My final session of the event is George Mathew’s General Session on the product and its roadmap. George recapped Alteryx’s laser focus on data analysts – who make up a big chunk of the audience here – and their focus on making the next best decision. It’s great to see the focus on decision-making not just on data and analysis.
George went on to talk about the typical project for a data analyst, often a solo effort with lots of pressure on time and resources. At the same time, he points out, analytics is increasingly differentiating winners. Bain, for instance, found only 4% of companies were really using analytics to drive their business but they were 2x as likely to be a top quartile performer, 3x more likely to act on their decisions and 5x more likely to make good decisions. To deliver the value of analytics in a timely fashion, he says,analysts need better tools – better than spreadsheets, better than web spreadsheets, better than custom code/scripting environments.
Alteryx 9.0 was delivered this spring with a focus on customer analytics, enterprise scale analytics and “unchained” analytics (releasing legacy analytics).
- Customer analytics is a key use case for analytics.
Companies are using more and more customer data to do customer analytics with demographics, purchase data, customer interaction data, 3rd party research and even mobile data , social media data, loyalty program data, POS transactions and in-store movement data. 9.0 therefore:
- Made it easier to get access to programmatic web data and easily bring it into the environment with a JSON Parser.
- New Social Media tools including support for Foursquare and Twitter.
- Linked to sales and marketing automation systems, adding Marketo to salesforce.com.
- Adding web analytics with an integration to the Google Analytics API
- More EDW support including Amazon.com RedShift
- Enterprise Scale Analytics is important as use cases get more complex
- Revolution has scaled the various R algorithms and 9.0 provided a workflow-based environment to access the Revolution R capabilities.
- The App building and visual query building tools in 9.0 were improved to make it easier to package up and share analytic apps that embed Alteryx workflows.
- The server product’s ability to scale out was improved.
- Unchained Analytics
- 9.0 provides direct access to SAS and SPSS datasets so that functions can be applied to existing analytical datasets.
With that we cut over to a demo focused on taking advantage of many of the new connectors to bring in a mix of legacy and web datasets, blending them and then creating predictive analytics using Revolution’s algorithms from them. The workflow can be parameterized (simply by adding additional nodes) and deployed as an analytic app in a gallery for others to use. Very nice.
After the demo, George came back to talk futures and started by discussing the management (and exploitation) of data. There’s clearly enormous value in data but this data has be focused on decision-making, on driving analysis that will help with decision-making. When you look, as George did, back over the history of data storage technology you see that storage technology is dramatically cheaper than it was even a few years ago. This cheaper storage is creating an age of data abundance. Many analytic tools, he says, started being developed when data was still scarce, when data storage was so expensive that data volumes were fundamentally lower. Newer tools, built on the assumption of data abundance, are therefore fundamentally different.
In addition organizations are finding that they can use Hadoop to manage the whole traditional ETL process. The availability of cheap compute power means that processing can be moved to the data and that all the data can be managed more flexibly, taking advantage of Hadoop’s facility for applying the schema of the data at read time rather than when the data is stored. The Hadoop 2 capabilities adds flexibility to this scale, especially around real-time in addition to batch processing as well as SQL/Search.
Cloudera is partnering with Alteryx and Alan Saldich VP Marketing of Cloudera came up to talk about the relationship. With Hadoop 2, he says, Cloudera is delivering an enterprise data hub. This allows multiple analytical techniques to be applied and ALL the data to be used. Cloudera and Alteryx have some joint customers and are working on some joint announcements. BTW iIf you are interested in R, PMML and Hadoop, check out this research paper I wrote.
Alteryx has been working with HiveQL and SQL (running onMapReduce) for a while. In 8.6 they added support for Impala for more direct access. With Yarn, part of the Hadoop 2.0 environment, has allowed more engines to be deployed on Hadoop. As Alteryx moves forward they plan to add support for direct HDFS2 access and expects to support in-Hadoop R execution as new R engines take advantage of the Hadoop 2 platform. Support for in-database functionality is also planned, with nodes allowing functions that are available in-database (or in-Hadoop) to be included in the workflow.
Another live demo of this capability followed, showing both live connections to HDFS and some execution in-database using new nodes in the workflow.
Data discovery is, he says, an increasingly disruptive element in the BI market. In fact Gartner feels that a majority of spend may move to data discovery and this is going to disrupt existing BI vendors. With that Tableau came on stage to talk about their joint work in data discovery. Growth, Tableau says, is coming from the move to self-service analytics with more empowerment of analysts and a mindset focused on problem solving. The combination of Alteryx’s data blending preparation tools, as well as the ability to easily enrich data in Alteryx, are powerful tools for Tableau users. Similarly Alteryx sees clear value in the manipulation and presentation tools in Tableau. The two companies have lots of premier partners in common as well as many joint customers. Moving forward Tableau wants to build on the interactive experience that is at the heart of their product, adding some proactive suggestions around data and what might help with a specific problem.
A final demo showed how to output the data into Tableau (or update an existing Tableau workbook) and interact with it.
That’s a wrap – lots of data discovery stories and capabilities with a nice focus on the importance of making decisions. George concluded with a reminder that there is a new analytic stack, encouraging his users to move forward and focus on the new way of doing analytics not legacy BI and analytics tools.