RapidMiner looks to step up advanced analysis business, adds to processing options Analyst: Krishna Roy 14 Jan, 2015 RapidMiner has its eyes down and focused on higher growth rates in its business this year, having closed out 2014 with a reported threefold increase in product revenue over 2013. The commercial open source advanced analysis player which transitioned to a so-called business source licensing model in late 2013 in order to increase product revenue has a number of plans in place to drive fresh growth. RapidMiner has also extended its data-processing and deployment options for advanced analysis as it continues to execute on a product strategy fashioned around a 'one platform to rule them all' approach. The 451 Take RapidMiner is a very different company than it was even a year ago. The restructuring of its business appears to have paid off, and the focus on indirect sales in 2015 should complement the momentum that management has built in direct product sales in recent years. While there are numerous vendors with an in-hadoop play for advanced analysis, RapidMiner's offering is differentiated. Moreover, the number of processing and deployment options the company now provides also singles it out from the rest of the advanced analysis crowd. However, cracking the all-important US market seems to be a work in progress. That endeavor could be aided by the infusion of some additional venture capital or another strategic tuck-in acquisition to diversify its portfolio into a complementary arena, such as BI. Context Copyright 2015 - The 451 Group 1
RapidMiner is looking to take its business to the next level in 2015, using indirect sales as a new avenue for growth. The firm began investing in a channel strategy in 2014, having shifted at the tail end of 2013 its business away from a pure open source play to a product-driven business model with a move to a new licensing strategy. Seventy percent of its business now comes from product revenue and 30% from services, where the percentage splits were reversed under its former open core model in 2010, according to management. The purveyor of advanced analysis software to data scientists and business analysts is also using a channel play to set the stage for further geographic expansion in 2016. Management reports that 50% of RapidMiner's revenue currently comes from EMEA, 42% from the Americas and 8% from the rest of the world. The German company opened headquarters in the Boston area in November 2013 to facilitate a more concerted sales push into the US. It reported 41% of revenue from North America, 47% from EMEA and the remainder from the rest of the world in our previous discussion last July. RapidMiner plans to use resellers to serve EMEA and also employ an indirect sales strategy for APAC. That said, we expect some regions of Europe to be tackled via direct sales, too, given the company already has offices in Hungary and the UK. RapidMiner's Budapest operation came about as a consequence of its reaching for ex-oem customer Radoop for its in-hadoop processing capabilities and expertise last July. The company has its headquarters in Cambridge, Massachusetts, and an office in Dortmund, Germany, which became a development operation after the company's headquarters moved to the US. Management is using a judicious hiring strategy for expansion purposes, with an eye to moving into the black in 2016. One notable recruit is Michele Chambers, who joined as president and COO in September. Chambers joined RapidMiner from MemSQL, where she was VP of marketing. Prior to that, she was chief strategy officer at Revolution Analytics. RapidMiner has also hired Glen Zimmerman as senior director of communications, reporting to Chambers. Zimmerman was previously part of IBM's analyst relation team for 'big data' and analytics. RapidMiner now reports having roughly 600 paying customers, compared with more than 500 accounts in July 2013. Management also notes a growing focus on Internet of Things (IoT) use cases from existing and new accounts. Although the company continues to amass customers in pretty much all vertical markets, it notes that IoT scenarios are often driven by manufacturers, which account for a significant proportion of its installed base. RapidMiner says it is also seeing escalating demand for IoT-based analysis in general a sector we think it is better placed to serve with its new offering for streaming data analysis. Copyright 2015 - The 451 Group 2
RapidMiner Streams was unleashed in December and will be built upon in subsequent releases. The initial version is integrated with the Apache Storm event-processing environment and designed for the processing of streaming data from social media, devices and wearable sensors. The aim is to enable Storm operations to be mixed with RapidMiner's operators to create predictive-analytic processes for scenarios including sentiment analysis and IoT-style analysis through embedding actions in manufacturing control systems. The company's Streams offering is the latest addition to its product portfolio, which was also augmented by the debut of RapidMiner Cloud last August. Although RapidMiner Server and Studio remain core offerings, the company has extended execution options and functionality in a bid to provide one platform for data