Reset password:

Strategic insights
Speed Matters - a lot!

Written by on May 8, 2009

If there is one thing I have learned over, and over, and over again, is that speed matters a lot. I have also learned (usually the hard way), that speed matters more today, than it did 10 years ago - even-though we now have broadband connections.

To put it simply:

If you are driving in a slow car, you accept that you will have a pretty unimpressive driving experience.

But if you are driving in a Ferrari, you demand that it goes fast-as-hell. Slow is not an option in a Ferrari.

It is the same on the web

If you got a slow computer, you accept that it doesn't always react instantly. If you got a slow internet connection, you accept that it might take a while for it to load the page.

But if you got a fast computer, with a fast connection, then your attitude changes. Now, you demand that everything happens instantly. Your sensitivity towards performance goes way up.

And since everyone now has a pretty fast connection, speed is a much bigger concern today, than it ever was in the past.

The Real Instant Coffee Generation

Starbucks recently came out with a new product, real instant coffee. It is nothing like the usual type of instant coffee, because this is real, fresh, coffee. Its not the pre-brewed dehydrated coffee that you usually get. It's the real thing, but as instant coffee.

Loading Video
(Note: You need Flash 9+ to watch these videos)

This is typical for the type of user behavior that we see today. We want the good stuff now. We don't want to wait. We don't want something 'fake'. It has to be the real deal, and we want it now.

We are all part of the real instant coffee generation. Waiting is not an option. A pre-loader is like looking at an old coffee machine slowly brewing coffee one drop at the time.

A Second is a really long time

I recently wrote an article about the 'a-second-or-so' problem. It is when people add small things that only take a second, and think that it doesn't matter. You know, it's just a second. People will not even notice. It is from now-to-now.

But a second is an insane amount of time, and it is the most expensive second you can ever imagine.

When Google slowed down 500ms, they lost 20% traffic. Amazon ties 100ms latency to 1% sales loss.

That is, if Amazon decides to add something to their website, that slows it down by 0.1 second, they would sell 1% less books. It equals $191,000,000 in lost revenue in 2008. That is a lot.

And it is not just about big sites like Amazon and Google. This affects every site.

I have experienced the same effect with many of the web projects that I have been working on.

A few examples:

Baekdal.com - slow means less fun

On baekdal.com I often experience a drop in page count when the site is slow. So if my servers are, for some reason, slower than normal, then people will not read as many articles. But if the site is fast, then people are much more likely to 'spend some extra time' exploring other parts of the site. The effect is much as 10% less 'exploring', just because the site is slow.

Media Management

We have also experienced the same thing with web applications.

One of the biggest web applications we have made is a Media Management System. It's a web app that makes it really efficient to see, publish and work with images.

Our backend developer have created a system that will convert images more than 4 times faster than what you can do manually. And it displays the files, in the browser, up to 5 times faster than Windows Explorer. It also uploads the files to a web server (if needed), twice as fast as what you can do with most FTP clients (like FileZilla).

So by all accounts, this is astonishingly fast.

And yet, every time we add just a tiny little thing that slows it down by a mere fraction, we get an immediate reaction.

A year ago, the app was about 2 seconds slower to use, and people started 'doing it the old ways' - even though it was still many times faster than what they could ever do manually. But it 'felt' slow. So people abandoned it.

We then redesigned a part of it, which gave us a good speed increase, and the high user satisfaction returned.

But think about it. 2 seconds makes the difference between an astoundingly fast system, and a system 'that we cannot use, because it slow'. That is a really expensive pair of seconds.

LIVE Image sizes

Another system I have made is one that automatically converts any image to whatever size you want - LIVE. So instead of manually having to resize and upload images, you just tell the system to give you this image, in that size.

It takes the system 0.3 seconds (in average) to convert the image, which is pretty fast. And yet, that wasn't fast enough.

When we put it into production we got a lot of bad reactions that the websites were too slow. It took way too long to load.

So I created a caching system that would pre-convert the images, cache them, and send that to the browser. Which means that it now takes only 0.01 second to load.

And guess what... People are now saying that we have a fast site. 0.3 seconds turns out to be a really long time.

The point is...

  • Speeds matters a lot... seriously! Every second counts. Actually, every millisecond counts.
  • Speed does not equal bandwidth. What matters is that everything must happen instantly. You can get a long way by pre-caching, or changing the order of things. Just don't make people wait for it.
  • Speed is not just about load times. It is also about how fast the interface reacts.
  • You can optimize 90% by thinking outside the box, and doing things in ways 'you are not supposed to'.
  • It's always about people, not technical stuff. If you have to make your servers 50% slower, to make people's experience 5% faster, People win!
  • It's not about overall performance. It's about getting instant feedback with what I am doing now.

...and finally...

  • Something that took 10 seconds a few years ago, has to take less than a second today. People do not want to wait.

For more technical tips on how to improve performance, take a look at: Even faster websites, by Steve Souders, Google (PowerPoint). Make Data Useful, by Greg Linden, Amazon (PowerPoint). Exceptional Website Performance, by Stoyan Stafanov, Yahoo (Slideshare)

Share on

Thomas Baekdal

Thomas Baekdal

Founder of Baekdal, author, writer, strategic consultant, and new media advocate.

Follow    

Baekdal PLUS: Premium content that helps you make the right decisions, take the right actions, and focus on what really matters.

There is always more...

The Economics of Individual Media »

ONLY FOR
SUBSCRIBERS

31
PAGES

What if Quality Journalism Isn't? »

ONLY FOR
SUBSCRIBERS

21
PAGES