original article at Forbes
Analytics has gone mainstream. We live in a world where #bigdata and #hadoop trend on social media. Most CEOs talk about being data-driven. Companies in a hurry to reap the rewards may treat analytics like a fad. Analytics is not a hairstyle, band or meme.
Getting value from analytics demands an infrastructure where data can be properly assembled, organized, and harmonized and people have self-service access to analytics. When all of those things work together, you gain a deeper understanding of the business and can do amazing things.
As often happens with fads, people are well meaning and enthusiastic, but not quite on the mark. Here are some signs that your analytics program may be a fad along with advice on how to create an analytics program with staying power.
1. I’ve given users a BI tool; that’s analytics
Simply providing your team with a BI tool is not analytics! Advanced analytics delivers statistically significant results that tell you what’s happening right now and what is going to happen in the future. BI tools are improving but most don’t have analytics of that caliber as part of their functionality and really focus on providing pretty pictures about what has already happened. Providing BI tools is an important step, but it’s not the same as having a full-fledged analytics program and employing experts with the statistical understanding to draw valid inferences.
2. We can use data from anywhere, just like it is
One of the key objectives with analytics is to gain insights using data from across the organization. You can’t run effective analysis on data that has already been aggregated or summarized at organizational tiers between the analysis and the original sources. In other words, ERP data or any other data source, which may have been aggregated in various ways, doesn’t cut it. Effective analytics is predicated on pure, clean data from the source, and the less aggregated it is, the more flexible, consistent, and statistically valid are the analytics you can perform.
3. Just use the freshest data you’ve got
If you have one data source that updates every day, another updated once a week, and a third that refreshes once a quarter, meaningful analytics can only take place if sources are synchronized and current at the time of analysis. Let’s say you are a manufacturer sitting down to sign a contract for a large order of widgets that include silver in manufacturing. If you calculate the contract using last week’s silver price, you could be in for a nasty multi-year surprise as silver fluctuates multiple percentage points every day. That example underscores the importance of having up to date data when you run analytics. Obviously there are exceptions to the rule (we have 50 states in the US, and that doesn’t change frequently), but be sure that your data is synchronized and fresh, especially where freshness matters.
4. We need the latest hot tool
Similar to the first sign, some think that getting users the latest hot tool really means they’ve bought analytics. Don’t buy a tool just because it’s the latest fashion. Provide users access to tools that they are good at using. If your enterprise standard is Tableau and MicroStrategy, and people are used to using Spotfire or SAS, you’re limiting their ability to get value from data if you take away what they are accomplished at using. Sometimes, the best analytics tool isn’t the most advanced one, but the one that people can use well. I’ve seen great things accomplished with Excel and access to the right data.
5. Keep it all under control
If you are going to be data-driven, don’t hide the keys to the sports car. You can get everything else right—the infrastructure, the tools, the data harmonization and synchronization—but if you lock the car in the garage, you’ve missed the point. If you build a hot rod of an analytics environment, don’t be afraid to let users take the car out for a spin. Of course you must pay attention to governance, but it’s surprising how many organizations get everything right, but then won’t provide self-service capabilities to let people benefit from access to data and tools.
Being data-driven means letting people take the data out on the track by providing them with self-service access and ad hoc queries. Only in this way can you and your organization answer new questions, explore uncharted territory, and find the insights that only unfettered analytics can offer.