Challenge assumptions. Don't take things as absolutes.
1. Culture Shift is Crucial
No two organizations are alike; there is no any easy or quick way to achieve advanced analytics. It requires a cultural shift in a company which is not used to this concept. For organizations which are accustomed to either guesswork or intuition for decision making, it may require a "company culture" shift to start becoming more of a data-driven organization. It definitely would be helpful if the C-Level executives are supportive of the data strategy efforts.
- Get to know the organization’s culture and people (data is easy, people are hard). Start with familiar concepts, data sources, and methodologies. Introduce new concepts, data sources, technologies, or methodologies slowly. Get buy-in wherever possible for the above. Invite open and constructive commentary to support an "analytic culture:
- Educate on how to "action" insights
- People and process — as well as technology
-Wash, rinse, repeat…
2. Identify Decision Points & Low-Hanging Fruit
Analytics works best at the strategic level and then the existing customer relationship part of the organization can benefit from these predictions. The customer service organization culture has to be already to benefit from the analytical capabilities. Talk and listen to the clients, understand their pain points, identify the "low-hanging fruit" where there is significant value for at least one group but the relatively easy implementation of the solution, then promote that initial success with other groups to build momentum.
- The first thing would be to identify clearly how and where you can use analytics in business decision making in business. Firms would invest time and money in analytics when the goals and benefits of using analytics are clearly defined.
- Start with a problem that will allow you to do one of two things or both: 1) data-driven discovery, or 2) theory-based exploration. Use that problem definition and then assemble a very skilled team to do a Proof of Concept. But, be very flexible to the "what if" analysis that is capable.
- Challenge assumptions. Don't take things as absolutes, especially as the data invariably has discrepancies in it. Grab the data from as close to the underlying production systems as possible as that usually represents "the truth."
- Amplify the best Practices: Often times, this internal marketing may involve doing separate demonstrations and presentations to numerous groups to help them become aware of the potential benefits, and then listening to them for new potential use cases. Not all the business problems can be solved by analytics, so it helps to build credibility to offer the best solution even if it doesn't involve analytics.
3. Get Data Ready
Essentially it comes down to where the company is along the analysis sophistication spectrum.
- Following are some of the important things to look at before even thinking abut analytics
a. Ensure enough historical/real time data is available
b. Data quality is a very important aspect.
d. A Clear understanding of the senior management efforts required as this requires extended duration to achieve results. Management should be ready to invest time and efforts.
- Get data standards agreed upon up front as it will cause the problem if the raw data is/was gathered and processed in a manner inconsistent with good BI practices. Start with the easiest to implement that will convince the company that this is the way to improve their business
- Use an interactive, visual analytic tool/environment in which allows you to do deep data inspection as well as scenario building. Prioritize the findings into some level of ROI or impact on the business. Push a team to validate and/or do root cause analysis of findings. Use those findings to drive business decisions, then as a team takes actions. If you don't take actions, then it's just interesting data - not analytics. For the POC, include at a minimum really good analyst, a data subject matter expert, and a business process expert (or someone familiar with the current business processes).
4. Implementation Scenario
Building up successes one step at a time, rather than invest in an initial huge, highly visible project which might take a long time to deliver but might not have any clear benefits in the interim
- Management MUST be open to transparency. Outlining a fresh vision and strategy based upon the adoption of critical thinking, highly effective communication and analytic methods may facilitate a robust change more effectively than making smaller changes. Sometimes, any lesser commitment really represents no commitment at all.
- Starting small and finding champions. Have a champion at the top who is convinced about this technique. This is since you need a significant amount of data to run any analytic technique. This requires investment into data management technology and analytic talent. Moreover, it’s only over a period of time that the results can be acted upon creating a collaborative analytic environment or even an embedded analytic system (fraud management system, revenue assurance system, commission analytics, and customer analytics) which are repeatable and drive consistent business value, for more mature companies
5. Measure the Right Things in Improving Decision Making
It is important to make sure that you are using the right inputs and a model that adequately fits the problem to carry out the analysis. Otherwise, the recommendations may not be optimal, make sure to measure the right things, and pass the "does this make sense" test.
- It’s both drive and measure to improve decisions. When focused on operationalizing analytics, it means to implement predictive analytics results in operational systems in real or near real-time, analytics can 1) Drive (automate) decision making, and by monitoring those decision results analytically; 2) Measure and 3) Improve the decision and therefore the overall system and business performance over time.
- Analytics can be very useful to measure "success" or improvement between different scenarios. Once more, it is important to be sure that the right things are measured to come to correct conclusions about the performance. Run old reporting and analysis side by side with the new comparing the two to show a change in the results where applicable.
- Data Speak: The business clients might not be interested in the actual technical details as to how analytics generates insights for decision making, but the actual measurable results should always be able to win over believers.
Properly done analytics should drive decisions. Properly done analytics will give you information that you wouldn't have otherwise. Decision making is both science and art, it takes both data analytics and intuition in order to make effective decisions; analytics is important, but do not forget common sense.