ECD Connect: What Usage Data Reveals About Practitioner Adoption
ECD Connect: What Usage Data Reveals About Practitioner Adoption
Context: The problem this platform is trying to solve
Early childhood education (ECD) in South Africa is under-resourced and fragmented. Most of the sector’s over 40,000 preschools operate informally, with no digital records and no way to systematically track whether children are attending or developing well.
ECD Connect is a live digital platform giving ECD practitioners tools to track attendance, monitor child development, manage finances, and access teaching resources. It currently serves 3,000+ registered users across South Africa.
This analysis examines what the platform’s usage data reveals about how, and whether, practitioners actually adopt digital tools in a low-resource context. Where do they drop off? Which features drive sustained use? And what does that tell us about how to design and prioritise features for this population?
Why this matters beyond ECD
Digital adoption patterns in the ECD sector mirror broader challenges in public-sector and emerging-market tech: how do you design for users who are time-poor, cautious about new technology, and operating without reliable infrastructure? The signals here have direct relevance to any product targeting underserved communities at scale.
Platform scale
Data covers ECD Connect App (open access) and ECD Connect Partner (white-label) tenants. ECD Connect Partner launched in October 2024 and ECD Connect App launched in April 2025. All data covers from launch to 20 May 2026.
3,319
Registered users
1,388
Converted users (valid preschool)
981
Preschools added
4,151
Children registered
Conversion rate: 42% of registered users complete onboarding to the point of a valid preschool connection. This analysis examines what drives conversion, which features sustain engagement, and what the data reveals about how practitioners actually adopt digital tools in this context.
Part 1: The onboarding funnel
How we define the funnel
To get meaningful value from ECD Connect, a practitioner must complete a seven-step activation sequence: registering, adding profile details, connecting to a preschool, and ultimately taking attendance. Each step requires deliberate action and represents a commitment to using the platform.
Note that a small subset of users engage regularly without completing the full funnel, likely using the app for resources or lesson planning rather than administrative tracking. The funnel below focuses on the core activation pathway.
The funnel below shows cumulative completion rates (OA and WL tenants).
Reading the funnel
The funnel reveals a two-stage dropout pattern that shapes the entire product strategy:
Stage 1: Identity friction (Steps 1-2). 35% of users never complete basic profile steps. This likely reflects a mix of users who don’t have documentation available at the time of signing up, non-ECD-practitioners who joined to explore the platform and people who joined out of curiosity after seeing the platform mentioned in their networks. User support data suggests a meaningful subset are prospective practitioners looking for information rather than active users ready to onboard.
Stage 2: Structural barriers (Steps 3-5). Connecting to a preschool and adding children requires institutional access. A practitioner who wants to use the app but isn’t the registered principal is effectively blocked. In addition, practitioners joining mid-year at larger schools also flag child registration as a friction point: adding a full class roster manually is a significant upfront time investment.
69% of users were last seen on the same day they registered, showing that the funnel doesn’t lose people gradually, it loses them immediately.
Product signal
Users who reach Step 3 convert at significantly higher rates. The biggest lever for funnel completion is shortening the gap between registration and first genuine usefulness, even in a limited trial mode.
Design implication
Steps 3 and 4 occasionally invert: some users create a class before formally connecting to a preschool. The data suggests some users naturally explore class creation before completing preschool connection. Making this pathway more visible in the onboarding flow could encourage more users to reach that point rather than dropping off earlier.
Part 2: Feature engagement
Being “converted” (having a valid preschool connection) is a proxy for meaningful adoption. The analysis below compares feature engagement rates between all registered users and this converted cohort, revealing which features are true retention drivers versus which are surface-level signals.
Reading the feature data
The gap between grey (all users) and green (converted users) bars is the key signal. Features where the converted bar is much larger than the all-users bar are used almost exclusively by practitioners who have fully onboarded. Features where the two bars are close together are used by both explorers and committed users, suggesting they have value even before a practitioner has completed setup.
Strongest hook feature
Community Profile is the most-used feature overall and shows uptake even among non-converted users (15.6% vs 24.3%). This suggests the community layer has broad appeal across different user types and onboarding stages, which is worth considering when thinking about how new users are introduced to the platform.
Core retention drivers
Activity Planning, Add Child, and Attendance show the largest gaps between all users and converted users, confirming these are depth features used by practitioners who have committed to the platform. Activity Planning in particular sits surprisingly high given it requires lesson planning behaviour change.
