SEO Testing – What Works and What Does not

How many times have you worked on a client’s site and thought to yourself, “I hope these changes work!” For many people in the SEO field testing is not a part of their toolkit. They make changes based on what has worked in the past and hoped it would work now. Only time would tell if the changes actually helped or hurt the site – and by the time you knew the results, so did the client!

Many SEO companies are a “one solution” company. They offer the client one solution to a problem, and don’t even mention the fact that there are several others that may produce better results. If you are lucky, in 6 to 12 months you’d have a chance to revisit the recommendations – but let’s be honest, how many times do we actually do that? Once we make a recommendation most of us stick with it and never look back.

To be fair, it wasn’t until recently that we had tools to help us make better decisions and to test our changes. The introduction of detailed web analytics (such as Google Analytics) has made the idea of SEO testing achievable. Search engines and crawlers also now index more frequently, and thanks to features such as XML sitemaps we can even control how often most of our content gets checked and crawled. We no longer have to wait 3 months until Google crawls us again, we can (almost) control it ourselves. All of this comes together to let us see what changes worked, and which ones didn’t, faster than ever.

So now that we know we have the tools available for testing, what type of testing should we do? What questions do we have to ask ourselves to make effective use of SEO? Some of the better questions I’ve come up with are these:

• Does alt-text on images matter? Should I stuff keywords in there?
• Are footer links effective? Should I have one, two, two hundred?
• How many links should I have from the homepage? Should I just link to main topics or products or should I try to link to a wide assortment of pages?
• Long or short title tags? Brand before or after the keywords?

These are just a few questions (and there are probably hundreds of other questions), but the answers to these questions are going to vary from site to site and demonstrate how testing can help point you to the correct answer.

There are also questions which won’t be as easy to determine, but still need to be addressed: Should I buy links? Should I cloak content based on user agent? Questions like these are difficult to test, but you should still be asking and trying to answer these questions as well.

Once you get your testing methodology down for SEO you now need to start monitoring the results. One of the best ways of doing this is through split testing. This is where you use different techniques on different parts of the site and see which one brings about the best results. For some sites you may also wish to look into A/B testing where you design the same page two (or more) different ways and then send a percentage of people who visit a particular page to the different versions. This works best when you are selling something as the results are more obvious through the sales figures.

Testing should be an integral part of any SEO process. If you aren’t doing it now you should start investigating how to do it and start rolling it into your projects. You will find that by investing the time up front to setup testing methodologies your final deliverables to your clients will deliver the results that keep them coming back year after year for your services.