Data integration - the foundation stone for measuring portfolio development impact

International donors and investors need to monitor the progress of their individual investments and the alignment with their internal strategies in order to show their impact. To demonstrate overall impact, they need to have a harmonized portfolio level view of those individual investments in order to support future decision making. Through the monitoring work with our donor, we have developed Theories of Change and Action models to derive appropriate metrics to measure their portfolio impact. However, this requires integrating and collating data from multiple implementing partners (grantees) to harmonize to a set of standard metrics.
Working with different grantees, each with their own individual data systems, creates challenges, often the biggest being people and processes. We have developed processes to facilitate interoperability and integration of disparate data sets, part of working towards FAIR data (findability, accessibility, interoperability, and reusability). We have written code based on an adaptor design pattern to convert grantee data to a standard portfolio format and harmonized grantee data to standard terms (e.g. livestock disease names). Through close partnerships with
grantees, we have explained the significance of this work and worked together to map data to a standard portfolio format. Instead of requiring a common portfolio format, grantees submit data in a format that is specific to their organisation. This reduces the risk of data errors and saves grantees time in data preparation. As such, this is an equitable process that supports those grantees not having in-house data capacity and ultimately produces higher quality metrics.
Interactive dashboards are used to disseminate the data which can be disaggregated by country, or domain specific terms. Metrics are based on both measured and modelled results, modelling is used to populate metrics where it is either not cost-effective or practical to gather specific field data or if impact is expected to occur in the future, so requires foresight and prediction. For example, we model the net economic impact of specific animal vaccines or therapeutics used by small-scale livestock producers. Gaps in available data to parameterise models are filled by literature and/or elicitation of expert opinion. Uncertainty in the data populating the metrics is communicated through a color-coded scheme. Clearly, the developmental and scientific rationale for the data collection and analytics are fundamental for an understanding of the socio-economic context of the portfolio, for which data integration lays the foundation.