M&V tutorial · meter before / meter after
Meter before / meter after: how fast VPP response is measured
When a virtual power plant (VPP) or demand response fleet provides reserves or regulation, there is no time to build a baseline from history. The grid operator reads the meter just before dispatch, reads it again after, and pays the difference. Here is how that works, and when it breaks.
VPP Economics has no affiliation with, sponsorship from, or endorsement by any utility, aggregator, platform, or device maker named here.
The short answer
- The simplest baseline there is. Whatever the meter read just before dispatch is assumed to continue. The reduction is that reading minus the reading during the response (NAESB).
- It is the method for fast products. Reserves and regulation use it because events are short and unannounced. NAESB allows only this method or metering the generator itself for regulation (NIST).
- Everything rests on one assumption: the load would have stayed flat. ERCOT allows it only for consistently flat loads, and applies no same-day adjustment (ERCOT).
- Short is accurate; long is not. In the example below, a flat baseline misses 3% of the true reduction in a 10-minute event, but 30% over three hours on a rising evening load.
From meter snapshot to measured reduction
Meter before, meter after
A worked example for a 1,000-home aggregation answering a reserve dispatch, using one-minute telemetry.
Swipe to see the full chart →
The response, minute by minute
The dashed line is the flat baseline; the green line is the fleet’s telemetered load. Measured reduction: 1.62 MW.
Swipe to see the full chart →
No history, no weather, no lookback window. Read the meter before, read it after, subtract. Here the fleet averaged 2.43 MW before dispatch and 0.80 MW once it settled, so it is credited 1.62 MW. That simplicity is why it settles reserves within days rather than weeks.
The flat-line assumption
What a flat baseline misses on a long event
The same fleet on a hot evening. Its load would have kept climbing after 4 p.m.; the flat baseline does not know that.
Swipe to see the full chart →
Credit missed as the event gets longer
Share of the true reduction a flat baseline misses, for the evening fleet above.
Swipe to see the full chart →
Over 10 minutes the load barely moves, so the flat baseline is close to the truth. Over three hours on a rising evening load it misses about 30% of what the fleet actually delivered. This is why grid operators use the method for reserves and regulation, and switch to interval-meter baselines for multi-hour capacity and energy events.
Which grid operators use it
Before and after windows by grid operator
Minutes from the dispatch instruction. Each operator picks its own windows to absorb telemetry jitter and ramp time.
Swipe to see the full chart →
ISO New England uses a hybrid: a 10-day historical baseline shifted to match the three 5-minute intervals just before dispatch, so the meter-before reading anchors a baseline that can still move with the day (ISO-NE). New York’s demand-side ancillary services program has been folded into its DER aggregation model (National Grid filing).
What it takes: fast data
Hourly or 15-minute meter data is too coarse: a 10-minute reserve event disappears inside one reading. So this method runs on telemetry, either a fast site meter or device data summed across the fleet. That raises the next question, whether the telemetry is right. ERCOT checks it: in its pilot, the fleet’s telemetered response must land within 10% of what the premises’ own 15-minute meters show (ERCOT). PJM requires one-minute metering for demand resources in reserve events and lets homes without it use statistical sampling, subject to approval (PJM).
Where it fits and where it does not
Unannounced reserve events
No warning means no chance to inflate the “before” reading, and 10 to 30 minutes is too short for the load to drift far.
Regulation
Response is measured second by second against the signal, so the “before” point is always seconds old.
Batteries and flat industrial loads
Steady loads and controllable devices hold their level, which is exactly what the method assumes.
Multi-hour peak events
Afternoon and evening loads climb with heat and occupancy. A flat baseline under-credits real reductions.
Events with notice
With hours of warning, a participant can run equipment harder just before dispatch. Rules use short, recent windows to limit this.
Cycling loads
Air conditioners and batch processes switch on and off. A snapshot during an off-cycle sets the baseline too low; PJM has a separate rule for batch loads (PJM).
How it shows up in the VPP Program Designer
The VPP Program Designer offers meter before / meter after as an M&V choice in Step 5. It needs device or real-time telemetry, and its estimated bias grows with the hours per event set in Step 3, following the curve above: small for a 15-minute reserve event, large for a three-hour peak event. Regulation and reserves in Step 4, the services settled this way, also require fast telemetry.
All six ways performance is measured
Baseline from interval meter
Past meter data, sometimes with weather, predicts what the site would have used.
See the tutorial →For short eventsMeter before / meter after
You are here. Usage just before an event against usage during it.
This pageFor fleetsStatistical sampling
A metered sample stands in for the whole fleet.
See the tutorial →For batteriesMetering device output
Measures what the device itself delivers.
See the tutorial →For firm loadsMaximum base load
Pays for staying below a pre-agreed level.
See the tutorial →For residential programsControl group
Compares dispatched homes with homes held back.
See the tutorial →Method and caveats
A tutorial for program designers, utilities, aggregators and other readers. It is not a substitute for a tariff or settlement manual. Worked examples use illustrative numbers. Operator windows are simplified for display, and some rules come from training or stakeholder materials that may have been updated. Checked October 23, 2026.
Missed credit is the gap between the counterfactual and the flat baseline, averaged over the event, divided by the true reduction. It depends on how fast the load is changing; the example assumes a hot-evening rise.