ForLab+
A platform that automates the government supply planning process with practical intelligence — so planners and logisticians can check, balance, and reason about their own data in real time.
The problem we set out to solve
Before ForLab+, supply planning across Ethiopia's 5,100 health facilities happened on paper. Each facility completed a quarterly supply plan by hand or spreadsheet. Regional teams collected these plans, often on their own timelines. All reconciliation happened in Addis Ababa — spreadsheet after spreadsheet, matched up manually. The national picture took weeks to emerge.
Consumption patterns that didn't match forecast — stockouts that followed predictable seasonal trends, demand spikes the models didn't catch — these anomalies surfaced too late. By the time a facility flagged a problem, the next quarter's medicines were already in transit. Ministry teams knew the issues were there. They could see it in the data. The limitation was tooling, not insight.
Planners spent more time on data collection and reconciliation than on reasoning about what the numbers meant. Logisticians worked with incomplete views. The same data existed in different forms in different systems — or different forms in the same system — and confidence in the decisions being made suffered.
Our response
We automated the existing process, not replaced it. Ministry teams use ForLab+ to do quarterly planning the same way they always have — but from one place, with real-time visibility into what each facility is reporting. We surface anomalies where they emerge: when a facility's forecast variance exceeds a threshold, when a regional aggregation breaks from historical pattern, when data quality issues appear. Every user sees the same numbers, at the same time. No new theory of operations. No software that expects you to work differently than you ever have.
Four core functions
Quarterly supply planning
Facility-level plans roll up through woredas and regions into a single national view. Ministry teams see what facilities submit, as they submit it.
Forecasting & reconciliation
Consumption, morbidity, and demographic forecasts, reconciled against stock on hand. Variance flags automatically where data diverges.
Data quality surfaced
The platform identifies incomplete, inconsistent, or implausible data. Fewer spreadsheets to chase, faster time to confident decisions.
Review and approve
Every plan has a complete paper trail. Reviewers query, comment, approve, or escalate without leaving the product.
Ask the system what it sees
ForLab+ learns patterns over time. Planners ask questions in English or Amharic and get answers grounded in their own data: Which facilities are likely to stock out next quarter? Where are consumption anomalies? How does this quarter compare to last year?
- Which facilities are likely to stock out next quarter?
- Show me consumption anomalies across the Oromia region.
- Compare this quarter's plan with last year's actuals.
- Flag reconciliation errors before submission.
High adoption was achieved without heavy training — by aligning with the workflows planners already had, and giving them faster answers inside them.
From Opian's implementation evidence. User voices will be published here with their consent.
Closing the loop, every season.
A forecast you never check against reality is a guess with a spreadsheet. ForLab+ treats every planning season as an experiment whose results feed the next one.
Every calculation is recorded
Each forecast keeps a full calculation audit trail — the method, the data, the assumptions. Nothing about how a number was produced is lost.
Forecasts are triangulated
Before a plan is approved, forecasts are scored against independent signals — consumption history, service data, warehouse issues. Plans that disagree with the evidence get flagged, not filed.
Judgment is part of the record
When a planner overrides an assumption, the override and its reasoning are captured. The next season starts from what people actually decided, not just what the model said.
Error is measured, then reduced
Forecast error (MAPE) is tracked per commodity against actual consumption, with explicit improvement targets in our 2025–2028 blueprint. Audited accuracy figures will be published as they are validated.
On the roadmap: machine-learning forecasting models trained on multi-country consumption data, as ForLab+ deployments expand beyond Ethiopia.
Evidence
ForLab+ has been in production at national scale across Ethiopia since 2022. The figures below are drawn from Opian's implementation evidence.
Forecast error (MAPE) is tracked per commodity inside the platform. Audited accuracy figures will be published as they are validated.
Who uses ForLab+
Deployment tiers
ForLab+ is licensed to ministries of health, typically with donor co-funding. All tiers include data sovereignty commitments.
Pilot at sub-national scale — one or two regions, one program area, limited integration.
Full national rollout across major programs. Integration with existing LMIS and DHIS2.
All programs, all facilities, deep integration, dedicated TA. Includes sovereign-deployment options.