DataSift allows customers to aggregate, filter, and extract insights from public social conversation on Twitter (including real-time and historical tweets back to January 2010), leading social networks, and millions of other sources.
Bundles range from $1,000 to $15,000 per month – pay-as-you-go options from $10 to $1,000 for trial streams and small data users.
Doug O'Reilly, VP of Insights at MWW Group, has been using DataSift for about eight months (the product was in beta for the first six).
How do you use it?
There are several options. DataSift has an option for creating stream definitions and storing data. You can also access the data via an application programming interface (API), which is the route we have taken. That requires our own data storage and analysis servers, which we host via Cloudant, and our own front end for the analytics and visualizations.
We developed the user front end and our proprietary measures of trust, relevance, share of advocacy, and a few others based on the data.
Problems can theoretically occur on a few fronts, but we haven't encountered any significant ones. Our own server cluster has 24/7 monitoring, and the DataSift API hasn't thrown us any problems since we finished our initial development. Because DataSift allows historical data access, we could fill in any holes if something did go down.
How does it serve your business needs?
It allows us to create and customize measurements specific to clients. For example, no one measures on trust, so I hired people with PhDs who created algorithms that score social content on how it contributes trust to a brand. One client wants to know what content is promoting advocacy. Again, nobody measures that, so we took DataSift data and built it out.
We created internal dashboards and now we can derive insights that no one else has for new business pitches and existing clients.
I can also get the full meta information behind a tweet – more than 100 pieces of data. DataSift automatically appends important information such as cloud information, sentiment, and semantic tagging, and I can add my own tags.
Because we have raw data, we can figure out what's driving a trend from a bottom-up perspective.
DataSift supplements other systems. Sysomos or Radian6 only present processed and analyzed front-end data – like a prepared meal. DataSift gives you all the ingredients to make a meal.
How does it integrate with your existing infrastructure from an IT standpoint?
It doesn't integrate per se. We built an infrastructure that's layered on top of DataSift with its own servers to help us take in and mine the data. We have tools to make it fully web accessible.
What are the main benefits?
Being able to build a measure and analyze data in a manner that aligns with clients' objectives. You shouldn't set objectives based on what measurements already exist. You should build measurements to match objectives, and DataSift allows that.
What are the main drawbacks?
If you really want the full value, you have to build a little bit of your own infrastructure. That takes commitment and investment. But it's completely worth it.
What would you like to see improved/added?
It's improving all the time. I'm looking forward to new analytics tools. Hopefully some sources will expand too – it's got many good sources, but it's not 100% comprehensive.
Gnip – provides access to more than 100 real-time social media data streams from dozens of social media sources, including all Twitter data. Enrichments augment raw data.
Radian6 – captures conversations across Facebook, Twitter, YouTube, LinkedIn, blogs, and online communities, and provides real-time actionable insights.
Sysomos – social media analytics; services include Heartbeat, a real-time monitoring and engagement tool, and MAP, a research and analytics tool.