Watch list
Finance tools and Progress Reports show low adoption across both groups. Finance usage might reflect a trust barrier: practitioners need confidence in the platform before logging sensitive financial data. Training and Self Assessment also show low numbers, though both features were released significantly later than others so direct comparison is limited.
Part 3: First-day dropout
What the data shows
69% of registered users were last seen on the same day they registered. This is the dominant pattern: most users open the app once and do not return. Among those who do return, conversion rates are higher, but the returning group is small enough that the overall picture is one of early and decisive dropout rather than gradual attrition.
Reading the chart
The same-day bucket dominates. Of the 2,276 same-day users, 713 (31%) converted, likely principals who set up their preschool in a single session and then went quiet. The remaining 1,563 registered and left without converting.
Users who returned after day one are a much smaller group but convert at higher rates: 47% at 1-2 days, 58% at 3-6 days, reaching 95% among users active for 180 or more days. The data cannot answer whether this reflects the platform working better over time, or simply that more committed users were always more likely to return.
Most dropout is immediate
69% of users never return after registration day. The barriers are likely structural rather than product-related: users who were not ready to onboard, were blocked by the principal model, or registered out of curiosity rather than immediate intent. Reminders and nudges have not moved this group in practice.
Design implication
Re-engagement campaigns are unlikely to recover users who left on day one. The higher-leverage investment is likely to be in the first session itself: reducing the time between registration and a first moment of genuine value, so more users find a reason to return before they close the app.
What same-day users actually did
Among users last seen on registration day, Community Profile was used by 10.4%, far ahead of any other feature. Activity Planning (3.3%) and Resources (2.6%) were next. Attendance and Add Child were used by fewer than 2%.
Community Profile is not the most prominent feature on the home screen. Users are actively seeking it out by name, filling in a join form, and still not returning. Something about the community concept draws people in. What they find there is not yet compelling enough to bring them back.
Community as an entry point
Community Profile is used by same-day explorers and long-term converted users alike. It is the platform’s most broadly used feature. The design question for a new platform is: what kind of community only makes sense inside a professional ECD tool and cannot be replicated by the social media tools and groups practitioners are already using.
Inside Community: what users actually do
Community is not prominently placed in the app but users who find it do engage with it. Of 642 who joined, 75% explored the directory and 66% wrote an About section. Engagement drops off before users reach the point of connecting with others.
642
Joined Community
484
Clicked ECD Heroes (75%)
64
Sent a connection request (10%)
20
Had a request accepted (3%)
Only 10% of practitioners who join the community send a connection request, and of 11,406 requests sent, 10,834 have no response. A small number of users are mass-sending requests that almost nobody responds to.
Profile completeness tells a similar story. 66% of community members wrote an About section, suggesting real intent to participate. Only 8% came back to update their profile after joining.
Design implication
The high rate of exploration suggests demand for professional connection. The current mechanic for acting on that curiosity is not working. A new platform should design lower-friction ways for practitioners to find relevant peers before asking them to commit to a formal connection. Filtering by district, school size, or role; shared interest groups; or lightweight reactions to profiles would all reduce the commitment required to move from browsing to connecting.
Part 4: Does organisational support change adoption?
ECD Connect operates across two models: an open access version (ECD Connect App, available to any practitioner in South Africa) and white-label versions customised for specific partner organisations (Ntataise, Khululeka, and True North). White-label users have an organisation behind their onboarding: trained support staff, existing practitioner networks, and in some cases structured rollout programmes. Open access users are entirely self-directed.
This section asks whether that organisational scaffolding makes a measurable difference to conversion and retention.
Reading the data
Two of the three white-label tenants convert at substantially higher rates than open access: Ntataise at 65.3% and Khululeka at 54.5%, compared to 40.8% for the ECD Connect App. Converted users on white-label tenants also stay active significantly longer. Ntataise and Khululeka converted users average over 70 days active, compared to 31 days for OA converted users.
True North is the exception: conversion rate (36.4%) sits below OA, and engagement was limited. ECD Connect’s partnership with True North has since ended. Their small sample size (33 users) makes any trend interpretation unreliable.
Across all tenants, the gap between converted and non-converted avg days active is stark. Non-converted users barely engage regardless of tenant type, averaging under 5 days across the board.
