One of my former sales directors once told me that despite having eleven different dashboards, she was unable to determine if the discount campaign from the previous quarter was profitable. She had all the necessary data; however, no one had tied that data back to her question.
The gap exists everywhere. Companies gather much more data than they use and can't figure out why their reporting is so expensive and has no value.
Start from the decision, not the data
The practical form of data analytics begins when someone makes a decision on a regular basis. Such as which leads should be called first. How much inventory to order again. Which customers will receive the retention call. And which marketing expenditures will remain after the next budget review.
Choose one. Describe how it is currently produced, the information the individual(s) utilize, and how frequently those individuals make mistakes. Only after you have described the current process, should you seek out data.
Teams that begin with an audit of everything they have gathered will develop a warehouse full of data but no answers.
Where it actually moves the numbers
Most businesses using business data analytics effectively will demonstrate some common trends in how they operate.
- Customer behavior. Who buys again? Who goes quiet? What were the quiet ones doing in the 8 weeks prior to going quiet? That is often enough to pay for the entire analytics effort.
- Pricing and margins. Companies already typically have revenue by product; however, this is not the case for margin by product after accounting for discounts, returns, and cost to service a customer. Margin by product rankings are generally quite different than those that the sales team may believe exist.
- Operational Waste. Where orders stop moving forward, where we repeat steps in our processes, and how much time we lose due to mistakes. While operational waste can seem less exciting than other types of waste, it often has the fastest ROI (Return on Investment).
- Forecasting is about making predictions on how much product you'll sell (demand), how much money you'll have (cash) and how many employees you'll need (staffing). A rough model based on the past two years of data will most likely be more accurate than a simple trendline in a spreadsheet, especially when it comes to predicting seasonal trends.
The part that gets skipped
Before most data analysis projects are even started, they're already dead on arrival. The reason for this is because the data used for the project is stored in four different systems that don't communicate with each other. For example, finance tracks a customer based upon when the customer received their first invoice; sales tracks customers based upon when the customer first contacted the company; and support tracks customers based upon when the customer created their account. There are three teams, each with a different number representing the same customer, all sitting in a room together and having an un-productive discussion.
Defining what things are is boring. However, there is a big difference between people trusting the results from your business analytics programs and people simply ignoring the dashboards you create.
Prior to beginning to create something, establish in writing what defines your customers; when is a purchase considered valid, and how do you handle returns. Once you have established these criteria, then you can link together the software programs that provide the information required by each criterion.
Dashboards aren’t the deliverable
A dashboard that no one is viewing on Tuesday was unsuccessful, even if it looks great. Business Intelligence and Analytics that are successful will be integrated into the workflow. For example, a "flag" in the CRM that indicates a drop in usage for a customer account; a "reorder" suggestion in the purchase screen; or an alert when a profit margin falls below the acceptable limit.
The decision-making process will remain unaffected unless the insight is provided at the time of making the decision.
What good providers do differently
Companies offering data analytics services differ greatly from each other. Companies that are good to work with will inquire about the decisions you need to make prior to asking for your company's data, will let you know if the question you asked cannot be answered based on the information you have available, and will provide a product that can be maintained by your employees after they leave.
Ask to see their data analytics solutions that have been in continuous operation for at least two years. Then ask if those same solutions are being operated today. If a vendor gives you a dashboard and then walks away, he's given you a picture of your dashboard.
Prove it on one decision
Take the decision you made when we began. For one full quarter, use both methods of analyzing the same data. Be completely honest in comparing the results, including the times that the data was incorrect.
If the new method is superior to the previous method, then continue using the new method. If the new method is not superior to the previous method, then you have found something with very little cost, and you are free to find another area to research.
A discount campaign won’t have its effectiveness evaluated by 11 dashboards; however, a single question asked and answered will.
