Methodology

How these numbers are built

Two tools on this page put a number on demand-side management. The benchmark sizes residential energy efficiency (the slower-acting half of DSM) by estimating the potential a utility's public data suggests is still on the table. The calculator sizes residential demand response (the faster-acting half) by estimating the generation a portfolio could help defer. Both run on public data and a short list of stated assumptions. Neither is a forecast.

Here is exactly what each one does, what it assumes, and where it stops. If a number looks off for your service territory, that usually means we're missing context only your team has: your programs, your avoided-cost curve, your load. That's expected. It's the kind of input that sharpens the estimate. These are screening figures meant to start a conversation with your planning team, not replace the work they already do.

What "micro-level" means here

A utility's resource plan models its specific load, its specific programs, and its avoided costs over twenty years. This is not that. These tools take a few public inputs and a handful of planning benchmarks and return an order-of-magnitude estimate in seconds. The point is to frame a question worth asking, not to settle it. Your team has data we don't. When you bring it, the estimates get sharper.

The benchmark

It estimates the potential annual residential efficiency value indicated by public benchmarks, and ranks each utility against peers of the same type (investor-owned and municipal, separately). It is an order-of-magnitude figure, not a measured shortfall.

The dollar figure is: residential energy sold (MWh) × the share still achievable each year × the share not yet captured × the avoided cost of a megawatt-hour. Every input is a planning assumption, not a measured quantity, which is why we show a range.

Residential energy sold (MWh) × share still achievable each year × share not yet captured × avoided cost of a megawatt-hour

In practice:

Residential MWh: the utility's own filed number, from EIA Form 861.
Achievable each year: about 1.25% of residential sales, a midpoint consistent with published achievable-efficiency ranges.
Still untapped: the utility's gap score, ranked against its peer group (investor-owned vs. municipal). It blends three figures the utility itself files with EIA Form 861 every year: residential customer base (40%), smart-meter coverage gap (30%), and reported energy-efficiency and demand-response activity (30%). The activity piece uses the utility's own reported MWh of efficiency savings and MW of peak demand response cut; a utility reporting little activity ranks higher on "untapped." These are self-reported numbers, not BlastPoint estimates.
Avoided cost: about $150/MWh, a blended energy-and-capacity value. When new capacity is scarce and expensive, the value of not needing it goes up.

For Georgia Power (about 2.5 million customers, 29 million MWh sold to homes), that comes to roughly $22 million a year, with a range around it.

What the benchmark does not know: your actual programs, your avoided-cost curve, your regulators, or your customers. A utility that already ran aggressive efficiency programs for a decade will show little untapped opportunity here, and that is the tool working as intended, not a knock on the program. A small municipal shows a small figure because it serves few homes, also expected. The benchmark sizes the opportunity in public terms. It does not tell you whether to act, or how.

The calculator

It estimates how much peak capacity a demand-response portfolio could contribute, and the new-build cost that capacity could help defer, under stated assumptions and before your own reliability constraints. It does not count the megawatt you still need on the rare day the program can't deliver. You set four inputs:

Summer peak (MW): your system peak.
Enrollment: the share of residential customers who join. The default, 24%, is the participation Portland General Electric reported across its flexible-load programs.
Program mix: direct load control (you cycle a device) versus behavioral (the customer responds to a price signal).
Avoided new-build cost: the default, $1,300/kW, is a planning midpoint for a new combustion-turbine peaker, consistent with the Lazard LCOE+ June 2025 v18.0 ranges.

The tool caps the addressable shift at 15% of your summer peak. This is a deliberately aggressive, best-case ceiling, not a typical result: published assessments put realistic demand-response potential at roughly 5 to 10% of peak, with 15 to 20% reached only under full-participation scenarios. The cap keeps the estimate physically grounded, so the tool never shows a shift larger than a credible program could deliver against your peak.

Two per-household numbers do the work. A direct-load-control household contributes about 1.2 kW at a typical event; a behavioral household about 0.8 kW. Both are directional planning benchmarks. Real yield depends on climate, device mix, event design, and how the program is run.

The avoided-capex figure is net of program operating cost over a ten-year horizon. It answers one question: if you met this much peak capacity through demand response instead of building it, what construction cost could you defer? It does not model your load shape, your interconnection queue, or what your regulators will approve. It is a back-of-the-envelope made interactive. Use it to decide whether a real study is worth commissioning.

Where the numbers come from

EIA Form 861 (2024, via Catalyst Cooperative's PUDL) for residential sales and load, smart-meter (AMI) coverage, and the utility's self-reported energy-efficiency savings (incremental MWh) and demand-response peak savings (MW); the 1.25%/yr achievable-efficiency midpoint reflects utility resource-plan filings and program-potential studies; Lazard LCOE+ June 2025 v18.0 for avoided generation cost; Portland General Electric, presented at the SECC demand-response webinar (May 2026), for the participation benchmark. ACEEE's State Energy Efficiency Scorecard is referenced for broader efficiency-program context.

If you want a number you can take to your planning team, the next step is your data: your load shape, your program history, your avoided-cost curve. We'll run it and show our work. Until then, treat everything here as a directional estimate, built in the open and ready to be pressure-tested against your data.