Reverse Engineering Search Engine Algorithms is Getting Harder

magnifytweezers-fleaIsolating ranking factors to see how changes affect ranking has always been the basis of algorithm testing.

After updates like Panda and Penguin, reverse engineering and even intentionally triggering low quality indicators has become extremely hard and unpredictable.

Google algorithms behind Panda and Penguin lack a single detectable threshold and none of the factors are used in isolation. Understanding what’s changing is key to future search engine optimization (SEO) endeavors.

What’s Changing?

Search engines are constantly evolving, but recent updates are radically changing the way in which Google handles unnatural behavior. Instead of manually defining the thresholds and conditions in which something is good or bad, they can now use scripts to define thousands of combinations that best describe spammy tactics.

Only in combination with various other signals is something good or bad and what is good for one site may be bad for another. Singling out ranking factors can no longer be done. Only ranking factors that occur in a large amount of patterns will be detected by collaborating SEO specialists.

In extensive pattern combinations it is even more reliable for Google to use time in the equation. “When has a certain change been made?” can now be used for individual signals. This makes the same activity turn out positive or negative just by accidental timing.

The Machines are Taking Over

Machine learning has always been something Google was hesitant to implement on a large scale. Current developments show that this is rapidly changing and the success of Panda and Penguin brings new found trust to expand machine learning capabilities to all Google algorithms. Search engines like Yandex show the potential of machine learning on search results and Google is very likely to push it to the next level when they get a feel for it.

How Does This Affect Daily SEO?

This development has a huge impact on SEO as we knew it. There is no longer an “ideal situation” you should carelessly aim for!

A quality website will become more of a mindset and less a collection of correct signals. It’s extremely hard for me to admit that it is becoming easier to create a great website and to do great marketing, than it is to fake it.

Until now it was easy to diagnose a website on individual symptoms to get a specific cure for each of them. From now on the advice will become more generic and the best cure will incorporate all aspects.

Like with an obese patient it is hard to determine what will kill him first. You know he is living an unhealthy live and you can focus on his heart, diabetes or kidney problems, but the patient is best of by bettering his live and losing weight. Because this is almost too logical and there are no shortcuts, a future proof SEO should be more like a dietitian, fitness instructor and health coach. Being there at every tedious step of the way.

The Near Future

More focus on the possibilities of machine learning will probably improve Google results enormously. At last Google can start to use sentiment analysis on search results. 

Google will be able to accurately mix universal results for my individual desires. Further personalization based on my friend lists is even impossible without elaborate machine learning capabilities.

Machine learning is here to stay and it will change SEO in unimaginable ways. Be ready for when it does, so start your first machine learning class today in Stanford machine learning 101.