Organisational support drives conversion
The data is consistent: users with an organisation behind their onboarding convert at higher rates and stay active longer. This is not primarily a product quality finding: the same product produces meaningfully different outcomes depending on the support model around it. For any new deployment of this platform, the onboarding support model is as important a design decision as the product itself.
OA conversion is strong given the context
40.8% conversion among entirely self-directed users with no organisational support is a meaningful baseline. These are practitioners who found the platform independently, registered without prompting, and nearly half completed full onboarding on their own. The OA channel is working; the question is what light-touch support could move that number further.
Room to grow with coaching features
None of the partner organisations have used the platform’s coaching feature, which enables structured support for practitioners. Khululeka has plans to pilot this. If coaching drives the kind of engagement uplift seen in the broader WL vs OA comparison, it could be a meaningful lever for both conversion and long-term retention.
Design Implications
A self-directed open access product will always face a conversion ceiling. The data suggests that ceiling sits somewhere around 40%. Pushing meaningfully beyond that requires organisational infrastructure: trained support staff, structured onboarding programmes, and principal-level engagement. Building a great product is necessary but not sufficient for adoption at scale in this context.
Part 5: How has conversion changed over time?
This section focuses on ECD Connect App (open access version), where monthly cohorts are large enough for meaningful trend analysis. White-label tenant monthly cohorts are too small, so overall WL vs OA comparison is covered in the section above.
Several onboarding improvements were made during this period: SSO login (Google and Facebook) to reduce duplicate accounts from forgotten passwords, removal of the “add practitioners” step from the OA setup flow, and auto-assignment of the principal to their class to address confusion identified in training sessions. The data below examines whether these changes are visible in funnel and conversion metrics over time.
Reading the charts
Conversion rate has trended downward as volume has grown, from 55-58% in the first cohorts to 30-32% in early 2026. January 2026 is the clear outlier (185 users at 97.8% conversion) driven by start-of-year intent rather than any specific product change.
The funnel chart tells a more nuanced story. Step 1 completion (adding a name) has been remarkably flat across all cohorts at 55-70%, suggesting identity friction is structural rather than something product changes can address through UX improvements alone. Step 3 (preschool connection) shows more variation and tracks closely with the conversion rate trend. Step 5 (adding a child) is low across all cohorts and appears to be declining in recent months, though this is partly a recency effect: newer cohorts have had less time to complete deeper onboarding steps.
Several onboarding improvements were implemented during this period. The funnel data suggests these changes have not dramatically moved aggregate completion rates. This is consistent with a broader finding: the primary barriers to funnel completion are structural (user intent, institutional access, and timing relative to the school calendar) rather than UX friction that product changes alone can resolve. Changes like SSO likely improved data quality (fewer duplicate accounts) without being visible in conversion rate metrics. The auto-class-assignment change is too recent to have accumulated enough data for a clear before-and-after comparison.
School calendar drives conversion quality
The January 2026 spike is the clearest evidence that when users register matters as much as how they find the platform. Practitioners who register at the start of the school year come in with clear intent to set up systems for the year ahead. Timing acquisition efforts around natural school calendar moments is likely to produce better conversion rates than volume-focused outreach throughout the year.
Growth brings in a broader audience
The downward trend in conversion rate as volume grows reflects the platform reaching beyond its initial early adopter audience. Early users sought the platform out deliberately. Later growth brings in users who registered more speculatively. This is a normal pattern in platform growth, not a signal that something is broken.
Identity friction is structural, not a UX problem
Step 1 and Step 2 completion rates have shown almost no improvement across cohorts despite product changes targeting the onboarding flow. The drop-off at these steps reflects users who register before they have documentation to hand, or who never intended to complete onboarding. This group is likely irreducible through product design alone and may require a different acquisition strategy rather than further onboarding improvements.
High conversion does not equal sustained engagement
The January 2026 cohort converted at 97.8% but retained at 8.6% after 30 days, consistent with other high-volume months. High intent at registration does not translate to sustained engagement. Users who complete setup quickly in a start-of-year window go quiet at the same rate as any other cohort, suggesting the barrier to return is not about initial motivation but about what the platform offers once setup is done.
