There are some limitations of using analytics in business. While predictive models can be built which predict a very high probability of an outcome, in the real world sometimes predictive models don’t pan out the way they expected. This is why analytics is an iterative and incremental process. You can always move closer to the objective truth or effect of a situation but it isn’t always 100% causal.
Inhibitors to successfully using analytics in business includes poor goal decisions that won’t add economic or other value to the firm. Also choosing independent variables first can often lead to poor decisions during goal definition. Not joining or adding in other data can also inhibit the success of using analytics in an organization. Choosing poor goals and associated dependent variables that aren’t tied to adding economic value can inhibit the success of analytics in a business unless it is related to new product development or other aspects of business that test proof of concept (if it makes sense financially).
Accelerators to successfully using analytics in a business includes broadening our analysis with more dependent variables, while deepening it with more independent variables. Using the five phase process while iterating and being incremental also accelerates analytics success. Joining in other data sets can also aid in the success of using analytics in business. Lastly having dependent variables that are tied to adding economic value to the firm are also important.
These are some very useful capabilities in SAS:
SAS data visualization – Data visualization allows complex data to be represented and presented to team members and relevant stakeholders in an analytics project in a meaningful simple way. It is most useful because it allowed our group to be on the same page and revealed insights in a straightforward and manageable way.
SAS correlation – Using R squared and correlation analysis through SAS revealed key insights into showing positive relationships between independent and dependent variables. It is most useful because it revealed a lot of treasure for our group and allowed us to focus in on what independent variables can to a certain degree predict dependent variables.
SAS analysis of variance (ANOVA) – ANOVA analysis allowed us to determine what percent chance the relationships that were revealed were due to random chance. It is most useful because this showed what was and was not statistically significant. This allowed our team to narrow down on the variables that were statistically significant which helped our analysis.
If you’re the founder of a new analytics company or the leader of a major new analytics initiative within an existing company, I would recommend focusing on strategy, new product development, marketing, social media, mobile, customer service, and financial reporting in terms of cash flow feasibility. These cross-functional analytics aspects are crucial to achieving sustainable competitive advantage in any industry.
Goals I would define:
Demand for a new product or service
ROI expectations for different marketing initiatives
Customer Service Rating of 4.5 stars or more
Average Customer Acquisition Costs
Average Customer Lifetime Value
Data I would collect:
Data on customer perceptions of a new product
Data on what leads a customer to leave a good or bad review
Data on acquisition costs and reasons for high or low acquisition costs
Data on lifetime value and why a customer would or would not be a repeat buyer
Types of Models I would develop:
Social media engagement models (expected likes/follows/comments per post/day/month)
Expected demand for a new product or service
Expected customer reviews of a product or service
Expected customer acquisition costs
Expected customer lifetime value
By using a five phase framework for analytics projects analytics leaders can successfully implement analytics-driven management and add significant economic value to the firm.
Phase 1: Define business analytics proposal, data required, data analysis approach, and decision making and innovation framework.
Phase 2: Define detailed plan for data collection and data analysis.
Phase 3: Present and discuss data collected, data analysis, insights revealed, and actions taken.
Phase 4: Present and discuss updates to data collected, data analysis, insights revealed, and actions taken.
Phase 5: Finalize your business analytics project and data-driven decision making and innovation; In-class presentation and/or demonstration
The following checklist is very valuable for optimizing this comprehensive end-to-end approach:
Throughout the five phases it is important to ask yourself “what action should I take based on this data?” If it is not clear more data and analysis needs to be done. The nature of business is iterative and incremental. It’s important to focus on dependent variables that have business impact; be a leader by focusing on business goals with high payout as well as dependent variables with high impact. Focus on the dependent variable first as independent variables can come later. It is also important to become column and row oriented; whether it is expressions of opinions for rows or rows as independent or dependent variables. Integrating other data sets into your data set is effective through joins and calculated columns. In data competitions adding data is very important to improve the overall results, which is equally effective for analytics projects in general.
Free variables does not necessarily cost any more money. For example using division, exponentiation, inverse, logs, and other non linear transformations is important. There are consulting situations where consultants get paid more to analyze multiple dependent variables. This is because it adds more value and insight to the customer or client, and it isn’t much more effort for the data engineer.
It is crucial to broaden our analysis with more dependent variables and deepen the analysis with more independent variables to reveal more relationships. The unit of analysis includes rows used in datasets is, for example, looking at purchases online individually or looking at transactions per day. Using analytics as the foundation for analysis and decision making (especially economic issues) ties in directly with broadening and deepening our analysis, thereby improving ROI and optimizing trade offs.
The five phases we used this summer are universally applicable. Overall business analytics has a universal framework of goal definition, data collection, data analysis, modeling, interpretation, and feedback allows us to be analytics leaders in any organization. Further analysis and research is needed to bridge the gap to the future in terms of how to optimize the processes throughout the phases and checklists aside from the aforementioned strategies above.