At the end of May I took a 10 day vacation to Miami, it was amazing. Before that, things had gone great with our sites, communities around the German card games Skat and Doppelkopf:

We had been climbing up the rankings steadily, being featured on page one for most of our relevant terms, even finally ranking #1 for one of them. I felt like I really deserved the vacation. Google didn't (or wanted to present a new challenge after):

For some reason, our site had been dropping for some of our main terms and continued to do so, whereas rankings for other terms, containing similar words, remained entirely stable.

Let's use the game chess and English words to make the example a bit clearer (and google.com instead of google.de). Our main keywords would be chess, chess game, online chess, chess online, chess online game. 

The general term chess starts going down (during my vacation) from a pretty stable #6 over #12 all the way down to #20. The 2 word terms "chess game" and "online chess" do something similar, just not so dramatically.  

At first I expected one of those penalties I have been reading about in this great post. But as "Play Chess online" remained totally stable and I could not think plausible reason why we'd deserve one, I ruled this out.

Based on Rand's advice I then checked on our backlinks, particularly on the ones that have "chess" as the link or alt text. I realized we didn't have very many, but the ones we did have had been there for quite a while and were still up. The pages and their own backlink structure didn't really show anything. Basically there was nothing that could make us drop so dramatically. 

After that I took a look at the sites that now ranked in front of us: I had not been monitoring them extremely closely but enough to see that they hadn't made any great effort in linkbuilding and/or dramatic changes to the sites. I noticed two new international sites that had chess in their domain, e.g. chess.dk or chess.ch . As I usually use the German google.de, sites from beighboring countries like Denmark or Switzerland could actually be relevant to searchers.

Theory #1: National Google sites are going for more international results. Makes sense, I guess, as there is always the option to only show sites in the respective language or from the country for the user.

This accounted for two positions, but not the drop all the way from the middle of the first to the end of the second results page. Hence, I went on analyzing the "new" pages' backlink structure. Generally speaking, it was pretty similar to ours, although the Mozbar and Google PR saw our site a bit stronger.  I dug a bit deeper and checked out the link text, since juice and trust hadn't explained the situation either. Most pages have the majority linking to them with chessdomain.com or www.chessname.com, and some other usual stuff like "click here" or "link". Nothing special there. When I looked at the optimized linktext, I finally found something:

Although the others had fewer links, they had more with just the word "Chess" than we did. Ours were mostly two or three word combinations like "chess online". It sounds like it would make sense to rank those sites higher that have more links with the exact keyword pointing to them. The interesting thing is, nothing had changed with the links or link structure- just with google.

Theory #2: With its update, Google changed the effect of 2- or 3-word terms. In my understanding, it narrowed the positive effect to rankings for the single words within the term. Basically, I believe they just treat the term like one word now.

I believe google has reduced or cancelled the juice that would boost the individual words within the 3 word term. The positive effect of the link for the words "chess", "online", and "game" were reduced. (I attributed them to receive originally 6%/3%/1% of total link juice, but these numbers are just random guesses to illustrate my point). I had read before, that Google wanted to deliver better results for 3-word searches, e.g. here. This would certainly rather support my theory.

Okay, you might think now that there are hundreds of other influencing factors, so really one can't be sure whether this is at all true. I wouldn't blame you.

But I have one more theory: In the image above, "chess online" and "online chess" rank somewhat similar before the change (4 vs. 6 or so). When I checked our backlinks, I found that -to my knowledge- we have no link using the term "online chess", but quite a few sites linking to us using "chess online". As you see, the term "online chess" dropped pretty heavily the last couple of weeks, whereas "chess online" remained stable.

Theory 3: Google puts greater weight on the order of the words in a multi-word-link, than they used to. Sort of like treating the term like one word.

Alright, I'll stop now, this post has gotten way longer than I had originally intended. What do you think about #1, #2, and #3? Did something similar happen to any of you out there? Does anyone have a good idea how to test this, considering we can't turn back time?

Thanks for reading, I would love to hear your thoughts!

Follow-up: On July 1st, about 6 weeks after "Chess" had started to drop it went back up from #14 to #6. Exactly one week before that it had gone back up from #18 to 14#. Also the other terms recovered and went back to their old position:

You may wonder now what we have done to "fix" this and get back into the top 10 results. Absolutely nothing!

Everything is the way it used to be, the same sites rank higher/lower than we did as before. So I am positive, it really didn't have anything to do just with our site.

I am sorry, if this is kind of disappointing in a way, but I figured I had to share it so everyone gets the full picture. My theory is still that google had changed how they treated 3-word-searches but for some reason went back to the way it was before or reduced the intensity of the change.