Market Brew

December 13, 2014

2014 Recap: Getting Really Close to Automated SEO

2014 Recap: Getting Really Close to Automated SEO

Market Brew’s CTO announces breakthrough automated SEO technology just recently added to the Market Brew platform.

What an Amazing 2014 4th Qtr

A few months ago, many of you know that we were touting that we had filed a patent on the idea of fully automating search engine optimization. As we have done in the past, we began in earnest to build these ideas into our already enterprise-grade platform, and I can report that we are further along than we thought we would be.

Now, with SEO, it’s not as easy as installing a plugin to your website, like Optimizely does for automating conversions. But…we are pretty damn close to automating the entire process. Let me take you through just some of the major enhancements to the Market Brew platform in these last few months.

A Customizable Search Engine Model

The idea of not only knowing the search ranking distances between results, but understanding what MAY be about to happen because of this, is an incredibly sexy proposition. It begins with our search engine model, a 7-year regression-tested model, and combines your own metrics and customizations directly into the calculations.

We ditched the idea of trying to force our search engine model onto everyone, and instead now let each digital marketer customize their very own version! If you think the model leans too heavily on links, you can now dial that back. Think the META Title doesn’t have much weight anymore? Well, you get the point. We also let you import your own revenue-metric numbers for both keywords and webpages. Everything we need for the next step…

Simulations, Simulations, and more Simulations

Once we have your data and model customizations, we use our ability to uncover search ranking distances and run thousands of simulations to determine precisely WHERE you should start optimizing. For instance, if you are ranking #4 and moving to #3 is close by, but moving to #2 is not, then our system will weight this predicted move accordingly.

A Literal TO-DO List

Now we can easily sort by this ROI-based optimization score to reveal exactly where you need to start optimizing. This is great, but we’re only on the 30 yard line — we’ve got the hardest part left!

I’ll admit, the concept of knowing EXACTLY WHERE you should start optimizing to move the ROI needle (not just the traffic needle), is pretty time-saving. You don’t have to run these simulations on our model by hand anymore, which saves hours of research. If only we could now determine WHAT to optimize…

Simulation Overview

With our new Simulations Overview screen, you can immediately see how close you are to moving up (or down!) in search rankings. The hotter the color, the closer you are. Knowing this information is great, but imagine having to do this for every keyword, and then deciding which opportunity to start with first? This is where the simulations really save you time.

Here is where our system and our competitor’s systems diverge greatly (as if they haven’t at this point already). What you will find on the mainstream platforms is that two disparate systems, like a ranking overview and a “website grader” tool, are combined together to give you actionable advice. This is very troublesome though. Sure, you rank #4 for “innovative software”, and you need to add more references to “software”, but seriously, what does this do to the overall traffic? Does it move you into the #3 spot? And HOW do you optimize?

Simulation Breakdown

Our system ties these two concepts together (ranking and actionable items), so they are the same system. By slowly breaking down the rankings into smaller questions (similar to an equation), we can reveal a very accurate course of action. Our search engine model will reveal EXACTLY how much semantic vs. link flow optimization needs to be done.

Because we took a bottom-up approach, and built our own search engine model, we can effectively reveal the query parser that is responsible for matching documents and returning them in order of “most relevant”. And you can see here that this particular simulation is saying that we need to focus on semantic (content), instead of link flow, in order to increase rank. We decided to show you a percentage breakdown, because it’s not black and white — sometimes it’s a little bit of both.

Semantic Breakdown

Why stop there? We can break this semantic scoring layer down into further sub-scores, which allows us to determine how the semantic score can be moved quickly. In this particular simulation, I’m wasting my time if I am working on the META Title.

I won’t post screenshots here, but you can click into each one of these areas, and the platform will show you exactly how to apply an optimization to your website. For instance, to optimize the Market Focus, we have a Market Focus Details screen that details out exactly how it’s calculated and how it can be optimized.

Link Flow Breakdown

Don’t forget the link flow breakdown. We haven’t figured out a neat graph yet, but here is the explanation shown on this screen for this particular simulation.

Note that the system tells you exactly how much link flow score can be increased — and where to increase it. In this example, there are two possible optimization options:

  1. Algorithmic Penalties
  2. Link Flow Distribution

We haven’t automated the explanations yet, but you can simply click into the algorithmic scorecard to view which penalties are hitting this webpage the most:

And if algorthmic penalties is not enough, you can easily move some link flow around inside your website and have dramatic results. Enter the new changes to the Link Flow Distribution screen:

Notice how the system shows you your current distribution of link flow, as well as the TARGET distribution. The target distribution is based on the keywords you’ve setup in your Analysis Group. It basically shows you where your link flow should be flowing. We take that one step further with the introduction of an automated matching system for your content — which takes your TARGET keyword and tries to find the closest match (think Levenshtein Distance), effectively automating the task of figuring out HOW to redistribute your link flow!

Of course, if your #1 target is the keyword “lawyer” and your home page is about “paperstreet web design”, the platform is going to show you this, so you can correct and re-calculate.

Summary

To sum up 2014, we’ve had a pretty crazy year. It started with a pretty insanely big idea, and ended with the beginning of the impossible. See you on the flip side in 2015.