← All dispatches
Analysis7 min read

Measured From Day One

South Africa's draft efficiency strategy says its earlier programmes were held back by the absence of verified measurement, and plans a baseline from day one. DARES is designed to pay its second mini-grid milestone on use measured a year after commissioning. Both depend on a record that starts on the first day.

Abraham Onoja
Abraham Onoja
Founder · Gen318
DWG-17 / TWO PLANS, ONE DEPENDENCY GEN318 / REV 01.00 A TARGET OR A PAYMENT IS ONLY AS GOOD AS THE RECORD BEHIND IT · THE RECORD HAS TO START ON DAY ONE SOUTH AFRICA · DRAFT NEES 2025–35 STRATEGY SEQUENCE, AS DRAFTED · TIMELINE NOT TO SCALE YEAR 1 · 2026/27 YEAR 3 · 2028/29 2035 M&E METHODOLOGY ADOPTED BEFORE PROGRAMMES SCALE OBJ. 5.1 · P. 55 MANDATORY ENERGY-DATA REPORTING REGULATIONS LARGE USERS: ENERGY PLANS NATIONAL ENERGY DATA REPOSITORY, VERSION 1 OBJ. 5.3 FIRST NATIONAL ENERGY CONSUMPTION SURVEY OBJ. 5.5 132 PJ AVOIDED ABOUT 37 TWH TARGET, NOT A MEASUREMENT “EVERY INTERVENTION … CARRIES A MEASURABLE BASELINE FROM DAY ONE.” — DRAFT, P. 55 NIGERIA · DARES MINI-GRID GRANT TWO MILESTONES, AS DESIGNED IN THE APPRAISAL DOCUMENT DAY 0 · COMMISSIONING +12 MONTHS SITE RECORD FROM DAY ONE: THE WHOLE YEAR THE SECOND MILESTONE RESTS ON NO RECORD RECORD STARTED LATE: FIVE MONTHS CANNOT BE REBUILT MILESTONE 1: COMMISSIONING AND CUSTOMER CONNECTIONS MILESTONE 2: CAPACITY UTILISATION FACTOR SMART-METER INTEGRATION REQUIRED TO CHECK USE THE CHEAPEST DAY TO START A RECORD IS THE FIRST ONE. GEN318 / DISPATCHES DRAFT NEES (DEE) · DARES PAD (WORLD BANK) · LATE BAR ILLUSTRATIVE SCALE 1:1 / SHEET 1 OF 2
South Africa's draft strategy plans a baseline from day one; the DARES grant is designed to pay its second milestone on utilisation a year after commissioning. The late-start bar is illustrative.Open full-size blueprint

On 27 October, South Africa's Department of Electricity and Energy and the World Bank Group hosted an online workshop on the draft National Energy Efficiency Strategy. 957 people registered for it. The draft sets a target of 132 petajoules of avoided final energy demand by 2035, about 37 terawatt-hours. Its diagnosis is more useful to Nigeria than its target.

The draft explains why South Africa's earlier efficiency strategies under-delivered. The constraint, it says, "has not been technical potential, cost-effectiveness or awareness of the opportunity, but the absence of adopted instruments, assigned responsibility, durable financing and verified measurement." Later it is blunter: "Timely and accurate data is the single most consequential enabler of this Strategy, and its absence is the most serious gap identified in Part Three."

This piece is about the last item on that list, verified measurement, and why it matters for a Nigerian programme that pays part of its grant a year after commissioning. Two notes first:

  • The quotations come from the draft strategy (Version 1, dated 1 August 2026), which DEE has published for public comment. I read it on 28 October. Page numbers are the draft's printed ones, and sources are listed at the end. The draft may change.
  • Gen318's commercial interest is stated at the end.

