I recently worked on a site that was previously well known but had fallen out of the rankings against the competition. What became really obvious was that content that had previously been included with the domain hadn’t been properly moved to a new host so there were lots of 404s and all of the traffic generated by established content evaporated. I grabbed old content from archive.org and began recreating it so we could at least get back some of the old authority. Some weeks passed and the site’s pages once again started moving back up the chain. But then I found a really odd thing happened.
The site needs to have authority across a focused set of keywords that all deal with the site’s industry. Working with what is considered best practices, I added optimized titles, h1s and meta descriptions for two phrase match keywords and made sure that those phrases had some weight in the body content. What happened next surprised me. After just two days, rankings for the wide variety of phrases I was running ranking reports on plummeted or disappeared altogether. Rankings for the specifically targeted phrases, however, improved slightly. The breadth of keywords that the site’s pages had ranked for fell off a cliff. Could it be that some specific part of the optimization was more to blame than the others? Time for an experiment.
I started by removing the custom meta description from every page in the site. The site was still thin on content and traffic was very weak anyway.
It turns out that this was exactly the problem. I am thinking that before I messed with the meta description, Google had been analyzing the entire content of the page better. As if the meta description told the algorithm, “Hey, just check the content to make sure it lines up with the meta description.” Without the md, Google looked at all of the content and said, “It seems that because of the content, this page should rank for x, y, z and these others.” Within 12 hours of deleting the meta description many of the phrases rebounded by as much as 40 positions.
If this holds true, using the meta description might be more useful for very specific pages that you don’t want extraneous traffic directed to like a specific product. However, for pages on sites where you are looking for maximum exposure, letting Google decide might be a better route. The other question is, will the pages without a meta description every rank very well for any specific phrase or will they just have broader appeal? That is an experiment for later and I can do some A/B to see what holds visitors better as well.