00:00:01 Introduction to Grow Great and the data challenge

It is an honor to be with you today to share a little bit about Grow Great's experience using ActivityInfo to move from a fragmented data system to a streamlined data management process. We will show how this has improved our data turnaround time, our responsiveness to funders and beneficiaries, and our overall performance.

I will share why we needed a unified data approach before handing over to Firas, who will take us through some live examples from Grow Great's ActivityInfo databases. We will end by sharing some key takeaways from our experience that we hope will be relevant for you, before leaving some time at the end for questions.

00:00:43 The context of stunting in South Africa and Grow Great's programs

Let me start by sharing a little bit about the South African context and what Grow Great does. South Africa has the unfortunate reputation of being one of the most unequal countries in the world; two-thirds of our population live in poverty and youth unemployment sits above 60%. In this context, one in four South African children are stunted. Stunting is when a child is too short for their age due to poor nutrition during pregnancy and the first two years of life.

When children are chronically undernourished, there is a lifelong impact on learning, health, and earning potential. In our recent National Thrive by Five Index, which tracks learning outcomes for children under five, stunting was significantly associated with poorer outcomes, leaving children on average five months behind their peers. Stunting is especially prevalent in communities facing poverty, inequality, and isolation. It is both a symptom and a preventable consequence of South Africa's deep inequality.

The Grow Great campaign began in 2018 with one goal: to halve stunting in South Africa by 2030. We focus on the first 1,000 days, from conception to the age of two. Our work spans three interconnected pillars. The first is community mobilization and advocacy, aiming to bring stunting onto the national agenda. Our second pillar, Grow Great Champions, trains and mentors community health workers to prevent malnutrition by supporting pregnant mothers and conducting growth monitoring. Our third pillar is Grow Great Flourish, which supports pregnant women and new mothers through peer-led mom and baby classes.

To date, we are active in 20 districts across all nine provinces. We have trained over 4,000 community health workers reaching over 420,000 children. We have 100 active Flourish hosts supporting 59,000 women. Emerging evidence shows that children visited by our trained workers have significantly lower levels of stunting, with more than 50% of those stunted recovering by two years of age.

00:05:00 The challenges of a fragmented data system

Grow Great is a complex systems-level campaign requiring a responsive Monitoring, Evaluation, and Learning (MEL) system. However, initially, our activities involved multiple bootstrap solutions: Google Forms, paper-based surveys, Excel sheets, and WhatsApp extracts. These were often reactive activities driven by funder demands. Without coordination, this led to significant fragmentation, confusion, and frustration.

We identified several common pitfalls. First, fragmentation had real implications for data quality and rigor. For example, we used to capture attendance registers via WhatsApp messages manually entered into different Excel spreadsheets, resulting in 96 different tabs of data in just one year. Second, data silos made holistic organizational summaries difficult. Simple issues, like using different district names across datasets, made it hard to match data and allocate resources effectively.

Third, data security and sustainability were at risk. When data is stored in multiple places or personal drives, it is difficult to manage access when staff leave. As we hold personal information for vulnerable groups, we must meet strict data protection criteria. Finally, multiple tools increased our turnaround time on data use, slowing access to insights and making reporting difficult.

00:08:33 A strategic approach to system change

Recognizing the need for change, we decided to tackle this by changing our approach and then our systems. Our approach evolved from funder-driven requests to embedding a culture of progressive learning. We need our data to answer three questions: Did we do what we said we would do? Did it make a difference? And what are we going to do about it?

We started in 2022 with a full situation analysis using a MEL gap analysis tool. We scoped out our current data flows and streamlined them into a list of potential databases. We then clarified our frameworks, indicators, and definitions. These steps helped teams reflect on the data pipeline and created internal buy-in for change. It also helped us map out our data model before seeking a software solution.

We chose ActivityInfo because it provided an end-to-end workflow for collection, cleaning, storage, and analysis. It also offered strong data security, including role-based permissions, and the team provided strong onboarding support.

00:12:16 Designing the solution in ActivityInfo

To understand the solution, we must define our terms. A system is any process that is defined, repeatable, and consistent. Information Management is the practice of using that system to provide information. Systems can range from manual to digital fragmented, to digital integrated. Grow Great had a good system, but it was fragmented. Our goal was to make it smoother, faster, and connected.

ActivityInfo's hierarchy starts with the database, broken down into folders, forms (tables), fields (columns), and records (rows). We used this structure to tackle Grow Great's challenges.

00:17:02 Overcoming data fragmentation and ensuring data quality

The first challenge was data fragmentation. The solution was a centralized relational data model. We designed a single source of truth for all common reference data. The main database holds reference lists for provinces, districts, sub-districts, and clinics. This feeds clean standards down to all other program databases, allowing us to aggregate results later.