What South Africa found when it went looking for data

The draft rates South Africa against sixteen indicators from the World Bank's RISE framework. It adds that its ratings are the authors' own and "do not represent official RISE scores, nor do they constitute an endorsement or position of the World Bank Group". Two of the red ratings concern measurement:

  • Large energy users. South Africa "has virtually no mandatory energy efficiency requirements for large energy consumers. There are no mandated energy audits, efficiency targets, energy management systems, or measurement and verification programmes for large-scale users" (p. 26).
  • Energy statistics. National energy balance information is not timely and lacks accuracy, and "There is also no centralised and up-to-date repository for energy data" (p. 28).

Its list of barriers is blunter: the available data "is incomplete and inconsistent, hindering evidence-based policymaking, baseline setting, and programme targeting" (p. 23). The executive summary shows what that costs: "Agriculture and power generation still require stronger data before numerical targets can be finalised" (p. 7).

What it plans to build, and in what order

The draft does not respond with a new technology. It responds with sequencing:

  • Year 1 (2026/27). Adopt a monitoring and evaluation methodology "before programmes scale, so that every intervention launched under this Strategy carries a measurable baseline from day one" (p. 55). Promulgate regulations on the mandatory reporting of energy data, and require large users to complete energy management plans.
  • By 2028/29. Commission the first version of a National Energy Data Repository and field the first national energy consumption survey (p. 56).
  • Money that depends on measurement.
    • A government-backed Super ESCO with "robust MRV protocols".
    • Utility efficiency obligations under "standardised measurement and verification under SANS 50010".
    • A simpler claiming route under the Section 12L tax allowance, which gives a deduction of 95 cents per kilowatt-hour of verified savings (p. 26). Smaller projects would use "deemed savings or standardised baseline approaches that cut M&V costs" (p. 40).

The phrase that matters is "from day one". A baseline is cheapest to record before an intervention starts; built later, it rests on bills and estimates. An operating record has no such fallback: a month a site did not log cannot be logged later.

The Nigerian programme with the same dependency

DARES is the $750 million mini-grid and off-grid programme the World Bank finances in Nigeria. As dispatch 12 set out, its project appraisal document says the grant for isolated mini-grids "will be disbursed against two pre-defined milestones: commissioning and customer connections, and upon meeting a certain capacity utilization factor one year after commissioning." The same document says that "smart meter integration with Odyssey will be required from mini grid projects in order to verify electricity consumption".

DARES is not South Africa's strategy, and I am not claiming it has South Africa's data problem. The appraisal document plans for metered verification, which is the kind of measurement South Africa says it lacks. The point is narrower. By design, the second DARES payment rests on a utilisation figure measured a year after commissioning. Like South Africa's targets, that figure is only as good as the record behind it, and the record has to begin at commissioning.

So the question South Africa has just asked itself in public is a fair one for every DARES developer to ask privately: when the one-year milestone falls due, what will the record for each site show, and how much of the year will it cover?

Three labels every reported figure needs

South Africa's draft and the Nigerian programme point to the same three disciplines. None of them is specific to any product.

