Kick off each Monday with the best news and ideas in social media.
Twitter Does 19 Billion Searches Per Month, Beating Yahoo & Bing (Sort Of)
How many searches does Twitter handle per month? We’ve finally got a figure: 19 billion. That has Twitter doing over nearly five times the queries that Bing handles and about 20% of those that mega search star Google processes. But hang on. Even Twitter cofounder Ev Williams says the figures aren’t all apple to apples.
At Twitter’s Chirp conference today, Williams said that Twitter does about 600 million queries per day. Catching him after his talk, I followed up to gather some more specific figures.
So Twitter does 18 billion searches a month, working off the 600 million per day figure? Williams said the monthly figure is closer to 19 billion.
How’s that compare against the major search engines? Working off comScore figures from December 2009 for worldwide search queries, we have:
- Google: 88 billion per month
- Twitter: 19 billion per month
- Yahoo: 9.4 billion per month
- Bing: 4.1 billion per month
Those are “property” figures, so any searches run on a Google property (Google web search, image search, YouTube) gets rolled up into the overall figure. Google was ranked tops in the world; Yahoo second, so that puts Twitter into the new number two spot.
Now for the caveats. For one, we’re comparing Twitter’s self-reported figures to comScore’s estimated figures. To date, comScore hasn’t reported Twitter figures. Twitter doesn’t even register on the radar screen.
This is most likely because of the second caveat. Most of Twitter’s traffic isn’t happening at Twitter itself. Instead, it’s happening through API calls — a system for partners to send a search to Twitter and get the info back. Ratings services like comScore typically don’t include such queries, instead focusing on traffic they can monitor happening at specific web sites.
In his talk, Williams said the vast majority of Twitter searches happen through API requests. So who’s leading in those requests?
No single partner generates a majority of Twitter’s search requests, Williams said, but there are some that are noticeable. In particular, he said TweetDeck and Seesmic (both makers of Twitter clients) generate noticeable amounts of searches. That seems to due to many people using those applications to run standing queries on subjects they are interested in (see How To Track Keyword-Based Tweets Within Your Twitter Stream).
How about Twitter’s big search engine partners, like Google and Bing, both of which offer Twitter search. Queries there don’t add into Twitter’s overall figure. That makes sense. Rather than those search engines placing search requests through the Twitter API, they’re instead pulling in tweets in the way they index web pages. They store those tweets in their their own databases and query that.
So that’s another “apples to oranges” caveat. I’d say relatively few people organize standing, automated queries that hit Google and Bing. In contrast, it seems like a sizeable number of Twitter users may be constantly issuing automated queries. That’s not bad — it doesn’t mean that those users are less important somehow as searchers. But it can contribute to Twitter’s figures in a way that the more traditional search engines don’t benefit from.
Another factor, Williams said, is many queries are generated by widgets inserted on web pages, widgets that basically pull back search results for display.
While the figures may not be directly comparable in terms of consumer actions, they are comparable just from an engineering standpoint. While Twitter’s taken plenty of flack for dependability issues, it’s a huge accomplishment that by and large, Twitter has been managing to deliver so many searches at such a scale to partners.
The growth is continuing. I also spoke with Twitter’s director of search Doug Cook, who said at times, queries per day reach 750 million — and he expects Twitter to have a 1 billion searches some time in the coming months [NOTE: Previous I’d written next month, IE, May — but I’d jotted that down incorrectly).
Cook also said that Google Reader is another noticeable contributor to the overall query figures.
How about the situation that today, Google announced that it will have a more comprehensive index of old tweets than Twitter has (see All The Old Tweets Are Found: Google Launches Twitter Archive Search?).
Cook was non-committal but gave me the impression that Twitter might be more focused on producing a more relevant search experience for recent tweets rather than expanding comprehensiveness for past tweets. That would be in keeping with the company’s recently launched popular tweets feature within search (see Twitter Quietly Rolls Out ‘Most Popular Tweets’).
How about Twitter Search itself? How much does it contribute to the overall queries that Twitter handles? Williams told me its in the low double-digits.
Tomorrow there’s a session on Twitter and search at the conference, so expect more news and perhaps more specific figures to come.