5 Signs Your Analytics Program is at Risk
Aug 5, 2015
Commonly, I see companies making an investment in “Big Data” – today’s favorite term within companies trying to gain an edge on their competitors. I once saw a post on my LinkedIn feed that said, “Big data is like teenage sex – everyone is talking about it; everyone is saying they are doing it; few are really doing it.” Just because you invest money into “Big Data” does not mean you are following the “All-in” approach and supporting it 100 percent.
Five key questions can help determine whether your company is geared for success.
1. Do we have the appropriate people?
Understanding that we need to start somewhere, it is not uncommon to start with one expert – this person can build a plan that will incorporate the appropriate employee structure, technical plan, and analytical working grounds. Companies sometimes struggle moving past this point – they do not see the perceived value given the work that has been accomplished to-date. They want to see data, advanced models and results immediately.
To move past this issue, you can develop a sound plan and communicate it to management – this is where you set expectations and needs.
2. Do we have Management’s initial buy-in?
Arguably the most important, is management in full support of developing an analytics program? Are they willing to give up access to certain sensitive data for the benefit of the company? On too many occasions, I hear executives and managers say “we can’t wait to start using analytics to help guide our business decisions – this is something we have been looking forward to for a long time.” But, when analysts try to gain access to the source system data, they are commonly met with the dreaded steel wall of “well, that information is too sensitive – we can’t give you access.”
This issue is one of the trickiest to move past, especially when you do not know what data is available in the source system of record. But focus on the common sensitive data elements – explain/show what you could do with that data and how the company would benefit from it.
3. Do we have access to source data?
While section two discusses management’s affect on an analytics program, this section focuses on the best results being driven from the most granular datasets. When constructing advanced analytics, the more granular the dataset the better. The most statistically accurate models will be derived from the lowest level of data because of its complete flexibility and number of data points. I have seen smaller companies use all-encompassing workflow systems of record that store sensitive human resource and financial information alongside data that analysts need to perform their job adequately. Much like in section two, management throws up the “Stop Sign” and blocks access to data, causing analysts to use higher-level aggregates from “canned” reports that are not all encompassing – this leads to inaccurate models that further lead to incorrect business decisions.
The way past this blocker is similar to the initial buy-in of management. How can management benefit from you having access to the data? Aside from improving the accuracy of advanced analytics, a benefit could be that one analyst/ database administrator could restructure this data to parse the sensitive information from the non-sensitive data. All while blocking future analysts from seeing any sensitive data, but still being able to access the data needed to perform their job well.
4. Are analytics driving operational strategy?
If you are not struggling with management’s initial buy-in, and you have moved on to the development stages of analytics. Are managers starting to use analytic results to drive operational strategy and better business decisions?
The key to ensure buy-in here is by creating analytics and visualizations that are easy to use within the already existing organizational structure. If management feels it makes life easier, they will communicate this message to their team.
5. Do we have reasonable expectations given our budget?
Some managers hear “Big Data” and equate that with all the information in this world tying perfectly together into a centralized environment while updating instantaneously. While this may be physically possible, it is not reasonable given the budget most companies are working with.
Once again, set expectations. If we have limited data and do not have a need for real-time data, why throw a ton of money into a really sophisticated solution that pulls and transforms data multiple times a day, pre-calculates data via a cube, stores data on expensive hardware in redundant locations for failover and server load purposes, has a business intelligence application sitting over the transformed data, and has integrated security – you are just wasting money. Instead, focus on a plan to make your solution scalable. Start by determining your business objectives. From this, we can collect and analyze the information, build conceptual data models, and begin to transform the data into a more usable form. This process can run nightly and be stored in a less expensive environment. Focus on the data – if you connect it properly and design a model that allows you to tie in new data easily, growing to a larger more advanced environment should not be a problem.
If your organization is experiencing one or more of these issues, please don’t feel as if “the sky is falling.” These barriers are extremely common in companies, especially ones that are in the early stages of building an analytics program.
Please stay tuned for my next post on a framework that will allow you to break through these barriers and make significant strides towards building an effective analytics and business intelligence program. In the meantime, I would appreciate you sharing any barriers that you have experienced/witnessed that hinder the success of these programs.
Learn more about the Marketing Change Management group