We do not "keep it" but we do build tools to estimate and license historical Twitter data:
You can go back to the start of Twitter in late 2006, but deleted Tweets are excluded. Some rules, like enhanced geo profile rule, do not go back before August 1, 2013.
You can create very simple search queries, or much more powerful filtered queries, including for country code and a variety of other geographical constraints:
If you create a free sifter estimate and you want to license the data, then we will put the results in your DiscoverText account, which is a subscription service. To export from DiscoverText, you need to obey the Twitter Terms of Service and also have a paid account on DiscoverText for at least one month.
Please let us know if you need further assistance using Sifter or understanding the Terms of Service. The Gnip Historical PowerTrack rules can be difficult to use at first. It really pays to start simple and build successively more complex rules. Since estimates are free, we encourage experimentation.
- First, please visit sifter.texifter.com to generate up to 3 free estimates per day. The system will send you an email with a cost that corresponds to the number of tweets (or "Activities" in Twitter's language). An example estimate email is shown at the end of this FAQ.
- If you elect to make a purchase, we will download all of the data available from Gnip's Historical PowerTrack and store it in a DiscoverText account.
- Our pricing model is simple:
- Three free estimates per day
- $25 per day of data retrieval
- $7.50 per 10,000 tweets
- Any Sifter purchase includes 14 days of Enterprise DiscoverText access. This enables up to three users to collaborate on the dataset.
Realtime Gnip Twitter collection via DiscoverText avoids the $25/historical day charge on Sifter.
- No refunds are allowed. Please see our terms of service.
- Please see http://sifter.texifter.com/Home/Pricing for more pricing information.
Remember, we have the only web-based estimate tool for the full history of undeleted Tweets. It is self serve at sifter.texifter.com
You can learn about it here:
There are a number of FAQs here:
If you do purchase access to data, we load it in a powerful Web-based system for text analytics, called DiscoverText:
You can use DiscoverText completely free for 30 days. We offer a 75% license discount for students.