You may have collected a variety of programmatic indicators and outcome targets, but at the end of the day, it’s the tasks carried out by frontline workers that will drive the impact of your program. From a program management perspective, it’s critical to define performance metrics for the people who are actually using the system. In this post, we’ll walk through some key things to consider when crafting performance metrics for a mobile data collection system.
A supervisor reviews performance metrics with a group of community health workers
The purpose of performance metrics
While a program may have hundreds of indicators overall, it can be useful to focus on only one or two key metrics for performance management. This allows expectations to be clear for users and can focus a team on strategies to amplify performance.
Not all metrics are equal when it comes to performance management, so it’s worth some extra effort to carefully consider your focal metrics. Some questions to consider:
- Is the metric something that the frontline worker can actually control? These metrics are also supposed to serve as incentives to improve performance. That means that if the frontline workers have no control over the outcome, hearing about it every time they check in with their supervisor can be demotivating.
- What is the linkage between this indicator and your desired project outcomes? We’re not saying to ignore them if the link is weak, but your frontline workers will be more motivated if they are shown how their work has a direct impact on overall outcomes.
- How will you respond to differing performance against this metric? Depending on the data you collect, you should have a plan to respond: Offer more tools and resources to struggling workers or opportunities to coach others for those going above and beyond.
- Is the metric susceptible to gaming? You probably want to encourage strong performance against your metric, but if the incentives are strong, people may find creative ways to meet the letter of the metric, but not the spirit. This is illustrated by the classic example of “curbstoning,” where enumerators are incentivized to complete the largest number of surveys and end up just sitting on the curb and filling them out themselves.
After you have identified some candidate performance metrics, it’s time to see how these map to the data you are collecting.
Sample performance metrics
One good place to start when looking at performance metrics is considering what metadata you have access to. For instance, in CommCare, metadata is useful because it is common across all apps. Some key pieces of data you have access to form start and end times, case updates, number of open and closed cases, and even information like GPS coordinates.
From these few basic parameters, you have a wide range of information that you can link to performance indicators. In fact, we built CommCare’s Worker Performance Reports around these metadata components.
A sample worker performance report for supervisors to review workers’ progress over time
For an example of how these are used, imagine your key metric is the number of houses visited in a month. In CommCare, this would translate almost directly to the number of cases that a worker updated in a month. This type of indicator is readily available in CommCare’s reports, and can be used to inform supervision in near real time.
Other metrics that can often be derived from metadata include the total number of form submissions (i.e. how many home visits were conducted), time spent filling out a form (i.e. duration of a counseling session), and total cases (i.e. number of clients registered).
Explore what data your tool readily offers, and you might find that by combining these data in a certain way, you can learn more about how your team is performing. If not, you can always explore what other data your program is collecting…
Sometimes metrics may be too nuanced to be captured by standard application data. For example, maybe you are interested in how many community members are attending group meetings facilitated by a frontline worker. In this case, you could check to see whether these data are being recorded for another purpose or look at opportunities to integrate a relevant question into the form itself.
In CommCare (and some other platforms), you can create a custom report to include these program-specific metrics on a frontline worker-level, or aggregated across a region.
Frontline workers vote on the metrics that motivate them the most
Using metrics for performance management
It’s key to test out how you plan to monitor and respond to the metric before launching. Sometimes a simple question does not translate cleanly into the desired metric our outcome indicator. For example, a simple question like “was this child immunized?” is likely much more complex than it seems on the surface, and it’s important to think through all of the possible interpretations and edge cases.
On a tactical level, once supervisors have access to their workers’ performance metrics, they can make informed decisions about how to spend their own time: Seeing certain individuals struggle to meet their targets can flag where valuable supervision hours and resources should be allocated. Additionally, high performers provide opportunities to collect insights and learn from their success.
Program-level access to performance data over time also enables testing of different strategies to boost performance and possible changes to the program’s execution. Different approaches to motivating workers may yield data that can be compared to a program-wide baseline, so supervisors can continue to iterate on their approach.
Remember, everyone is different, so exploring the motivations and environmental factors that affect your team’s ability to perform effectively will set you up well to improve the performance of your program over time.
For your consideration…
In one study, community health workers using CommCare were provided feedback about their key performance metric: visiting each of their pregnant clients every month. This feedback included both their progress against this metric and how their progress compared to the progress of their peers.
During a Randomized Control Trial, the community health workers who used the feedback system made 21.5% more home visits than those who did not. Qualitative feedback suggests that many of them found it very motivating to see how much work remained to do, and they liked having a sense of how well they were doing compared to their other CHWs. This dramatic performance boost is notable in that just giving workers their data is very inexpensive compared to other more intensive types of supervision.
So, if you’re worried about the time and investment of tailoring a supervisory approach to individual workers based on their performance, rest easy. Sometimes, all you have to do is share the data and workers manage themselves.
Sharing the data
Most reporting is developed with an assumed audience of program management staff and supervisors. As we have noted, these data can be critical to informing various programmatic decisions – both in the short- and long-term. Additionally, the case study above shows an example of where closing the feedback loop back to users can amplify performance.
However, even the beneficiaries of a program can benefit from access to their own metrics. Many reporting systems are very hierarchical: The frontline worker enters data into a system, and it makes its way up the chain into increasingly aggregated reports. Rarely do programs consider sharing that data back with the source, but providing data on key performance metrics or outcome indicators back to the community can also be motivating and demonstrate accountability to the people that are served by the program.
Go find your worker performance metrics
Digital systems present a unique opportunity for more dynamic program management through monitoring performance metrics. The data they collect can aid supervisors, improve resourcing decisions, and support and empower the workers themselves.
If you are still curious about how to determine the right metrics for your own program, check out this article on identifying your data requirements.