Think Wikipedia dominates the Google rankings? Think again. Amazon is king, but watch out for Youtube.
In the last couple weeks, Search Engine Land and Search Engine Watch have been trading blows with duelling studies over how much Wikipedia dominates Google search results. First, there was a study by Intelligent Positioning (and SEWatch) that claimed that 99% of all Google.co.uk searches have Wikipedia in the top 10. Then, there was a counter-study by Conductor (and SELand) claiming that just 46% of all Google US searches have Wikipedia in the results. Then there was another study published by Search Engine Watch claiming 84% of results have wiki on page one. Finally, in the comments of the Conductor study, someone pointed to a Slingshot SEO study that claims only 28% of searches point to Wikipedia.
So, there’s a bit of a controversy. Each of those studies is based on a sample of between 1000 and 3000 searches. As it happens, I have access to the SpyFu SEO Recon database which lets me do exactly the same research instantly against a sample size of 699 Million SERPs, a sample size at least 250,000 times larger.
I can also produce the same results spanning over 5 years for both Google.com and Google.co.uk. So, I can see how Wikipedia has moved in the SERPS over time. (BTW, we have a screenshot of every Google SERP we’ve ever pulled.)
That’s when I came up with the idea to see how Wikipedia fares versus other big players like Amazon, Youtube, and Ebay. Those are crazy interesting results.
There’s one problem: There are a lot of ways to have a biased result (one-word keywords, highly informational vs transactional keywords, location, etc). So, each study is pointing out the biases of the other study. Let me do everyone a favor and point out the biases in my data:
- 1. SpyFu’s keyword selections are more likely to be keywords that people advertise on; they are much more likely to transactional.
- 2. SpyFu’s keywords are unlikely to be very low search volume. We choose the highest volume, highest competition keywords first.
- 3. Our customers can add keywords they’d like us to track. Since they are more likely to add keywords that are related to a product or service it biases our keyword pool further toward transactional keywords.
One quick thing that I did to partially mitigate the transactional keyword bias is I ran all the single-word phrases (over 589 thousand of them). That’s a direct one-to-one comparison against the original Intelligent Positioning study. But, of course, I’m just trading one bias for another. That said, it gives us more than one point for triangulation. Really, I *can* eliminate biases pretty elegantly, but it would take a couple hours and in the interest of getting something out I’m foregoing it. If the overwhelming response in comments is that I need to eliminate the transactional bias more, then I can.
Okay. Now on to more results:
First things, first. This is a direct one-for-one comparison against the original study posted by Intelligent Positioning and reported by Search Engine Watch. Just like them, we used only single word search terms and ran them against Google.co.uk. The difference is, instead of running a snapshot of 1000 searches, we ran ~589,000 single-word searches every month for 4 years. My results show about 69% rather than 99% of Google SERPs returning a wikipedia result on page one, and 20% rather than 56% with wikipedia as the first result. It's notable, though that when I run the exact same single-word report on the exact same 589k keywords, but for Google.com instead of Google.co.uk, only 60% of pages (9% less) have wikipedia on the first page. So, the UK peeps are indeed seeing more of the wiki.
Next we have a biased, but large scale validation of the Conductor Report. The reason this report is biased is because SpyFu keywords are somewhat biased toward transactional keywords -- we cover pretty much all the keywords that anyone would buy on Adwords, and since customers can add keywords to our database, those tend to be their best converting keywords. Conductor reports that 46% of searches have a Wikipedia result on page one; but, they also say that 34% of transactional searches have a Wiki result; my results show 32%. My data is based on 348 Million SERPs; theirs is based on 2,000 SERPs. Acknowledging my data's bias, if I were a betting man, I'd bet that their estimate is about 20% high... and what I mean by that, is the real number for transactional searches is 28%, the real number for informational searches is 48%, and the real average number is 38% (vs their estimates of 34%, 60%, and 46%, respectively). Pretty close, though; my numbers more or less validate their numbers.
Oh, I almost forgot about the Slingshot SEO study. Theirs is pretty solid too. Since theirs is based on "keywords serviced for real clients, which provide a sample representative of more competitive, valuable, and realistic SERPs" , their study is "biased" in a similar way that my data is biased. As such, their results are very similar to mine: 28% vs. 32%. So, mas validacion.
On to my favorite part.
I decided to see how Wikipedia fared against other big players, Amazon, Ebay, Yahoo, and Youtube. What you can see in the chart above is that we should spend less time worrying about how many SERPs Wikipedia dominates, and worry more about Amazon; Amazon shows up on on page one more often than Wikipedia, and has for over a year. More importantly, it's still moving up, while the Wiki is flat and all the others are either flat or in decline.
Okay, let's see how it looks for the same domains, but for the Top 50 results.
Again, Amazon is a powerhouse that either beats or ties Wikipedia depending on the month. There are a couple more stories here:
Yahoo has been on a consistent decline -- personally I'm surprised at the footprint that Yahoo still has (although, I guess people still use AOL, right?). Yahoo has lost about 50% of it's page one and overall (Top 50) Google results in the past 3 years.
In the past year, Google has increased Youtube's share of top 50 results by 4x. Check it out:
So, yeah. Keep an eye on Youtube.
Okay. To sum up:
- Intelligent Positioning - sorry guys, I tried, but I couldn't confirm your results even with a side by side comparison. But, Wikipedia shows up in *at least* 10% more page one results on Google.co.uk vs Google.com
- Conductor - Nice work. I can confirm your work. But, I think your numbers are 20% high.
- Slingshot SEO - Also, good work. I can confirm your results, too.
- Amazon is all over the SERPs; Wikipedia is not the SERP king.
- Yahoo is dying slowly.
- Youtube is trending.
Let me know in the comments if there's something else you'd like to see. If you post quickly, I'll run a query and post the results.
BTW, I wrote a follow up to this article called 5 Years of Social Media in Google Rankings: Tumblr to Overtake Twitter in the SERPs
VJ Dost asked in the comments below why Amazon is showing up in more search results than Wikipedia? Are they getting a more unique pages indexed? VJ was dead on. Check it out:
So, the synchronicity that you see (both Amazon and Wikpedia surging at the same time) is almost certainly due to major increases in SpyFu's data collection. But, no matter what, you see Amazon pulling away -- getting more unique pages indexed than Wikipedia. With all this data, there's still the data bias that I talked about earlier. If I had to estimate that, though, I'd put it at 20-30% of this chart -- not 50-70% that is the difference here. So, yeah, I can completely validate VJ's idea here. Amazon's rise is about number of uniquely indexed pages (that are delicious to Google).
Search Engine Land summarized this article and posted it on their site, and in the comments James Finlayson posted something that made me want to clarify some things.