The first Panda update alone changed over 12% of search results. If you’re web traffic has taken a dip in its aftermath, you may have a CEO putting your neck on the block, customers getting frustrated at not finding you or a wife threatening to nullify your pre-nup. You need to get around it, and fast. How can you do this while keeping everyone at bay?
You know the old saying “if you can’t beat ’em, copy ’em”? Think like the proverbial scamp and copy what Google did (or as best as you can quickly and cheaply). Panda’s machine-learning algorithms are based on the results of a questionnaire fed to a test panel of mostly students. Fortunately, Google have released a lot of the questions. Which means that you can cut out the Mountain View espionage missions and focus on replicating the tests yourself in 3 easy steps (thanks to SEOMoz and Webmaster Tools blog for the info).
1. Find Your Guinea Pigs
The first step is to find people that can do the questionnaire for you. It shouldn’t be you (you’re biased, trust me) and it shouldn’t be subject matter experts. Just your average, run-of-the-mill users. You don’t need to ask thousands of people (unless you’re trying to build your own search engine). Just 15 or 20 will be plenty for you to get an idea of what you need to do to get your conversion optimisation cranking. If you can’t find any research subjects or want to move fast, you can find a market of low-cost, anonymous test subjects at Mechanical Turk (https://www.mturk.com/mturk/welcome).
2. Ask Them The Questions
The questions asked are simple Yes/No questions. They are designed to be asked at 2 levels – the domain level and the page level.
Domain Level: These are general questions about your website. The answers can be asked after allowing the users to browse around the site. I recommend just showing them the homepage so that you can get that quick “gut instinct” answer.
- Would you give this website your credit card details?
- Would you trust [insert website topic here] info from this site?
- Would you recognise this as an authority?
- Is the site cluttered?
- Would you share this site on a social site or email it around?
- Was the site updated recently?
- Is there a phone number available?
- Are there testimonials or case studies?
Page Level Questions: Find a decent looking page on your site – neither a broken page nor a prime, first-born-child-tradeworthy page. Then ask your test subjects the following questions about it. But remember, you shouldn’t expect to get a “yes” to all of the questions as it’s not always appropriate. For example, a product category page wouldn’t be a candidate for a print article.
- Would you trust the information on the page?
- Are there clear and descriptive short headlines?
- Does it seem like it was written by an expert?
- Are the articles copmprehensive and engaging?
- Does it have any obvious spelling, grammar or punctuation errors?
- Does the information seem factually correct?
- Does the content sound natural or is it tailored to search engines?
- Is the content biased towards a particular point of view?
- Does the article provide original content or information?
- Does the article contain insightful analysis?
- Would you bookmark this page?
- Are there excessive ads on the site?
- Would you expect to see this in print?
- Do you think people would complain when they looked at this page?
- Does it reference an author?
3. Action The Findings!
OK, so now you have an idea of where the weaknesses are in your site and its content. The next step is to ACTION IT! Get your web designers, developers and plebs updating the information before running another round of tests to see how well you improve. The additional benefit of this is that this will help your conversion rate. So don’t be a sucker – measure your current performance, make the changes and then measure them again to see the ROI!