Part 6: Principals vs practitioners
Of the 3,319 registered users, 1,223 are identifiable as principals, 160 as practitioners, and 1,932 remain unclassified: users who registered but never completed enough onboarding to take on either role. The 8:1 ratio of principals to practitioners shows that the platform is currently used almost entirely by principals, and practitioner adoption is entirely downstream of principal engagement.
Reading the data
The chart shows feature usage rates within each role. Practitioners show higher rates on classroom-facing features; principals dominate on administrative ones.
The platform is principal-led
With an 8:1 ratio of principals to practitioners, the platform currently functions primarily as a tool for school leaders rather than classroom teachers. Practitioner access depends entirely on principal setup, meaning principal recruitment and onboarding quality is the single most important variable in overall platform adoption.
Practitioners are the classroom users
Once onboarded, practitioners use every classroom-facing feature at significantly higher rates than principals. Attendance is used by 40.6% of practitioners vs 10.2% of principals. Calendar (29.4% vs 9.4%) and Progress Reports (17.5% vs 5%) show the same pattern. Community Profile is the one exception, used at nearly identical rates by both roles (26.3% vs 25.9%), suggesting it serves a different need.
Design Implications
Designing around the principal workflow is the right starting point: principals are the gatekeepers, the multipliers, and currently the primary users. But practitioners who do get access engage deeply with classroom features. A new platform should make it as easy as possible for principals to bring practitioners on, then design the practitioner experience around daily classroom workflows.
Part 7: Feature sequencing
Among converted users who used at least one feature, what did they do first? This matters for onboarding design: if there is a dominant entry point, that feature should be prominent in the first session experience.
The analysis covers all converted users with at least one trackable feature event. Users whose first action was creating a class (a required onboarding step rather than a discretionary feature) are excluded so the chart reflects genuine feature choices.
Reading the chart
Community Profile is the first discretionary feature for 16.8% of converted users, nearly double the next feature. Attendance (7.8%) and Add Child (6.9%) follow, reflecting the core administrative workflow. Activity Planning (6.2%) shows a meaningful group drawn in by the content and curriculum layer before anything else. Everything else sits below 4%.
Note that 54% of converted users have no trackable feature events at all, meaning they completed preschool setup but never used a tracked feature. These are likely principals who set up the preschool structure and then stepped back, or users who completed setup in a single session and did not return.
Community Profile is the dominant entry point
Among users who go beyond setup, Community Profile is the most common first feature by a significant margin. Combined with its strong showing in the overall feature engagement data and among same-day users, this is consistent evidence that community is the platform’s primary hook. It draws in explorers, it draws in converted users, and it is used equally by principals and practitioners. Whatever form community takes in a new platform build, it should be prominent from day one.
Two distinct entry paths
The sequencing data suggests two types of converted user. The first goes straight to administrative setup: Add Child, Attendance, Income Statements. The second explores content and community first: Community Profile, Activity Planning, Resources. These are different user intents and may benefit from different onboarding paths rather than a single linear flow.
Part 8: Feature breadth and retention
Users who engage with more features stay active longer: median active days rises from 0 for single-feature users to over 200 for users who engage with six or more. The relationship is consistent and steep, but the direction of causality is not clear.
Reading the data
Users who engage with only one feature have a median active window of 0 days with more than half last seen on registration day. At 3 features the median jumps to 42 days. At 6-8 features it reaches 90-200 days. The 10 users who engaged with 9 or more features averaged 296 days active.
The paper system problem
Many ECD practitioners already manage attendance, finances and child records on paper. For these users, a digital equivalent asks them to change a workflow that is not necessarily broken, on a device that costs data. Whether the value exchange is worth it is not obvious from the practitioner’s perspective.
Genuine new value vs digital data collection
The features with the strongest adoption (Community Profile, Activity Planning, Resources) have no direct paper equivalent. Features like attendance registers and income tracking replicate existing workflows. Even where these features offer a return in the form of automated reports, absence alerts, the value only materialises after significant data entry, which is itself a barrier for practitioners who already have working paper systems.
Design implications
Every feature in a new platform build should pass the practitioner’s ‘so what?’ test: what does the practitioner get back from using this, beyond the act of using it? Features with no paper equivalent have a natural answer. Features that replicate existing workflows need to earn their place by offering something the paper version cannot.
Implications for product design
This analysis covers a platform serving 3,319 registered users across South Africa, from launch through May 2026. The findings are specific to this context but the patterns they reveal are relevant to any digital product targeting practitioners in under-resourced, low-infrastructure settings.
