Cloud Analytics: Six Tips For Success
Cloud Analytics: Six Tips For Success
Doug Henschen is former Vice President and Principal Analyst where he focused on data-driven decision making. Henschen’s Data-to-Decisions research examines how organizations employ data analysis to reimagine their business models and gain a deeper understanding of their customers. Henschen's research acknowledges the fact that innovative applications of data analysis requires a multi-disciplinary approach starting with information and orchestration technologies, continuing through business intelligence, data-visualization, and analytics, and moving into NoSQL and big-data analysis, third-party data enrichment, and decision-management technologies. Insight-driven business models are of interest to the entire C-suite, but most particularly chief executive officers, chief digital officers,…...
Read moreWe still live in a hybrid world, but data analysis is moving steadily into the cloud. Here are six tips on how to succeed with online BI and analytics.
The cloud movement that started more than a decade ago has turned into a stampede in nearly every category of enterprise software. It’s no different in business intelligence and analytics, as I observed last month, but there’s no one set path or approach to bringing data-exploration and analysis into the cloud.
One argument is that data analysis is moving into the cloud because more and more data lives there. Google Analytics and social networks like Facebook and Twitter, for example, keep generating more data. And storage-and-analysis options like Amazon Redshift are among the fastest-growing cloud services. In fact, data is originating and being dispersed all over the cloud, according to Adam Hall, head of technical product marketing at Google Cloud Platform, so companies are placing bets on multiple platforms.
Google’s case for moving (big) data analysis into the cloud is succinct: “Focus on insight, not infrastructure.”
“We’re seeing customers setting up [virtual] containers that they can run anywhere, and most assume they’re going to be running in multiple clouds, including public and private,” said Hall, one of several executives on a panel discussion I recently had the pleasure of moderating in San Francisco.
Another panelist, Brad Peters, chairman and chief product officer at cloud BI vendor Birst, offered a six-point recipe for cloud analytics success, which I share below along with other useful observations from the event:
1. Start with a clear use case. Tor Stahl of Practice Fusion, the practitioner on our panel, took this approach when he waded into cloud-based analysis. Practice Fusion is a Web-based electronic health records platform for doctors and patients. Due to HIPPA compliance concerns, it has, until recently, managed all of its data in on-premises company data centers.
Since 2006, Stahl, the company’s data services & business intelligence program manager, has used Tableau Desktop to analyze Practice Fusion’s HIPPA-regulated data and Tableau Server to analyze aggregated (sans personal health information) data. Earlier this year, Practice Fusion chose cloud-based Tableau Online as the platform for a new online service for pharmaceutical companies.
The new service provides insight into the use and effectiveness of competitor drugs within a cohort of patients with a specific diagnosis. Before launching the service, Stahl says Practice Fusion ensured that a third-party firm could de-identify and certify that the data sets used could not be reverse engineered to reveal identities.
Practice Fusion’s new service was a good fit for the cloud because it was destined to be consumed outside the company’s firewall. Indeed, the most common adoption pattern for Tableau Online is “organizations that need to share and collaborate with data outside of the firewall or when they want to support mobile employees, such as salespeople,” said Ellie Fields, VP of product marketing at Tableau.
2. Focus on outcomes, not features. Amid all the hype about big data and next-generation technologies, it’s easy to dream about state-of-the-art capabilities. But Arsalan Tavakoli, VP of customer engagement at Databricks, the commercial company behind Apache Spark, relayed a cautionary anecdote about a retailer who inquired about the Spark Streaming engine. “When I asked him about the need for real-time data [on orders], it turned out they updated their shipment data only once a week,” Tavakoli said, noting that the company would have little opportunity to act on real-time insights.
Think about the outcomes that are desired versus those that are practicable. Unless those latest-greatest features, like streaming analysis, can really make a difference in your business, maybe it’s better to settle for more routine performance specs.
3. Take an iterative approach. If you’re moving to the cloud adopt the ways of the cloud by embracing agile, iterative development. As on-premises practitioners know all too well, if you spend too much time defining requirements and getting everything right, users are likely to shrug their shoulders and say their needs have changed by the time you deliver new functionality. Start small with a single analysis. Then add depth and breadth incrementally trough rapid release cycles, gathering user feedback during every step along the way.
4. Build the right team. Peters of Birst advises that you need a strong project champion who understand the business case and the data available as well as a capable “ninja” who has the technical skills to refine and combine the data and develop the analyses.
5. Ensure adequate resources. Resources include time, talent and treasure, and they’re essential not only during the initial implementation, but also on an ongoing basis. Projects and programs that are starved for these resources are doomed to failure.
6. Enlist a visible and supportive executive sponsor. The higher-level and more visible the sponsor, the better.
MyPOV on Cloud Analytics Success
Most of the points above apply to on-premises projects as well as those in the cloud. If there’s one theme I heard from every panelist, it’s that agility and reduced resource requirements are the cloud difference makers.
Have a use case you want to get up and running quickly? Think cloud. Want to take an iterative approach? Spin up sandboxes and pilots in the cloud. Ready for production? Add compute and storage capacity at the push of a button. Worried about building the right team? Don’t add server and storage babysitters to the list; go cloud and focus on the data and the analysis.
Need to ensure adequate resources? Here’s where you’ll have to ensure that your executive sponsor is signed on for the cloud rather than vested in on-premises agendas. In my view most enterprises are entering into and will remain in a hybrid world, but they’re on a one-way street to the cloud. So many executives I talk to – including Stahl of Practice Fusion – say they have cloud mandates, even if they accept that some things will remain on premises.
Health care businesses and banks are among the more conservative types holding back on cloud deployments, but that’s changing. When I pointed out that Google, Birst and other cloud players like GoodData have achieved HIPPA compliance, Stahl admitted that he’d like to move more (non-PHI) data and analyses into the cloud. For now, Practice Fusion has only recently launched that first cloud analytics use case. Next comes the steady, iterative expansion into new areas.
