knowing the
causation of downtime, and the fact that
we did all of this this based on average revenue rather than certain times of the day. Obviously the time of day does play a factor. Amazon probably makes more money at 9AM ET than they do 6AM ET, though the times zones tend to balance some revenue for additional international customers of some of these companies.

The purpose however, is to prove that even if our numbers of off by a margin of error of even close to 10-15% (which we are putting our faith in some of the accuracy of their data), the amount of revenue lost due to downtime is still a a fascinating number that is usually higher than the cost of outsourcing DNS to an enterprise provider. We want to help eCommerce sites keep their sites up, and given the timing of this point of the year, the data seemed more relevant now rather than later. My apologies if this data seems questionable, and it’s totally understandable. I simply took the information we had, applied some basic math, and just tried to put the numbers into words for those that are interested. Thanks again for reading, and I hope the feedback and clarification helps answer some things from all of you.

Cheers,

Ryan
http://twitter.com/Ryohara Ryan O’Hara

Thanks for your post. I thought I’d help clear this up a bit.

Source of data?

See Tom Denniston and my comments here.

How do you determine lost revenue without knowing the products sold?

We don’t know what products were sold, so good point…but what we do have is their 2010 total sales revenue. We were then able to determine how much revenue these companies make a minute. This number doesn’t include peak hours, but a site that is a top 500 revenue maker should think any downtime is unacceptable.

Average is high compared to mean, are there maybe some sites in the mix that are outliers (went down and never came back)?

Great questions!!! All of the sites came back online to become available in 2010. Tom and Mike personally went to each site to check to ensure they stayed in business. However, there were sites that brought down the average. Some sites how unavailability as high as 17%, while others, including some enterprise DNS users had zero downtime.

Downtime versus maintenance time?
Root cause of outage?

We were unable to determine this…however….using a network like DynECT does give users the capability to do failover, or load balancing so they can set up a mirror network to keep things smooth sailing as they update and do maintenance. We’re big fans of redundancy here if you couldn’t tell haha.

Its
not as simple as orders per hour = orders per hour missed. Lots of
customers go away for a while then come back because of brand loyalty or
uniqueness. If I prefer Amazon or want a T-shirt from Threadless it
is not very likely I won’t be back to purchase later. Any stats on the
spike that followed the outages?

We don’t have that data unfortunately, but Tom’s write up in the comments section does explain a little of how they determined the web site unavailability.

Not that I doubt the huge outages, just that I am one of those who get irked by random use of stats in corporate blogs.

I’m right there with you Mr. Random. Hopefully our comments help clear it up a little bit more. The most important part for us is to share our numbers with our readers, hopefully start discussions with users about what causes their downtime, and also see if there is a play to work together closer.

ps. “Here’s an infographic if you’re more of a visual learner.” is in there twice but only 1 has the link.

I’ve forwarded this to our webmaster. Thanks for noticing and sharing!

Ghhoffman

As a doctoral learner, I have now been trained to send up the BS flag when statisics are not substantiated. I do not doubt the source, I would just like to see a reference that would allow me to verify the stats.

Samual Clemens (AKA Mark Twain) dais there are three kinds of lies:
1. Lies
2. Damned lies
3. Statistics

BYW, I cannot see the graphic either.

http://twitter.com/Ryohara Ryan O’Hara

I love Mark Twain, and I love people that question the unquestioned. Thanks for reading the blog. If you want to take a look at the numbers they determined…here’s a link where I believe you can purchase them. http://www.internetretailer.com/top500/

If you want to understand how the numbers were gathered, please see Tom’s comment on this page. Hopefully this transparency can help you determine whether you choose to believe or not believe the numbers we ran. We simply took the data, applied some math, and shared it with our users.

For the record, Dyn doesn’t publicly endorse this book. Tom, Mike, Raj, and I ran through these numbers on our own free time and wanted to share the data with our readers/eCommerce sites we care/know about. Thanks again for reading, and more importantly, asking for clarifications.

Here’s the link for the infographic link…please let me know if this works: http://dyn.com/infographic-how-downtime-financially-impacts-top-ecommerce-websites/

MinuteByMinute

I saw the infrographic and I can give you a LOT of leeway with the statistics as well.
In my business, downtime for phones or data/internet would “cost” us about $12k an hour – that would be “peak time”, say between 9 AM and 7PM. Compare that to what Amazon or Zappo’s or WalMart or Target or a brokerage firm/stock trader or any other top 500 retail website would do in an hour.
Time would be the only relevant statistic – I’m fairly confident in saying any US web site that is down at 3:47 AM for 8 minutes is nothing compared to being down at 9:47 AM for 8 minutes.
It abosulutely makes no difference what the source of the issue is: webserver, database, bandwidth, broken connection or nuclear destruction.
Being down is devastation. Period.
We are all dependent on so many variables – DNS, the guy using a backhoe in Los Angeles, the weather and so many other bad things you do not want to discuss.
I monitor both my home and work network. I do not monitor it out of obsession – I do it becasue my boss wants to know what we are paying for at work, and I want to know what I’m paying for at home.
A monthly report shows our websitedata connection, since we host our own site, is at 99.897% uptime. This is within the parameters of the contract we researched and signed. Home is at 99.8%. As I said the times the connection is broken is critical – if we have one user try to use the website and make a payment at 2 AM it is bad, but hopefully recoverable – I we have 342 users that cannot make a payment between 9-11 AM…well, I better not be at fault.
Outliers would help, of course, but generally, from my limited knowledge, aren’t they usually cast out of the figures but used for reference?
http://twitter.com/tddenniston Tom Denniston

By popular demand, I dug up some clarification on how these statistics were gathered. The book referenced is the Internet Retailer Top 500 Guide, 2011 Edition. The methodology, as described by the good people at Internet Retailer, includes the following:

Financial Data: Where possible, web sales statistics were provided by the companies listed. When not provided by the company, IR estimated sales based on traffic, assumed conversion rate and average value of a sale for that retailer’s category/specific industry.

Traffic and website availability/downtime: data from comScore Networks Inc. and Compete Inc. were used when specific figures were not supplied by the retailers.

Conversion Rate: If not provided by the retailer, IR used category/industry data and analyst interviews to formulate estimates.

Average value of sale: When not provided by the retailer, IR based figures on averages within the category/industry and input from market analysts.

WITH ALL ESTIMATES, RETAILERS WERE GIVEN THE OPPORTUNITY TO RESPOND TO/REFUTE FIGURES

I hope this helps!

-TD

http://twitter.com/Ryohara Ryan O’Hara

Thanks for clarifying that Tom. I appreciate it. If anyone has any questions, let us know and we can try and put you in touch with their team.

http://twitter.com/Ryohara Ryan O’Hara

Here’s the infographic for those having trouble seeing the link.
Perry

Really good tips. I’m a ecommerce developer in NYC and remember this when the holidays are coming up.

Black Friday: Proof That Ecommerce Uptime Truly Affects The Bottom Line

Lost Time

Lost Revenue

The Holidays Are Approaching

Related Posts

Enjoyed this post? Share It!