“Learners reached” answers an important question: who had access? It does not tell us what they experienced, learned, retained or did next. The figure becomes misleading only when it is made to carry claims it cannot support.
Good impact communication does not make every number large. It makes every number legible.
Define reach before reporting it
State the reporting period, participant definition, attendance threshold, duplicate-removal method, programme dosage and data source. Distinguish registrations, unique participants, active learners, completers and returning learners.
Layer 1: Reach and participation
Track who entered, how often they participated and whether access was equitable. Disaggregate by relevant characteristics and compare intended with actual participation.
Layer 2: Experience and implementation
Was the programme appropriate, usable and delivered as intended? Record session completion, equipment availability, teacher adaptation, learner engagement and barriers. UNICEF’s Ghana implementation research uses appropriateness, adoption, fidelity and feasibility to examine how digital learning works in real classrooms.
Layer 3: Capability
Use authentic tasks rather than only self-report. In coding, ask learners to predict, debug and modify. In robotics, assess system models, prototypes, tests and explanations. In AI literacy, assess verification, bias recognition, disclosure and responsible decision-making.
Layer 4: Practice and transfer
Can educators apply the method in their own lessons? Can learners transfer an idea to a new problem? Observe practice, review artefacts and use follow-up tasks after the initial programme.
Layer 5: Progression and durability
Track what happens next: further courses, school projects, subject choices, mentoring, leadership, continued teacher use and institutional budget or timetable allocation. Sustainability is visible when routines and ownership persist after intensive external support reduces.
Build a minimum viable evidence plan
- Write one clear outcome for each audience.
- Select one or two indicators that genuinely test that outcome.
- Collect a baseline where change is being claimed.
- Combine quantitative evidence with artefacts and observation.
- Protect personal data and collect only what will be used.
- Schedule reflection and adaptation—not only final reporting.
Avoid common measurement traps
- Using satisfaction as proof of learning.
- Using a showcase team as evidence for every participant.
- Changing indicator definitions between reports.
- Reporting cumulative reach without a period or denominator.
- Collecting large amounts of data with no decision attached.
- Hiding implementation failure instead of learning from it.
Report with proportion
If the evidence shows access, say access. If it shows short-term skill improvement, say that. Do not imply long-term employment or national transformation without the design and evidence required to support those claims.
Download the STEM Programme Evaluation and Impact Framework.
Why the largest number is usually the weakest claim
“Learners reached” is useful for describing distribution, but it says almost nothing about what happened after contact. A learner who watched a demonstration, attended one session and completed a term-long project may all appear as one person in the headline total. Without dosage, participation and outcome definitions, the number cannot guide improvement or justify comparison.
This does not mean every programme needs an expensive experimental evaluation. It means every claim should have evidence proportionate to its ambition. If the claim is access, document access well. If the claim is capability, observe capability. If the claim is transformation, expect a far stronger design.
Build measurement around decisions
Before selecting an indicator, name the decision it should inform. Attendance might help adjust scheduling. A debugging task might reveal a curriculum gap. Classroom observation might change coaching. A maintenance record might determine whether the operating model can scale. Data without a decision becomes reporting theatre.
Separate four commonly confused ideas
- Output: what the programme delivered.
- Outcome: what changed in capability or practice.
- Contribution: the credible role the programme played among other influences.
- Attribution: the stronger claim that the change was caused by the programme.
Honest contribution evidence is more valuable than careless attribution. Credibility grows when limitations are visible.
A compelling evidence story
Combine a small number of disciplined measures: who participated and for how long; what authentic task they completed; how performance changed; whether they used the capability elsewhere; what teachers or schools sustained; and what the programme learned from failure. Add participant work, observation and explanation so readers can see the change rather than merely trust the headline.
The goal of measurement is not to make the programme look certain. It is to make the next decision wiser. Organisations that can explain what did not work—and what they changed—signal greater maturity than organisations whose dashboards contain only success.
What a credible public impact statement contains
A strong statement names the reporting period, participant group, dosage threshold, method for avoiding duplicate counts and source of the data. It separates learners from teachers, exposure from completion and direct delivery from partner-supported reach. When capability is claimed, it describes the task, rubric or observable evidence behind the claim.
This discipline does not weaken the story. It gives serious partners a reason to trust it. Precision is especially important as programmes grow across schools, regions and delivery partners, where inconsistent definitions can turn a dashboard into false confidence.
Applying the evidence ladder to my own numbers
Across InovTech’s wider portfolio, we report 25,000+ learners reached, 2,500+ educators trained and 100+ schools engaged. KodeVR reports 10,000+ active learners. These figures answer important reach questions, but they do not answer every impact question. I should be held to the same standard I recommend to the sector.
The next evidence layer must show dosage, completion, capability and transfer: how frequently learners practise, what they can build or explain, whether teachers apply the training, how participation differs by gender and school context, and which institutions sustain delivery after direct support reduces. Earlier programme records documented girls as 45% of more than 20,000 participating learners; that figure becomes more useful when connected to meaningful technical roles, completion and progression.
One field moment captures the difference between a number and a learning question. A child encountering a robotics kit for the first time froze and whispered, “So this thing is real?” The moment is moving, but exposure is only the first rung. The evidence challenge is to follow what that child can imagine, build and explain next.
