Google ranking changes, It's not always your actions

[goodroi]

Many people worry about their website rankings. Personally I'm more interested in traffic vs rankings but I'm getting off topic. So I'm often contacted by people that are worried because their website rankings have changed. Lately it seems the flux in Google rankings has increased. Here are the results from an oversimplified test to show the flux in Google's search rankings.
Step 1 - Select a unique text string that was not published on any website and returned blank search results.
Step 2 - Place it on some sites so now only these sites filling up the entire search results
Step 3 - Make no changes to the sites and see what happens to the rankings.
Step 4 - Share on WebmasterWorld to demonstrate the concept that rankings can change due to Google and not because of our positive or negative changes even if my example is a bit oversimplified & imperfect. Notice how the different sites which weren't changed moved in the rankings for over two weeks.
Apologies to color blind people who might not be able to easily follow the different colored sites

[aristotle]

This shouldn't be surprising, since google reportedly makes small adjustments to its algorithm almost everyday. But there are also other reasons why a site's rankings could change even if the site itself remains unchanged. For example:

-- External factors such as the site gaining or losing backlinks

-- Changes made to competing websites

It's hard to devise a simple test that completely excludes off-site factors.

[NickMNS]

Here are few issues I see with what you have presented:

Problem -1

Step 1 - Select a unique text string that was not published on any website and returned blank search results.

Google uses semantics and context to extract meaning of web page and then match meaning to intent to return the most relevant pages to the user. Taking some random "unique" text will not allow Google to this matching in a reliable way. Thus the result of the test will not be representative of actual searches, instead it will be testing behavior of a corner case.

Problem -2

Step 2 - Place it on some sites so now only these sites filling up the entire search results

Are these new sites or established sites. What else is on these sites. How and where are they hosted? Is Google seeing these as unique entities or copies? If they are established sites, then you have the influence of the site's history and its own ranking factors.

Problem -3
What is a ranking? Is it for mobile device, desktop devices. What is the search volume for the search terms. Are the tester's searches the only one's?

Problem -4
Google does not need to change anything and you will certainly still have changes in "ranking" due to other exogenous factors. Like the impact of other sites making changes, or semantic changes like say an artist releasing a hit song that includes some part of your text in the lyrics.

What exactly are you testing, what is the hypothesis? Was it accepted or rejected?

[goodroi]

Something else was being tested and I can't share that information. I thought it would be helpful to show how rankings can change without any intentional changes to off page or on page factors. As I already said this is an imperfect test because this manufactured situation does not happen in real life. Honestly every test has some level of imperfection. If for nothing else it is a good reminder that Google makes over 500 updates every year and there are only 365 days.

To answer a few more questions for any curious people.
-No change to backlinks of any site. All sites were owned & being controlled.
-No changes to competing websites because this manufactured situation (aka unique text string) prevented any other sites from ranking. This involved about 20 test sites.
-Search volume is non-existent for this unique search term.
-This pretty chart is for desktop results, mobile results had similar levels of flux.
-The public was not involved in searching or clicking results. Technically it is possible but extremely unlikely a member of the public randomly guessed this unique text string.

[seoskunk]

Conclusions? Google auto rotates.....

[goodroi]

There are many possibilities and it could be a combination of factors. Here are some specific ideas in no particular order...

1) Google auto rotates your rankings to see if it changes the CTR and/or bounceback rate. This could help them identify which sites are better satisfying users.

2) Google averages more than 1 algo change a day so every day your ranking might change if your site has a very close ranking score to other sites.

3) Google AI was trying to understand the semantics or other element of this manufactured situations and kept reshuffling.

4) Google monitors the user activity and while conducting the test there was not a steady and identical amount of time on site and/or ctr for all sites.

5) Google is very sensitive to page speed and was shuffling sites depending on the changing server speeds.
(I doubt its #5 but I mention it along with the others to demonstrate the wide range of possible theories and opinions to explain this)

[Lexur]

@goodroi I think your test is a great experiment to check the premise of the post's title.

I would add a 6) posibility, including the mindset of big corp employees there:

6) engineers change the results because, although they have no an exact idea about what they are doing, it seems to be smart and gives no clues to executives, competitors and cheaters, so they can add one more bonus to their astronomic salaries.

[goodroi]

@Lexur It seems you have let your personal frustrations cloud your professional comments. That is not a wise or profitable option. You imply that Google PhDs have no idea what they are doing. That is highly illogical to say the least. Yes, Google makes mistakes but they are not clueless morons. They have short term & long term strategies that they are trying to achieve.

In life it is wiser to respect our adversaries and not let personal frustrations cloud our professional judgement. Let's be careful to stay on topic and avoid focusing on our personal frustrations.

[seoskunk]

I don't think Google care that much about organic results, its just randomised backfill. They let there algo's decide the order of results and then auto shuffle.

AI , continuous changes, wouldn't be surprised at all if its BS

"If you believed they put a man on the moon..."