Data Driven Decision Making

Not long ago we discussed the Value of Meaningful Data Collection. When there are clear rules for the values being entered in any given attribute, that information gains clarity and therefore its potential value also increases. That clarity and value takes on added importance for any company striving to make data driven decisions.

Gone are the days where changes, whether to an individual business process or to a company’s entire roadmap, happen based simply on an idea. Now that concept needs to be backed by solid data that shows the directional change will result in improved results for one or more Key Performance Indicators (KPI) . But what do you do if you can’t trust the quality of your data? Or maybe the data you currently collect doesn’t result in any metrics that directly relate to your company goals. What good is all that data you’re collecting if you can’t act on it?

If this sounds like your company, it may be time to take a serious look at what and how data is being collected. Is your data transformable into actionable metric reporting?

The “What” in Data Collection

It’s always interesting to see how a company has designed their change order form. When looking at their historical changes, many attributes are consistently left blank, incomplete or incorrect. What’s their purpose? Generally, all data collected during any business process should serve one or more of the following three purposes:

Definition – Attributes can provide details that define the specification of a part, a change being made, a quality incident, or any PLM record. This includes any information required by approvers to sign off, such as cost, planning, safety, regulatory or ESG concerns.

Actions/Tasks – Attributes can record actions taken, such as material containment already performed for a particular nonconformance. Attributes can also provide instructions or requirements for actions yet to be taken, such as defining a change implementation plan prior to releasing an ECO.

Reporting Capabilities – Attributes can categorize and summarize PLM records in order to provide meaningful feedback, like tracking quality trends and KPIs.

If the data you‘re collecting doesn’t serve one of these primary purposes, take a deep dive into why you’re collecting it and what its usefulness can be to your company. You might find that you can trim away attributes that don’t contribute to your processes or reports.

The “How” in Data Collection

If the information you need for reporting is buried inside free text attributes, it can be a struggle to pull out the relevant data. Converting text to list attributes whenever possible will help alleviate this problem. If the list values don’t provide enough data for the business process, a companion text attribute can be used for supplemental information.

When I see a text attribute in a customer’s PLM, and suggest it be converted to a list or multi-select list, I’m often told that there are too many different values to try to maintain a list. This doesn’t need to be an obstacle. In most PLM systems, adding a value to a list takes almost no time at all, and should not hold up the processing of any records.

More importantly, if the data is too varied or inconsistent, it can usually be organized into categories that would provide strategic information at a higher level.

Transforming Text to List Values for Actionable Metrics

There are many methods to harmonize inconsistent text data into stable list values. Here’s one of my favorites.

Try taking the historical data and lumping them by similarities into 5-10 groupings. This can actually be a fun team or cross-functional team exercise. Take a sampling of the text values and put each one on their own sticky note. Then, on a white board, have several team members move them around into groupings where the answers seem similar or related. Some stickies may get moved around several times until they find a home.

Now that you have your groupings, find a name that categorizes each one. This may take some workshopping, and it’s particularly helpful to keep in mind which KPI(s) this attribute might contribute toward. Once you have those category names, you’re ready to create a list attribute that will contribute far more reporting value than the original text attribute.

One of the strengths of this hands-on method is that it not only fosters positive synergy within a team, but it also gives them stakes in the result. Being involved in the process of change encourages employees to adopt that change readily and with commitment. Few programmatic or brute force methods to harmonize your data will give you these additional benefits.

Finally, for any attribute, new or old, it’s critical to have clear guidance to help the users enter that data properly. See my previous post, the Value of Meaningful Data Collection for more details.

Ready, Report, Act

After you’ve reviewed the types of data you’re collecting, you’re ready to refine your reports. You’ve set the stage by creating a PLM environment that supports metric reporting. And what about that brilliant idea you had for a process improvement? You now have the tools to create reports to support that idea, or to point you in a different, possibly surprising direction. Either way, your decision will be driven by real data. And that’s a decision you can trust.