We’d like to share an illuminating anecdote that illustrates why it’s important for every employee to understand a company’s data strategy. The unlikely subject? Breakfast sausage.

A French food services company saw orders of breakfast sausage spike dramatically at one site until it became the only item ordered. Nothing was wrong with their data pipelines and it was unlikely that croissants had fallen out of style, so the team dug deeper. 

What they found was that the point-of-sale system allowed the worker to punch a single button for an item rather than enter the item’s full price. Breakfast sausages cost the same as croissants, and because the breakfast sausage button was more strategically located on the register … well, you can guess the rest.

We can all probably recall a time when we were involved in a transaction and an employee prioritized speed over data accuracy. Perhaps you signed up for a new mobile phone contract and the assistant was supposed to ask a myriad of questions beyond the information required to produce a legal contract. Instead, they sped through the optional fields with dummy values and got you out the door with a smile—and gave their data team a headache.

Data practitioners often see data as columns and rows, but the reality is that data is just as much a story of human action. As data professionals, if we don’t understand the incentives, environments, and processes for employees to record actions accurately, no amount of fancy JOINs or value imputations will save us from poor data quality.

Even with proactive monitoring to flag anomalies in data distribution, data trust has to start with  your people. In this post we’ll discuss strategies to turn your workers into data assets.

Understand the data collection process 

The first step in turning frontline workers from a data liability into a data asset is to develop a deeper understanding of the data collection process, including:

  • What data is currently being collected?
  • Why is this data being collected?
  • Who is collecting this data and at what point in the customer journey?
  • Who is responsible for the data quality at this stage? What is the feedback loop?

This will require reviewing business processes, evaluating the health of your data, and talking with those familiar with each collection process. You may want to consider structuring your team by domain (if they aren’t already) or establishing liaisons, as this coordination will likely be an ongoing process.

During the review, ask yourself who is the most likely to provide the best data inputs. That could be the frontline worker, the manager, the upstream business owner, or even the customer. 

Reduce field fatigue and align incentives

Armed with this new knowledge of the data collection process, you can now begin to fight field fatigue. As data professionals we are naturally curious and understand the power of data, so we have a tendency to want to collect and hoard it—even without an immediate use case. 

While we understand the value, we can sometimes be guilty of overlooking the cost. For example, our colleagues in marketing have always been acutely aware of the cost of each additional form field and its impact on the rate at which users complete or abandon their submission.

Form abandonment is one thing when we’re talking about a piece of marketing collateral, but when the data is being gathered in real time with the customer, either on the phone or on location, it becomes a central part of the customer experience. Each frontline employee now must weigh the direction to gather that data against their mandate to provide an exceptional customer experience.

While there are situations when it is difficult to balance the added value of the data with the cost it imposes, in most cases you will not be faced with such a devil’s bargain. 

You can trace the data lineage to see how data flows across the company, and in many cases you may find the end destination is not a customer-facing data application, but an unused table in the corner of your data warehouse.

Tracing your data lineage from Snowflake in a data observability tool from Monte Carlo. 

At the very least, understanding these data flows and how the data is ultimately used by your internal business consumer helps you better align expectations. Creating data SLAs can help by codifying aspects such as how frequently the data needs to be updated or by creating and placing it within different data reliability tiers

Show, don’t tell, why the data is so important

Another important way to align incentives with frontline data collectors is to create a full feedback loop and show why the data they are collecting is so important. 

Think of it as the carrot and the stick. For the carrot, you can illustrate the art of the possible with data literacy programs to show how data can lead to improvements that make their job easier. It could be the automation of a formerly manual, tedious task or optimizing workflows—like making sure the croissants are placed closer to the coffee since those two items are frequently ordered together.

The stick doesn’t have to be punitive, though. It can be as simple as letting people know that your team is aware when the data collection process goes awry, along with a request that they re-dedicate themselves to more accurate collection. 

The workers in our opening example didn’t start off punching the breakfast sausage button every time; they ramped up to that point after they didn’t receive any pushback. How important could that data be if no one complained when it was wrong? This is part of the reason why time to detection for data anomalies is so important.

This approach will be particularly effective and credible when combined with your efforts to reduce field fatigue and make frontline workers’ jobs a bit easier. 

Improve data trust by investing in a people-first data strategy

Investing in data literacy and the proper incentives to maintain it can involve a considerable amount of resources, but the cost and consequences of poor data (and untrustworthy data) are even higher. 

If you don’t treat data like a human problem, teams can find themselves burning resources on a hedonic treadmill of data quality initiatives that attempt to fix the symptom without focusing on the root cause at the source. They can also lose the trust of the organization in the data product they produce, which once lost can be difficult to regain.

Our recommendation is to invest in a people-first data strategy and turn your potential data liabilities into assets. You can thank us later. By the way, is anyone else as hungry as we are?