Ungerboeck Digital - Blog



Machine Learning vs. SEO

Tyler Bishop
Jan 26, 2016
machine learning and SEO

So, at this point we all know that there was a major update to the Google algorithm a few weeks ago. We also know that this change officially incorporated “Panda” into the core algorithm AND that we’re a lot less likely to hear about major updates like this in the future.

Well, hooray! Google has finally figured out exactly the right ingredients for perfect search, so we can all stop trying to stay one step ahead (or just plain keep up, if we’re being honest) and focus on what we know works, right? Not quite. Sorry to have gotten your hopes up.

The implications of the latest news out of Mountain View are not that there won’t be any further changes to the algorithm, just that they aren’t likely to be the kind of major updates that require comment or send huge shockwaves across the web. From this point on, change is likely to be continuously incremental and semi-predictable. Does that sound at all like the difference between say, a human and a robot? Exactly.

What this means is that the days of patching in updates in dramatic rollouts are a thing of the past. Updates will probably soon be more like how user updates occur in Facebook. Facebook makes changes or tweaks on a daily basis, but they are rarely noticed because every time users login they are experiencing the latest version in real-time. This makes changes seem less dramatic and prevents major hiccups or errors.

In fact, the similarities between Facebook’s application updates and Google’s search updates don’t end there. Both companies have invested a lot in the virtues of machine learning to carry their innovations forward. This kind of learning has the ability to significantly alter the way searches have been performed for nearly a decade. We may actually be in the midst of one of the most dynamic shifts in internet search methodology ever.

Rise of the Machines

Although it’s only recently that it’s become a topic of discussion among anyone but hard-core search geeks and computer nerds, machine learning is hardly a new force at Google or in SEO, in general. It’s been tooling around in the background for some time, but is now poised to take center stage.

In the last three or four years alone, machine learning played a role in helping the algorithm to figure out intent, language vs. keywords, how to identify entities of knowledge and how to connect them with language and keywords, etc. In other words, it’s all gotten way beyond meta titles, H2’s and quality links.

These days, you can watch your fortunes rise (literally!) in the rankings by paying attention to the same things the machines are using to determine what separates a good list of search results from a bad one:

  • Bounce rate
  • Short click vs. long click
  • Repeat queries
  • Page 2 results

The bottom line: machine learning means it’s less about what goes in (page speed, keywords, anchor text, etc) and more about what comes out. What this describes is a naturally smoother, adaptive process.

SEO: Not Dead, Just Different

I titled this post as a competition that pitted SEO against machine learning (because I'm an SEO guy!), but that’s not exactly accurate. ML isn’t going to kill SEO so much as change it significantly–in my opinion—for the better.

Having an algorithm that essentially determines for itself what is and isn’t important sounds like a potentially scary thing, but that’s only if you don’t know what its end-game is. For Google, that has always been abundantly clear: provide the best, highest-quality, most accurate results for any given search query. As long as you remember who’s doing the searching, you know the answer. Optimize for humans and everything’s going to be a-ok.

What Matters The Most?

First and foremost, it all starts with good content, and a quality website. There’s simply no way to get ahead in the rankings now or in the future if you have a poorly designed site and poor quality content.

Building a site starts with the architecture and bleeds down into a form of storytelling. How do you take the most fundamental elements of your site and position them up front? Do you break down this overarching theme into small components and elaborate on them? Have you built cornerstone content around these elements? These are questions you need to ask to build a good site. But, building a great site isn’t enough on its own.

Your site must be supported by great content. Content written for people that supports the messages you’ve built your site to support. Think of it like peanut butter and jelly.

Beyond that, SEO experts of the future are going to want to focus on things like:

  • Click through rate relative to ranking position

    • For example, if you’re ranked #2 for “top IT conference,” and your CTR is higher than the average CTR for results in the #2 spot on a results list, that’s a good thing.

  • Amplification rate

    • Yes, that means social shares.

  • Low bounce rate
  • Return traffic

These are just a few examples of the kind of quality and results-oriented factors that are going to become increasingly important with machines at the helm.

I do want to note that the things that have traditionally been important in SEO (links, keywords, meta data, etc.) are not going to become completely insignificant. It’s just that it will be more useful to think of them as the new baseline. You’ll always have to start there, but those things won’t be what get you ahead or keep you there anymore.

Can We Help?

You bet we can! Producing compelling content, tracking important metrics, answering tough questions. The team at UD can do all of that and more. Contact us today to get started.

Let's Talk

Looking to know more? Drop us a note through the form below.

636.300.5606  |  100 Ungerboeck Park, O'Fallon, Missouri, 63368