The Google Dance Explained

Dance with me or

…of Hummingbirds, Pandas and Penguins: Oh My!

Algorithm Overload

One of the Semantic Mastery partners, Hernan Vazquez, posted in our G+ MasterMIND group a couple of weeks ago. The post included a link to a DiggityMarketing article which talked about three reasons why your rankings could be showing negative movement.

I am not sure if all of our MM members got around to reading the post or if they realized the implications of the post. So I wanted to do a fairly thorough follow up for our membership and followers.

The Dance

First of all, we’ve already come to expect movement in the SERPs after making changes on a property. It also happens during or after a link building campaign. We call it the Google dance. However, the article yesterday put a name and face to the Google dance:

Google Dance Patent from USPTO

Yes, I’m a Geek

I must admit that I am a Google patent geek. I have spent hundreds (maybe thousands of hours) researching G patents and trying to extract as much information from them as possible in order to try to understand why.

I feel like a reporter sometimes because I am constantly asking “how” and especially “why”? I already know who, what, where and when. It’s the last two that consume most of my time.

But the title of this patent and its Summary say everything you need to need to know. The patent is titled “Changing a rank of a document by applying a rank transition function.” I suggest going to the patent and reading the Summary and Detailed Description so you can see the scope of what Google is doing for yourselves.

Don’t gloss over this section!

Of the utmost importance to us is what I will call the 20/70 rule, which you will find in FIG. 7 within the Detailed Description of the patent.

It states:

FIG. 7 is an exemplary time-response graph illustrating the influence of changes in link-based information on the rank of a document according to this other implementation consistent with the principles of the invention. As shown in FIG. 7, the rank of a document may initially decrease in response to a positive change in its link-based information. After a period of time, the document’s rank might rise to its new steady state (target) value. Like FIG. 6, the time line shown in FIG. 7 may be represented in days in one implementation consistent with the principles of the invention. In other words, the document’s rank may decrease for a period of approximately 20 days before settling in on its new steady state (target) value (e.g., 1.0 in FIG. 7) in approximately 70 days after a positive change in its link-based information.

While two exemplary rank transition functions have been described above, implementations consistent with the principles of the invention are not limited to these transition functions. In other implementations, transition functions based on time delays, pre-computed piecewise time-series, or a process that examines time after a change and indicates no effect, positive effect, or negative effect may alternatively or additionally be used.

Let’s break it down:

1. An exemplary time-response graph illustrating the influence of changes in link-based information on the rank of a document.

Yes, it’s all about links and will remain this way for the foreseeable future – even the latest Ranking Score patent and the Trust Rank patents and algos, which take trust and topical trust metrics into account, are all about links.

2. As shown in FIG. 7, the rank of a document may initially decrease in response to a positive change in its link-based information.

Once gain: LINKS! Now let me expound on this because this is where it gets fun. You set a regular schedule for whatever it is you’re going to do to get your website ranking for in your niche. You’re moving along, the property is ranking well, then it suddenly drops instead of moving up in the SERPs.

What do SEO’s generally do?

They add content, edit content, move content, hit it the property with more links, then go and do some more on page, off page, change titles descriptions and anything else they can do to recover lost rankings.

Meanwhile…

3. After a period of time, the document’s rank might rise to its new steady state (target) value.

Make sure you pay attention to #3 because you’re being told right in the patent that the “document might rise to its new steady state (target) value.”

At Semantic Mastery, we’ve often discussed the “Google Dance.” We have said time after time that you simply need to let things “stew” for a while. We’ve been speaking from experience and testing.

You have to remember that we are in the trenches on a day-to-day basis. We don’t just speak about clients, web properties, rankings, SEO and Online Marketing. We live it.

Although I have spent an inordinate amount of time researching Google patents, I had not run across this particular patent.