integration, data discovery and preparation, as well as model building, validation and scoring. A 'one platform to rule them all' strategy continues to be the focus going forward. RapidMiner Cloud joins RapidMiner Streams and the company's Radoop offering for in-hadoop execution. The company's cloud service is all about enabling users to off-load a piece of the advanced analysis process, such as model building or sharing, from the RapidMiner Studio desktop code-free GUI to Amazon Web Services. The offering includes connectors to cloud data sources and a repository. It requires a subscription to RapidMiner Studio 6.1. Users also need to buy cloud credits for computation and processing requirements. Activation is $78 and includes the cloud subscription, which is $39 per month, and 100 cloud credits for $39. RapidMiner's cloud offering includes so-called crowd operator recommendations, which essentially cull insights gleaned from predictive models created by the company's open source community, in order to aid the analytic process. The Recommendations aspect is earmarked for further development this year. Management notes that it now has roughly 200 operators, which have been contributed by the RapidMiner community, within its advanced analysis platform. They include the so-called Wisdom of the Crowd Operators in its cloud service. The company cites more than 1,500 operators in total, including about 250 operators that tap into internally developed machine-learning algorithms. Competition SAS Institute and IBM/SPSS are dominant vendors in the advanced analytic sector and, in our opinion, provide frontline competitors to RapidMiner. While management acknowledges both vendors as rivals, it also cites situations where RapidMiner is complementary to predictive-analytic wares from these industry titans. Copyright 2015 - The 451 Group 3
In big-data advanced analysis situations involving Hadoop, RapidMiner reports that it is seeing greenfield opportunities. However, we wonder whether the company will bump heads with Alpine Data Labs and Splunk at some point. Alpine Data is similarly focused on providing discovery and predictive analysis of data in various environments, including Hadoop and purpose-built analytic databases, without data movement. Moreover, the startup has recently introduced a collaborative front end to enhance its appeal to business analysts and users a core target market for RapidMiner. Splunk's Hunk offering is positioned as an analytics platform for exploring, analyzing and visualizing data stored in Hadoop. While RapidMiner may have more of an advanced analysis slant to its Hadoop offering, IoT scenarios are where we see the company squaring up to Splunk. It is also worth noting that SAS and Big Blue have in-hadoop advanced analysis offerings, as do startups Datameer and Platfora, which may also appear in RapidMiner bake-offs in the future. We also think KNIME and Pentaho provide a rivalry when it comes to other open-source-based options for predictive analytics. TIBCO Software is also a potential competitor. TIBCO went private in December and peddles its Spotfire visual-analysis offering, which contains the S+ statistical programming language for advanced analysis. SAP/KXEN and Oracle are the other big guns in advanced analysis and, as such, are competitive. RapidMiner may also face Alteryx, which is similarly focused on making predictive analytics user-friendly. Other potential competition comes from machine-learning startups looking to simplify and scale advanced analysis in order to take it to business analysts and users. RapidMiner says it doesn't encounter the likes of BigML, Ayasdi, Wise.io and Nutonian, which are focused on this end game, because it provides an end-to-end advanced analytics stack with data preparation part of the mix. It will be interesting to see if this competitive situation changes as RapidMiner augments and plays up its machine-learning capabilities. SWOT Analysis Strengths Weaknesses RapidMiner has successfully evolved from an open source vendor peddling services to a purveyor of advanced analytics software. RapidMiner Streams opens the door to additional IoT use cases, while the range of deployment and processing options on offer in general make it more compelling than it was previously. RapidMiner is still a minnow. The company hasn't yet been fully able to crack the all-important US market. Opportunities Threats Copyright 2015 - The 451 Group 4
Additional venture capital would help to scale the company's business and enable it to better execute on its product roadmap. Another strategic tuck-in acquisition could add size and product girth to compete against advanced analysis heavyweights. SAS and IBM are often incumbent in the departmental enterprise where RapidMiner is largely focused. The range of in-hadoop execution options for advanced analysis is proliferating as is the number of machine learning startups providing the makings for a tougher competitive environment in the future. Copyright 2015 - The 451 Group 5
Reproduced by permission of The 451 Group; 2015. This report was originally published within 451 Research's Market Insight Service. For additional information on 451 Research or to apply for trial access, go to: www.451research.com Copyright 2015 - The 451 Group 6