Implementing an S&OP System? Here’s How to Make Your Data Work for You

A look at some of the S&OP implementation best practices for managing your data.

Now I’m not going to say, “Ask not what your data can do for you, but ask what you can do for your data.” In the past, I was under the impression that going through a system implementation process of setting up product hierarchies would mark the end of the process.

I must now say, there’s more to the story.

For the S&OP system to work as expected, there is the need to make your data work for you. This process requires a careful examination of the types of data that’s essential for visibility, isolates demand variability and ultimately give your product velocity.

Types of Data to Examine When Implementing an S&OP System

Enterprise Data

This is the data that’s essential to run your business. It’s the stuff consultants spend dark hours making sure it’s normalized and structured in a hierarchy for financial reporting and logical for the supply chain. Assume it’s all in working order when the last software consultant leaves the building. Now it’s all up to you to maintain.

As Ronald Ireland said in his book “Supply Chain Collaboration,” “My experience has shown that most companies vastly underestimate the number of errors that occur in execution systems.” This begs the question, why not include defect prevention in the data maintenance plan?

What are data defects? These would be the missing or incorrect data entries causing those in the supply chain to correct them before execution can begin. This slows down material velocity and requires extra resources to figure out the mess. (And please, stop rewarding people for fire-fighting. Could they be the arsonists?)

Instead, create alerts to scan and detect these defects before they cause problems. This can be as simple as missing characters in a sales code to create an alert or code in conditional logic to alert planners when a set of illogical conditions exist. The more you tailor this to the workings of your business, the more valuable these alerts become. Think of the time you save in more accurate orders that process quickly or better demand plans that help reduce unwanted inventory and safety stocks. Companies exist to do this for you, but the more you become involved, the more you save and increase material velocity through making defects visible.

Planning Data

The ability to think in terms of what causes disruptions or loss of material velocity helps identify the defects. Here are some examples, and definitely not all:

Assigning production to the wrong asset, inaccurate processing times, staging out of order
Assigning Ship to Customers to the wrong warehouse
Inaccurate lead times, or missing lead times
Duplicate entries of Ship to Customers
No assigned replenishment type, or incorrect replenishment type
Not re-aligning customer demand to the proper warehouse or item when substitutions occur

Again, designing a process to periodically scan for these defects saves a lot of headache in untangling these errors as they snake through the system. Think of all those who claim forecasts are always wrong. Are they doing their fair share to remove the defects? Could they be arsonists in the making?

Descriptive Statistics or Create your Own Data

This is where you create data from existing data in the system and attach it to a meaningful element. Create tables with this information to assist in the speed of description. Here are some universal examples to start your thinking towards your applications:

For views of inventory, the data element Item-Location (SKU) becomes your meaningful element. Take a 12-month data stream and measure standard deviation and mean. Divide the standard deviation by the mean and call it COV (coefficient of variation). Segregate your SKUs by a hurdle rate for your business, say 0.30. If the variability of the SKU is greater than 30% of the mean, call it HIGH VARIBILITY. Now you have a descriptive statistic for that SKU. Get fancy and call those SKUs below a COV of 0.30 LOW VARIABILTY. Create a hurdle rate for volume and you’ll soon have a four-quadrant system showing: High Variability/High Volume; High Variability/Low Volume; Low Variability/ Low Volume; and Low Variability/High Volume.
Create conditional logic for stock outs and alerts. (If Make to Stock, and Inventory =0, then it’s Stocked Out.) Compare to the volume variability grid to determine the importance to the planner. Add open orders and in-transit inventory with conditional logic to create a hierarchy of stock outs.
Create a table of strategic customers for each SKU so planners know which SKUs are more important when defects occur.
Count the number of months out of 12 the product ships for each SKU. Then divide this number into 12 to create the Average Demand Interval. 12/12 =1, so these SKUs show the highest velocity. 12/4 = 3, so this SKU moves every 3 months. Good to know when the customer complains.

Data Mining

According to the APICS dictionary, data mining can be defined as, “the process of studying data to search for previously unknown relationships. This knowledge is then applied to achieving specific business goals.”

Here’s where you tailor descriptive statistics to your business through downloading data from the system of record and engage in “torture”. Excel Pivot Tables offer a great tool for discovery and by pivoting a pivot table, you come close to what SQL developers can do for you. Once designed, leave the execution and automation to the developers and more capable tools.

Note: As we progress in this series of memoirs, we’ll define more of these descriptive statistics to improve your decision making by making them visible in terms of volume and variability that ultimately increases your material velocity.

Enjoyed this post? Subscribe or follow Arkieva on Linkedin, Twitter, and Facebook for blog updates.