Make your data collection match your impact KPI 

A simple rule of thumb helps you choosing where to start: If your KPI is about access, the data may already exist; you just need to count it. If it’s experience, you generally can’t observe it directly, so you ask your stakeholders, either through a short survey or a quick feedback question. If it’s about change, comparison is the starting point: collecting the same data at more than one point in time can show a direction or trend.

A good idea is to start with existing records, as building from there tends to be more powerful than designing a comprehensive system from scratch. In our bootcamps, we help you put this into practice:

By (1) building a flexible data collection plan around your existing KPIs, (2) identifying the right customer touchpoints to collect your data from, and (3) helping you use what you already gather, to impact operational decisions, and not just for reporting.

Use the same data to answer a more meaningful question

Even once you’ve matched a method to your KPI, there’s often another shift available before collecting anything new: you can look at your available data through a different lens.

Take an impact start-up that already logs customer registrations and IDs (which is a typical approach). Read through an output lens, that log answers a simple question: how many new customers were acquired? Read through an outcome lens, the same log can answer a more meaningful one: What stakeholder demographic gained access, and where did they come from (e.g. female users from rural areas)? That’s the important shift from a counting exercise to understanding who benefits from your product or service.

Even if your log doesn’t yet capture basic stakeholder characteristics like gender or location, that information is usually close at hand. It’s a small addition to your registration process rather than a separate data collection effort.

If a large share of your stakeholders turns out to come from rural areas, for instance, that alone can be an early signal about where the service is proving most relevant. None of this requires new infrastructure, just a different question asked of pre-existing data.

When you can’t observe the outcome directly, look for a proxy 

Not every outcome can be measured directly, which is often where one might default to a survey. There’s usually an easier route: measuring something observable that serves as a credible stand-in for the outcome itself.

For example: A solar company cannot verify a household’s full electricity bill. But it can verify, from its own system, how much grid electricity a home no longer requires. This means that the solar company can use avoided expenditure as a proxy for financial relief. The test for a good proxy is whether you have credible evidence linking it to the outcome that you want to demonstrate. 

Ask fewer and better-timed questions

There is often a temptation to collect everything that might be useful one day. We get it: More survey fields, more follow-up questions – a fuller picture can all feel worth having. However, this is a temptation that’s worth resisting if you strive to make collecting impact data lean and meaningful. A useful rule of thumb: if you don’t already know what you will do with a piece of data, don’t collect it. Larger datasets aren’t inherently more rigorous. They’re simply harder to analyze, and a bigger ask of the stakeholder providing the information.

This has a real cost. For example, a report by Root Capital on designing impact evaluations in a way that they create value for the stakeholders involved showed what happens when stakeholders are asked the same battery of questions again and again: Some participants became visibly frustrated; others grew suspicious of who was asking and why, and some simply gave whatever answer would end the conversation fast. Fatigue can quietly compromise the data you collect, in addition to limiting the number of responses you can collect.

To fix survey fatigue is less about clever design and more about restraint. You must ask only what your KPI need and attach it to a moment that is already happening. This can be an onboarding step, a renewal, or a support call. This is way better than creating a new interaction.

“In our bootcamps, we’re not trying to set up a new research department inside a social enterprise. What we’ve seen is that the simplest data collection methods are usually the ones that actually survive contact with a busy team and real customers. That’s the balance we care about during all sessions: rigor without losing simplicity.” 

Louisa Wolf, Trainer & Senior Associate, Impact Management at Roots of Impact