Seven findings stand out with direct implications for how a next-generation ECD platform should be designed.
1. ECD Connect has meaningful traction in a difficult context
A self-directed platform serving practitioners with no mandatory adoption requirement has converted 42% of registered users and built a base of over 1,200 principals. White-label tenants with organisational support convert at 55-65%. Practitioners who onboard fully engage deeply with classroom features. Community draws broad engagement across user types. In a low-resource context where digital adoption is challenging, these numbers represent real traction. The lessons are worth carrying into a new build.
2. The first session is the only session for most users
69% of users were last seen on the day they registered. Reminders do not bring them back. The decision to adopt is made in the first interaction rather than over time. This means the first session experience is the highest-leverage design investment in the entire product. Everything that reduces friction in that first session, from faster preschool connection to earlier access to useful features, will have a larger effect on adoption than any re-engagement campaign.
3. Organisational support matters as much as product quality
White-label tenants with active organisational support convert at 55-65%, compared to 41% for self-directed open access users. The product is identical. The difference is the support model around it. A new platform that relies on self-directed adoption will likely hit a ceiling around 40% conversion regardless of how good the product is. Structured onboarding support, principal-led rollout, and active partner engagement are the primary driver of conversion beyond that ceiling.
4. Timing of acquisition matters as much as volume
Conversion quality varies significantly by when users register, independent of product changes or channel. Users who register at the start of the school year convert at dramatically higher rates than those who register mid-year. Outreach timed around natural school calendar moments (start of year, start of term) is likely to outperform volume-focused campaigns spread throughout the year. A new platform should build its acquisition strategy around the school calendar, not just channel reach.
5. Principals are the gatekeepers and the bottleneck
The platform has an 8:1 ratio of principals to practitioners. Practitioners cannot meaningfully use the platform until their principal has completed setup. Where principals engage, they bring their staff with them. Where they do not, the school does not engage regardless of practitioner willingness. Any new platform should be designed first and foremost for principals, with practitioner onboarding as a downstream consequence of principal engagement rather than a parallel track.
6. Community is the strongest hook, but needs a clearer purpose
Community Profile is the most-used feature overall, the most common first feature among converted users, and the only feature used at equal rates by principals and practitioners. It draws in users who never return and users who become the most engaged on the platform. This signals a demand for professional connection. However, the design challenge is building something that adds value that existing tools (WhatsApp, Facebook, etc.) cannot already provide. These could include verified professional credentials, contextual content, district-level peer connection, structured resources.
7. The platform has two distinct user types that need different design
Principals manage the school: finances, staff, compliance, reporting. Practitioners manage the classroom: attendance, child development, lesson planning. The data shows these are genuinely different use cases with different feature preferences, different active windows, and different onboarding needs. Treating them as a single user type in product design produces a platform that serves neither particularly well. A new build should design the principal and practitioner experiences as related but distinct, with a clear handoff between them.
The core design question for a new platform
What can this platform offer that a practitioner cannot get from paper, WhatsApp, or the tools they already use? The features that show the strongest adoption are the ones with no paper equivalent: community and peer connection, structured curriculum resources, professional development. The features with the weakest adoption are the ones that replicate existing paper workflows without giving anything back. A new platform should lead with features offering new value and earn the right to ask for administrative data entry only once practitioners trust that the platform works for them.
Methodology note
All data is aggregate and anonymised. No individual practitioner, preschool, or child data is reported. Analysis covers ECD Connect App (open access, launched April 2025) and ECD Connect Partner white-label tenants (launched October 2024), through 20 May 2026.
Conversion is defined as having at least one valid preschool connection established. Users are deduplicated by phone number, taking the most recent account where duplicates exist. Feature engagement reflects lifetime usage through the analysis date.
Role classification is based on platform structure: principals are users with a connected preschool carrying a valid registration code; practitioners are users connected to a principal’s preschool. Users who completed registration but did not satisfy either condition are classified as unknown.
The “days active” metric reflects the window between registration date and last login, not total days of use. It is a proxy for engagement duration rather than a precise measure of activity.
Analysis by Kim Tichmann, Data & Product Strategy Lead, ECD Connect, DGMT.
ECD Connect is a product of DGMT. All views are those of the author.