Why is the number of data items (e.g., tweets) different from the Sifter estimate?



Unfortunately, the Sifter service and website has been decommissioned as of September 30, 2018. Thanks to our 6,970 users who created 16,128 free estimates from the complete, undeleted history of Twitter between 1/14/2014 and 9/29/2018.

Please contact Twitter for approval of future academic or commercial use cases. If you can get an approved use case from Twitter, we can still help you work with the data inside DiscoverText.

@DiscoverText remains open and is still the top-ranked text analysis platform on the Internet.

All paid jobs prior to the decommissioning will still be honored.


I created a rule to search for Twitter tweets to get an estimate so I could estimate the number of tweets that would be returned and the cost. The estimate provided around 35,000 tweets, however the final information only has 31,000 tweets. Why is the actual number of data records different from the estimate?

Twitter provides an estimate based on six 1-minute samples over a 24-hour period. The actual amount of data (activities) available cannot be known in advanced; it can only be estimated based on this sampling technique. The query rule is run against an ever-shifting population of Tweets and deleted Tweets using an estimate provided by Twitter.


Have more questions? Submit a request


Powered by Zendesk