DWG-18 / ONE REPORTED FIGURE, THREE LABELS IT NEEDS GEN318 / REV 01.00 ILLUSTRATIVE VALUES · ONE FACILITY, ONE YEAR · SEE DWG-17 FOR WHY THE RECORD STARTS ON DAY ONE REPORTED SAVINGS, YEAR 1 412 MWH WITHOUT LABELS: ONE NUMBER THAT CANNOT BE CHECKED OR ADDED MEASURED 300 MWH DEEMED 112 MWH 1 WHAT KIND OF NUMBER MEASURED: 300 MWH SANS 50010 / IPMVP, ADJUSTED BASELINE DEEMED: 112 MWH AN ESTIMATE · REPORTED SEPARATELY MISSING: STORED AS MISSING NEVER RECORDED OR TOTALLED AS ZERO 2 HOW MUCH DATA ARRIVED DELIVERED: 96.2% 8,425 OF 8,760 EXPECTED HOURLY READINGS TRUSTED: 99.1% 8,349 OF 8,425 RECEIVED PASSED CHECKS NEITHER STANDS IN FOR THE OTHER A 100% CHECK PASS CAN HIDE 75% LOSS 3 HOW WELL THE BASELINE FITS R² 0.86 · GUIDE > 0.70 · MEETS CV(RMSE) 11.4% · GUIDE < 25% · MEETS NMBE 0.0% · GUIDE ±0.5% · MEETS FIT ON OUTPUT, WEATHER, OCCUPANCY UNCERTAINTY ±18% AT 90% MEASURED: 300 ± 54 MWH GEN318'S RULE: A BASELINE OUTSIDE THE GUIDE VALUES GIVES NO NUMBER; THE FIGURE IS REPORTED AS INSUFFICIENT. GUIDE VALUES: LBNL RECOMMENDED GUIDANCE (2019), NOT PASS/FAIL IN THE SOURCE · UNCERTAINTY: ASHRAE GUIDELINE 14 FSU A NUMBER WITHOUT ITS LABELS CANNOT BE ADDED TO ANOTHER NUMBER. SAY WHAT KIND OF NUMBER IT IS · SAY HOW MUCH DATA ARRIVED · SAY HOW WELL THE BASELINE FITS GEN318 / DISPATCHES ILLUSTRATIVE · NOT A MEASURED FACILITY · ARITHMETIC SHOWN IN FULL SCALE 1:1 / SHEET 2 OF 2
DWG-18. One reported savings figure, with the three labels it needs: what kind of number it is, how much of the expected data arrived, and how well the baseline fits. Illustrative values.Open full-size diagram

1. Say what kind of number it is. The draft's deemed-savings route for small 12L claims is sensible: measurement costs money, and small projects cannot carry it. But a deemed saving is an estimate. If deemed and measured savings are added together, a national progress report cannot say how much of the total was measured. Label each figure as measured or deemed, and keep the two in separate lines.

2. Say how much of the data arrived, and never record a gap as zero. A meter that should report every fifteen minutes and reports once an hour has delivered a quarter of its data. A check on what did arrive can still pass at 100%. Completeness has two parts: the share of expected readings that were received, and the share of received readings that passed checks. Neither can stand in for the other. South Africa's complaint that its data is "incomplete and inconsistent" is close to these two parts. A gap should be stored as a gap: a zero reads as "no consumption" or "no supply", and lowers a baseline or raises an outage count without anyone deciding it should. This is cheap to require in a reporting regulation and very expensive to correct in a national database afterwards.

3. Say how well the baseline fits. A facility's consumption can fall for reasons unrelated to the measure: lower output, a cooler year, fewer occupants. A savings figure should say whether its baseline was adjusted for those, how well that baseline fitted the data, and how wide the uncertainty is. ASHRAE Guideline 14 and LBNL's acceptance guidance describe the usual measures: R², CV(RMSE) and NMBE.

Gen318's interest

  • Gen318 builds software that keeps operating records for solar, battery, generator and metered sites in Nigeria. It reports data completeness both ways and keeps missing data as missing. We have a commercial interest in this argument.
  • Our product only partly matches this. It does not yet adjust a baseline for output or weather, or report the baseline's fit. Today's diesel-savings figures use a flat baseline and are graded as modelled.
  • We have no commercial presence in South Africa. We are preparing a comment on the draft, on method only, along the lines of the three labels above.
  • None of the three labels requires Gen318. Any developer, regulator or funder can write them into a reporting template with the systems they already use.

One question for DARES developers. Before your first site is commissioned, decide what its one-year record will contain and how you will show how complete it is. The cheapest day to start that record is the first one.

Corrections to anything above are welcome and will be listed.

Sources

Fetched and read on 28 October 2026. The draft strategy, the comment notice and the World Bank appraisal document are primary sources.

RelatedSite records from commissioning for DARES developers →
Abraham Onoja
Written by
Abraham Onoja

Abraham is a CTO based in Abuja, building AI systems for the energy sector. He writes the dispatches — data-heavy, sourced, and from the ground the mini-grids stand on.

Read the thesis →hello@gen318.com

Connect the analysis to your operating workflow.

Request a demo