In a recent study, Stone Temple Consulting made an attempt to crack the way Google indexes tweets. The main reason to conduct this study was to ascertain the likelihood of Google using signals from Twitter for ranking purposes, however, in the process a lot of other information were revealed:
- A significant percentage of tweets is not indexed by Google.
- The tweets indexed by Google are highly biased to people with 1 million followers or more.
- Indexing is not particularly fast for even those high authority accounts
Here are the details of the findings:
Twitter Indexing Information
Twitter’s IPO filing report mentions, Twitter handles more than 500 million tweets per day on an average. The image below displays the pair of Google search queries used to find out the number of Twitter pages Google has in its index:
Between the two queries, less than 1.5 billion pages are seen, which is not a large number when 500 million tweets per day are taken into consideration. To get more accurate results, the process was broken down further into the number of Twitter pages Google was indexing per month, the study found this:
To accomplish this, the search query operators shown in the graphic above was utilized by first clicking on “Search tools”, followed by “Any time,” and “Custom range.” The calender feature can be used for selecting a range of dates. This was also done on a month-by-month basis for each month from January 2012 to June 2014. This is what was found:
Detailed Research on Indexing of Tweets
In this part of the study, an analysis of the indexation of posts for 963 different Twitter accounts was included. Twitter and Google APIs were used to pull the last 20 tweets from each of these accounts and track the indexation levels in different ways. The numbers of the accounts included in the study were broken into categories as follows:
- 26 accounts with more than 5M followers
- 9 accounts with 3M to 5M followers
- 23 accounts with 1M to 5M followers
- 20 accounts with 500K to 1M followers
- 71 accounts with 100K to 500K followers
- 199 accounts with 10K to 100K followers
- 615 accounts with <10K followers
Aggregate Indexing of Tweets Over Time
The data was analyzed to check the percentage of tweets being indexed, without considering the number of followers. The details are mentioned in the graphic below:
In aggregate, out of 10,453 tweets, 326 were indexed at 3.12 %.
Also analyzed was the indexation levels for tweets more than one week old:
The indexation level for tweets more than one week is relatively low. Out of 19,389 checked tweets 701 were indexed at 3.62 %.
Breakout of Data by Follower Count
Here are the results for breakout of data by follower count:
The survey shows indexation level of tweets for users exceeding a following of 1 million+ is relatively high. If you are within the 10,000 or below followers group, there is a 0.22% chance of your Tweets being indexed.
Indexation of High Profile Tweets Over Time
When the indexing levels of high profile tweets over time on a day-by-day basis was analyzed, this is what was found out: Even tweets from high profiles are not indexed quite fast by Google.
Why do Tweets Get Indexed?
The tweets were broken down into different categories to see the impact of indexation. For the analysis, five Twitter profiles which had the largest number of followers, and five Twitter profiles with the most inbound links were reviewed.
For the profiles with the highest follower counts, there was 80% indexation and five profiles with the strongest link profiles, there was only 20 percent indexation.
Also analyzed was the number of indexed tweets being news oriented or with a link. The details are mentioned in the chart below:
The investigation was repeated by examining the makeup of non-indexed tweets. The details of the investigation are mentioned below.
Twitter Links are NoFollowed
The source code for a tweet was also analyzed. As with Facebook, this link is NoFollowed, so a PageRank is not passed by the link:
This is common on social media networks, largely due to the content being user generated, and this makes the value of that “endorsement” suspect. You can read the full report here.Why Do Some Tweets Get Indexed by Google and Others Donâ€™t, Explains New Study!,