Evolutionary Data Quality: A Response to AI Disruption

Written by Emilio Arocho on September 18, 2018

Artificial intelligence is so embedded into our everyday lives we don’t notice it anymore. We don’t see machine learning as end users, we see its major outcome: convenience. Shortcuts. Streamlined digital experiences that require less attention from us.

Similarly, when we think about how artificial intelligence relates to the association space, it’s natural to focus on outcomes. We see disruption to our service models, risk to be managed, a race to find highly relevant applications for the technology before our competitors do.

Today, let’s focus on a major opportunity that comes along with AI disruption.

There’s something associations can be doing right now to not only make the best of AI disruption, but also reap benefits from it in the future. And it doesn’t cost thousands of dollars or require the adoption of a new product. In fact, it’s probably something your organization already cares about deeply: managing your data quality.

Adopting an evolutionary view of your data quality not only helps your organization in the present, it engineers future opportunity. One common application for AI is to make insightful connections between internal and external data sets. This is not an easy or automatic process, but you’re far better prepared for it with high quality data that reports impactful, contemporary truth about your organization, its programs and its members.

What makes for good quality data? One of the most important features of a quality data set is flexibility. As the rate of change at our organizations increases, static database fields and option sets are less acceptable than ever before.

DATA SHOULD BE FLEXIBLE

If you can be flexible, and seize opportunities to integrate new data on the fly without undermining historical records, your organization has enhanced ability to facilitate its evolution.

These need not be sweeping changes, either. You may decide to incorporate a new data point and implement it in your database on a trial basis at first, surveying a subset of your community. Then, once you have some data in your system natively, you can analyze it in concert with all the other data and decide if it’s something you’d like to collect more widely and permanently.

DATA SHOULD BE DOCUMENTED

By this I mean more than database administration procedures, but true data provenance.

Where does each data point in your system come from? Are there any biases in it? Are there wrinkles in your data collection that your organization recognizes and works around? Do you maintain records for when your process of collecting data changes?

Don’t underestimate the value of documenting these insights for future reference and historical record. This allows the evolutionary character of your data to be better reflected in your planning and strategic thinking.

When your association undertakes a new project or considers undertaking a new project, how long does it take before your data comes to bear some level of influence?

DATA SHOULD BE PART OF EVERY CONVERSATION, EVEN WHEN IT’S NOT

Your data deserves a seat at the table when your organization is planning and managing.

However, it’s okay if you don’t currently have data to inform a conversation. That itself is a data point! What matters is that this fact gets noticed and evaluated somewhere in your planning and execution process.

You will find that when you think in evolutionary terms about your data, it naturally becomes a part of more of your conversations. This is a wonderful thing!

In closing, evolutionary data quality is not an achievement or destination. Instead, it’s an ever-present process of assessing how your association collects, documents and uses information. And it sets you up to be able to move quickly and capitalize on that data, whether via day-to-day operations or feeding your data through a machine learning algorithm. Stack the deck in your favor!