We picked up a new client yesterday, one that had previously shopped around and tested the waters with some other providers out there.  One of the claims he was given from a competitor was,

 “Our sites load within 150 miliseconds (they actually average about 1100ms). Our competitors on average clock in at over 2,000 miliseconds.”

I confess, on the first load, our site takes about 3,000 miliseconds [3 seconds] (if you’re lucky which includes all background loading), and our customers average about a 1,500 milisecond first-load-in (now some sites with hundreds of high res product images on the home page cough client knows who I’m talking about cough maintaining even a 5,000ms first-load is rough).  Sounds bad right? Well, it all depends on what we’re trying to accomplish.  Obviously we have invested a lot of web page real estate into supporting and interacting with our customers right from our website where it is generally most convenient for us and our customer.  This takes time to load (SSL for security, JavaScripts for live chat, encryption and decryption for passwords and credit cards, etc), and we offer the same services for our clients.  We don’t just have simple info-wall web page(s) like “A Competitor” offers to their clients and has for themselves.  No, our sites are interactive, media rich, and you can even live-chat with one of our support personnel right from our website! And I haven’t even started to touch shopping cart and catalog websites (which some how isn’t presented in this competitor’s portfolio list).

But here is where I drew the line for this client and explained, “Ok, now the claims are just, plain, false”:

 The speed is because of our lean tools platform coupled with our optimized server.

The speed is due to there being nothing or very little on the web pages, in fact what this competitor is now referring to is server response time.  Essentially, this competitor is trying to compare Formula 1 Race Cars (very small, skeleton web pages) to Rockets (very large, interactive, and media rich web pages). Sure, that Formula 1 will win on short-range tests even if it has a really poor engine, it has no mass and little inertial forces to overcome.  The rocket on the other hand, has a lot of mass and inertial force to overcome, but in a test where the engine of that rocket is given full range, it, by design, will win the race.

The test I ran indicates that our servers respond 0.2 seconds faster against this competitor. And when comparing this competitor to a group of 10 competitors, they came dead last in server response times, we came second, most definitely not a result indicative of claims of optimization.  Now, I will be up for the next 3 days figuring out how to tweak our servers to come in first! Seriously though, I really do thank this client for the tip!

I’m a network guy — if it has something to do with the internet or its network technology, I must learn anything and everything about it!  That’s who I am.  So it befuddles me why this competitor decided to place his servers in New York while doing business in North Carolina.  You see, all of NC traffic on major networks such as Comcast, Time Warner Cable, Verizon, AT&T, etc goes through the Atlanta, Georgia Public Peer.  What that means is any communications to New York has to go, first, south to Georgia, and then ride the Level 3 pipe to New York to that peer to exchange with the providers there (literally going around your elbow to get to your thumb).   This is why we chose to put our servers in Georgia and Florida.  Both are sitting on major public peers and in a direct linear route from NC, so packets take half the time to get back and forth to us here in NC than say New York, the Mid-West, or California.

Also, always ask your web developers one, single, simple little question:  Do you have root access to your servers?

You’ll find the most common answer is no.  Most developers don’t have an IT staff to manage a server plus a staff to design/develop their sites. This means that they can never guarantee or assure you of any standard of performance because they’re using a hosting provider that stuffs as many accounts as possible on a single, over-burdened machine.  Our company only ever allocates five clients per CPU core.  Because we also have root access to our servers, we can always customize the environments specifically to our customer’s needs.

In summary, if you’re shopping for a site, do be very careful about claims of loading speed or optimizations, etc.  In fact, run your own tests; open their client’s site and test their load speeds for yourself while evaluating for yourself whether or not you believe that site’s content and media is worth the load time.  And ask questions.  If they have nothing to hide they’ll love nothing more than to sit back and answer every question you can toss at them.