In fact, the one that I had been looking into was an earlier patent (http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-adv.htm&r=1&f=G&l=50&d=PALL&S1=08244722&OS=PN/08244722&RS=PN/08244722) titled “Ranking documents”. Although this patent also has the “20/70” rule, it completely escaped my attention. If you look at the patent, you can see that information about the Google Dance has been available since August 14, 2012.

But look at the actual filing date! Google has had a way of identifying ranking manipulation by SEO’s since June 30, 2005. But those were the glory days of SEO and we could get away with murder back then. But Big G was already onto the game and cooking up a very sneaky way to build a better mouse trap and catch us in the act.

4. [T]he document’s rank may decrease for a period of approximately 20 days before settling in on its new steady state.

Simply stated, SEO’s would hit a web property with links, on page, off page or “any means necessary” and G would rank it. Then we’d see the Google Dance, meaning that Big G would drop what you had ranked from the SERPs for no reason at all it seemed to us at Semantic Mastery. It was just G randomness in action is the way we looked at it.

Buy we knew well enough to continue just as before. If we already had a steady schedule we were following to rank a property, we knew well enough not to veer from it. We would also rank a property and simply leave it alone.

There were times when we would see our properties settle on the second page and not move. But we knew that all we had to do was continue with our schedule.

Nothing coincidental

As you can see, there anything random or coincidental at work. The dance or drop in rankings was done purposely to check and see if there was any activity unusual linking activity during the period of the drop – 20 days according to the patent. From start (ranking the document) to finish (end of the dance), the process can take up to 70 days – longer if you hit the property while it’s dancing.

Now we have a definitive duration for the Google Dance!

Implications

You go in and rank a property (Profile, Free blog, Self-hosted blog, YouTube video, what-have-you) and it hits high and maybe even page 1 bottom 5. You don’t like that position of course since you know top 3 is where the most action happens. You set a schedule for your silos to be built out, niche keyword targets, link building, etc. But it doesn’t budge or it “flatlines”, suddenly can’t be found, and you go “WTF did I just do?” mode.

What’s should you do? 

Main thing is: DON’T PANIC! You do nothing except what you have been doing all along. You now know you’re in G’s cross hairs. This doesn’t mean that your entire property is going to be penalized if you veer from your script. It does mean, however, that your rankings will get sandboxed and you are going to have a very difficult time getting your rankings back.

This is what we have been saying at Semantic Mastery from day 1 without really knowing this patent existed.

This Humpday Video talks about the Google Dance in detail at around the 16 minute mark:

We knew from testing and experience that a property would start to dance and then come back to the original or better position. We also knew from testing and experience that this process could take one month to six weeks. And we knew that it could even take longer, but we also knew enough to just let things be.

SEO’s are an impatient panicky lot:

If rankings move just one spot down, there’s a scramble to correct it.

If G says there’s an update, then the world is coming to an end and SEO’s start scrambling and looking for answers.

If a property starts dancing, it must be stopped and corrected at all costs.

If something isn’t ranking, throw the book at it.

If it still isn’t ranking, more is better.

And it’s all wrong, according to the patent.

Again, if you’re an SEO, what should you do when your property starts dancing? Do nothing other than what’s already on your schedule.

For those of you who know me, I have been at this game for well-over a decade. The only people I know who have survived and thrived in this game for at least a decade and much longer are those who have learned to wait.

If you can’t wait things out, you get found out, your rankings drop, your clients leave and you’re out of the SEO game. I know it’s difficult to wait when you have clients calling you at all hours and telling you they want rankings NOW! But you have to learn to the waiting game or your client will have NO rankings and you will have NO work!

So wait; everything will begin to settle and clear. Now you analyze, ask how and why, and test – but NOT on anything you care about losing. If you’re testing on client sites or your own money sites, you’re creating a recipe for disaster.

Now test again and analyze. Meanwhile, those who have panicked and continued throwing everything at their properties in the hopes of recovering their rankings will never be seen or heard from again. This leaves you a better playing field with less competition and much less work to do deliver results or get the rankings you want.