The second challenge was data quality. We built smart relational forms with embedded controls like validation rules, required fields, and relevance rules. For example, sub-districts are not open text fields but reference fields pulling from the master list. We use specific field types like date fields for birth dates and calculated fields for age to prevent errors. Relevance rules ensure questions only appear when necessary. While errors can still happen, these controls catch many issues at the point of entry.

00:21:00 Streamlining reporting and visualizing insights

The third challenge was slow turnaround time and reporting. Unified live reporting allows us to create reports across different program databases and consolidate historical data with new live data.

In the live demo, we addressed three practical questions. First, to count community health workers and Flourish hosts by province, we used the "Host Information Master" form. We created a pivot table to count records aggregated by province and district, filtering out inactive hosts. We then added a calculated measure to count community health workers from a separate database. Because both forms use the same master geographic lists, the report automatically groups them together.

Second, to track the total number of moms year over year, we had to combine live 2025 data with historical data. Instead of migrating all old records, Grow Great imported summaries into a "Process Indicators Summary Table." We then used a calculated table to combine these historical summaries with the live monthly validation data.

Third, to compare progress against targets, we included targets in the summary table. This allows us to visualize performance using traffic light coloring (Red, Amber, Green) to see if targets are met. These reports can be presented in a notebook format, which includes headings, text, pivot tables, and maps, telling a complete story of the data.

00:34:59 Data security and role-based permissions

None of this matters if the system is not secure. ActivityInfo is ISO 27001 certified. We utilize multi-layer security, including Single Sign-On (SSO), secure data centers with redundancies, and encryption in transit and at rest.

Crucially, we use a fine-grained roles and permissions system. We can decide who can view, edit, approve, or delete data at both the record and field levels. This ensures that "least privileged access" is a practical reality, protecting beneficiary data.

00:36:45 Poll results: Common challenges in M&E

Looking at the poll results, 47% of attendees identified data silos as their main challenge, which is the classic digital but fragmented problem. Poor data quality and slow reporting were also significant issues. Interestingly, security was also highlighted, which is a critical challenge often overlooked until it is too late.

00:39:36 Key takeaways and best practices for implementation

Reflecting on our journey, we have five key takeaways. First, don't confuse movement with progress. Fragmented systems require immense energy to maintain. Pausing to reflect and sharpen your skills is a worthwhile investment.

Second, fragmentation is solvable but requires a clear, flexible plan. We mapped our systems to identify gaps and motivated for additional capacity; our M&E team grew from two to five people. Co-creating the roadmap with teams helped foster ownership.

Third, do not let perfection stop you. A perfect data model does not exist; treat it as a permanent draft. We updated our model multiple times as we learned, but we didn't deviate too much from existing workflows to ensure adoption.

Fourth, look for quick wins. For example, consolidating our Flourish baseline surveys allowed for immediate longitudinal analysis. Migrating our paper-based assessments reduced turnaround time from six months to two weeks. Sharing basic dashboards early on helped embed the system into the culture.

Finally, good data systems are a strategic organizational function, not just a "nice to have." They build efficiency and trust. Cutting M&E capacity during funding cuts only undermines organizational stability. We need to streamline collection so we can spend more time using data to drive social change.

00:49:39 Q&A session

Do you have a relational database where you store your data? Currently, everything is done through ActivityInfo as an end-to-end solution. However, ActivityInfo can act as a hub and connect via API to Excel, Power BI, and other software. Before migrating, we consolidated data into Excel databases to clean up fields, which made the migration simpler.

Does the system allow for collecting qualitative data? Yes, ActivityInfo allows for multi-line text fields for narratives. While we haven't built specific code for analysis within the platform yet, we store and manage qualitative data alongside quantitative data, ensuring security and easy referencing.

How do you manage the correction of wrong data once it is in the system? The system helps identify errors during import via validation rules. Once data is in, we use functions like scanning for duplicates to identify, compare, and merge or delete records. The system also has built-in redundancies, allowing us to retrace steps and recover data if mistakes are made.

How can data be used for future planning? Having data accessible means that during strategic planning or reporting, you know exactly where you stand regarding reach and implementation. It also speaks to an organizational culture of wanting to use data. We are trying to embed routine data reviews to build organizational resilience.

How do you combine previous data with new datasets? For some data, like active hosts, we migrated the records. For historical data used only for comparison, we used calculated tables. This involves selecting columns from the historical summary table and the live data table, transforming the structure to match, and combining them using a UNION function.

Does the system allow for data quality checks like completeness and cross-validation? Yes, validation rules are central to the system. For example, you can prevent entering a date of birth that would make a child older than the program limit. While you cannot eliminate all errors, these rules significantly reduce them.