Erin Shellman

Improving data capture and applications in the field.

Values and operating principles

Before proposing solutions for people working in challenging field conditions, it is critical to define core values and operating philosophies so that we can reason about which solutions align with our values and which do not.

Principles

The following are examples of operating principles:

  1. Automate as much data capture as possible. Humans are bad at data entry, and computers are great at it. As much as possible, we should build workflows that play to our strengths and help us overcome our weaknesses.
  2. Tools work for people, people don’t work for tools. Too often, workers in the field are asked to use clunky and impractical tools to produce datasets that don’t improve their day-to-day. But tools are meant to ease our workloads, not make already busy people a means to an end. Further, when tools create friction, people work around them.
  3. Quick and effective service administration is more important than data collection. The first priority is the client experience and data considerations are secondary. Any solutions that improve data collection at the cost of service administration are non-starters.

Streamlining data entry

The following is a collection of ideas to ease data entry and improve data quality. The ideas comprise both technological and operational approaches.

  1. When possible, work in pairs. One person interacts with the client, while the other takes notes and does data entry.

  2. Conduct data entry on tablets or laptops. Digital forms with limited, pre-populated fields shrink the universe of options and reduce misspellings and duplication. Likewise, Type Ahead capabilities embedded in digital forms allow service administrators to autocomplete things like names when working with repeat clients. Digital forms can also automatically correct addresses and reduce entry errors.

  3. Voice to text to transcribe in-take conversations and enter data. Advancements in AI and language models make it possible to automatically transcribe spoken word into text and parse text into structured data. If service providers are using tools like tablets and laptops, it is possible to collect data both from traditional structured forms and from conversations.

  4. Online and in-person training. Often front-line workers don’t see the downstream application of the data they collect and so don’t know how it’s used and why data quality is important. Staff trainings should go beyond practical training to illustrate the big picture. Further, not all service providers are focused on the same goals, so training is an opportunity to build alignment towards a shared outcome.

  5. Multi-purpose in-take forms and processes. Hand-in-hand with training, re-using forms and processes across service areas can help standardize and simplify the data fields that are collected.

  6. Record Linkage and related algorithms to probabilistically identify duplicated records. There are many algorithms capable of using incomplete data to identify duplicate records.

Piloting in the field

Understanding current working conditions and building credibility with direct service providers is essential to the success of any of the above ideas. Significant time should be spent at the outset shadowing service providers to understand their realities and to build an intuition for what types of solutions will and won’t work. During this time, it is important to interview service providers to learn about their pain points and priorities.

Once the landscape and key players are understood, it’s critical to move quickly to pilot solutions. Pilot studies provide the opportunity to test interventions before investing significant time and money. They also give collaborators in the field a chance to give feedback and shape the solution. Only when a pilot has demonstrated success should we proceed with productionizing and/or scaling to other sites.

Finally, it’s important for the improved data quality to feedback to service providers and lighten their load. How that looks will depend on the specifics of the situation, but that could be with services that help them manage clients or appointments. Success of the intervention hinges on whether it creates direct value for the people in the field.