The business guide to website performance

by Mike on May 19, 2013

Where did the phonebooths go?

Where did the phonebooths go?

Ever considered the speed of your website? Probably not – it’s just one of a hundred trivial details that don’t matter much.

Google disagrees.

Google’s experiments show sluggish performance affects how often their customers use their search engine. They even decided to use your site speed to calculate your search ranking.

The truth? Speed might seem best handled by the same intern that fetches your sandwiches, but it’s likely a core part of your business. Performance isn’t small work like picking coding standards and checking commit logs.

So speed matters. Now what?

How the internet works, kind of

How the internet works, kind of

1) Throw out your mental model for web speed

Most people think of a web site as a big bucket of data. If you want the bucket to move quicker, reduce the data size or increase the size of the pipe.

This mental sketch is useful, but 80% of the problem doesn’t fit the plumbing diagram. There are a boatload of complex issues, including:

  • If the web site calls for files in the wrong order, the computer sits on its little digital hands
  • The data bucket contains many smaller buckets and each takes time, regardless of size
  • Unless you’re careful, the computer has senior moments and asks again for “known” files

The bottom line? Speed fixes are different from most business and computer problems – it’s difficult to wrap your arms around the options. Web performance optimization is still a new sport – challenge your internal team to ensure they’ve dug far enough into the possibilities. The experts aren’t there just for implementation – they also can think of useful directions that are beyond us laymen.

Bugs better viewed from a little distance

Bugs are best viewed from a little distance

2) Don’t measure website speed with the naked eye

Site speed can confound. Don’t even bother surfing the site yourself. You don’t see the same website your customers do.

There are many important differences between your experience and theirs. Factors include physical location, connection speed, and cached files. Just remember, your website is different for everyone.

You can’t even rely on customer feedback. Few people are cranky enough to call in about website speed. Users don’t consciously notice when they hit problems. They might even blame their own computers or connections. You may not hear about bad performance, but you can spot the lower sales numbers.

Where's the kaboom? There was supposed to be an earth-shattering kaboom!

Where’s the kaboom? There was supposed to be an earth-shattering kaboom!

3) Get an assist with speed measurement tools

Speed tools are the computer equivalent of secret shoppers – they report the information you can’t check for yourself. There are really only two types of speed tools – synthetic testing and Real User Monitoring (RUM).

  • Synthetic uses a computer to simulate one of your users. It’s easy to set up (try a free tool now), can check from different connection speeds, and is especially useful before your site is live.
  • RUM samples load times from your actual users – who each have different connections, locations, caches, etc. Reality might be poles apart from synthetic, so this method is considered state of the art. Google Analytics (also free!) is based on RUM.

Synthetic is simpler to set up. It will quickly show what your site looks like for a user with a normal connection in a different location. But only RUM can report reality – and this might be surprisingly different.

Make sure to start measuring pronto (hopefully, with RUM). You need a good baseline to judge improvements.

Taking the wider view

Taking the wider view

4) Gain some perspective

Speed measurements show the size of the problem, but the numbers are “flat”. They don’t answer your real question – “Is my site fast enough?”.

Google’s experiment shows how speed affected Google’s customers’ search usage, but your question shouldn’t be about Google.

Your site has different products, different customers, and different choke points. You might even generate value different ways than by selling product. However, your customers will also show their speed preference. Everyone votes with their wallets, just in different ways.

One rule of thumb is to aim for two seconds. Another is to use your competitors’ websites for benchmarking. However, the true answer varies for every site and every page.

For more color, try Web Speed Ledger. This analysis shows profit improvement from speed across your top pages (and more accurate RUM numbers). Use this info to weigh the potential profits against costs, set budgets, and know where the “good enough” line is.

All systems go ...

All systems go …

5) Take action

There are a hundred different directions to take – content delivery networks, backend optimization, image compression, and so on. Check in with your developers to see what steps they feel are useful. Just remember that speed technology is still being worked out. The solutions might not be obvious to your team – and the implementation may be tricky.

In the short run, outside expertise might be the most cost-effective path forward. Reach out beyond your walls if needed. But keep your long-term need for speed in mind – you’re not looking for a one-time fix.

Gradually pull this knowledge inside your doors. Maybe your team can benefit from speed meetups or other training. Include speed profit impact in the team’s annual review.

And spread speed thinking wider. Designers should consider image size, while product managers can weigh the speed cost of new features. Once you assign a dollar value to speed, people will pay attention.

Good luck! Speed may be a goldmine of new profits for your company. And you just might make the web a little better while you work that vein.

Previous post:

Next post: