Every vendor selling energy optimization software will hand you a case study claiming 20–30% savings. The problem is that most of those case studies have bad baselines, short measurement windows, or are from facilities with obviously low-hanging fruit that your building may not have. Before you sign anything, you need a framework for building your own projection — one that does not depend on accepting the vendor's math.
This is the framework we walk through with prospective customers before they start a pilot with us. It is not complicated, but it does require some homework on your end.
Start With a Clean 12-Month Baseline
The most common ROI calculation mistake is using the last three months of utility bills as a baseline. Seasonal variation makes short baselines meaningless. A building in North Carolina that measures its baseline in October through December will systematically understate summer cooling costs and overstate heating efficiency.
You want at minimum 12 months of interval data — ideally 15-minute interval reads from your utility meter or submeters, not just monthly billing totals. If your utility provides AMI (Advanced Metering Infrastructure) data, request it. Most commercial and industrial accounts in ISO-NE, PJM, and MISO territories can get 15-minute interval history going back 24 months through their online portal.
From that data, calculate your load shape: average weekday kW by hour, average weekend kW by hour, and your monthly peak demand values. The load shape reveals where optimization is possible. A flat load curve — HVAC running at consistent levels overnight and on weekends — is often a sign of a badly programmed BMS. A spiky curve with sharp morning ramp-ups points to night setback schedules that create demand peaks at recovery time.
Identify the Savings Categories That Apply to Your Facility
Energy optimization software can generate savings from several distinct mechanisms. Not all of them apply equally to every building. Before evaluating any vendor, identify which categories are relevant to you:
Schedule optimization — shifting equipment run times to avoid peak periods and reduce runtime overall. This applies to nearly every building, but the magnitude depends heavily on how outdated your current schedules are. If your BMS was commissioned in 2015 and has not been touched since, schedule optimization alone can save 10–18% on HVAC energy. If you already run tight schedules, the incremental gain is smaller.
Demand charge reduction — flattening 15-minute peak demand to lower your demand charge tier. For industrial accounts on tariffs where demand charges represent 30–50% of total cost, this is often the single largest savings lever. Commercial accounts on simpler tariff structures benefit less.
Pre-conditioning — cooling or heating a space before occupancy starts, then reducing HVAC activity during peak occupancy, when the thermal mass of the building sustains comfort. This requires accurate occupancy prediction and weather forecasting, which is where predictive platforms differ from purely schedule-based tools.
Demand response participation — automated load shedding during utility-called DR events in exchange for bill credits. The value depends on your local utility's program structure. If you are on a C&I DR program in a territory like NYISO, ERCOT, or PJM, this can add meaningful revenue. If your utility has no active DR program, this category contributes nothing.
We are not saying every building will see savings across all four categories. A single-story light-manufacturing facility in a mild climate may only benefit meaningfully from demand charge reduction and schedule optimization. A multi-floor office tower in a hot climate with a demand-response-eligible tariff has all four levers available. Be honest about which ones your building actually has access to.
Building Your Own Payback Projection
Once you have your load shape data and a sense of which savings categories apply, you can build a rough payback model. Here is a simplified version:
Take your trailing 12-month energy spend. Apply a conservative savings percentage — we suggest 12–18% for a mid-sized commercial building with a reasonably modern BMS, or 18–25% for an industrial facility with older setpoint-based controls and high demand charges. That gives you your annual savings estimate.
Then get quotes on the software cost — both the implementation fee (if any) and the annual subscription. Divide the annual software cost by the estimated annual savings. If that ratio is below 0.4 (meaning the software costs less than 40 cents for every dollar it saves), the economics are compelling. If it is above 0.6, push back on either the savings estimate or the pricing.
For a concrete example: a 200,000 sq ft office building in Raleigh with a $180,000 annual electricity bill might project 15% savings, or $27,000/year. A subscription that costs $10,000–14,000/year represents a strong ROI. A $24,000/year subscription starts to look marginal. These are not real numbers from any specific customer — they are illustrative of the calculation structure.
What a Pilot Should Prove (and What It Cannot)
A 30–60 day pilot is the standard way vendors let you test before committing. But most pilots are structured in ways that inflate the measured savings. Watch for these issues:
Weather-adjusted baselines: If the pilot runs in April and the comparison period is February, savings will look artificially high because April is lower-load season. Any credible pilot uses degree-day normalization to compare equivalent weather conditions. Ask the vendor directly: how are you normalizing for weather differences between the baseline and pilot periods?
Short measurement windows: Demand charge reduction requires observing at least one full billing period — typically a calendar month — to see the effect on your peak demand reading. A two-week pilot will not show demand charge savings.
Equipment state at baseline: If any HVAC equipment was replaced or recommissioned right before the pilot, the baseline is contaminated. The new equipment would have reduced energy anyway, independent of the software.
A well-structured pilot should produce: a baseline load shape for the comparison period (weather-normalized), a measured load shape during the pilot, and an attributed savings calculation with the methodology explicitly stated. If a vendor cannot give you that in a pilot summary, treat the results as unverifiable.
The Discount Rate Problem
One thing facility managers sometimes overlook: the discount rate for energy savings projects within their organization. Many companies apply a hurdle rate of 15–25% when evaluating capital projects. Energy software, however, is a subscription — it is an operating expense, not a capital expenditure. That changes the financial framing significantly.
An OpEx that returns $2 in savings for every $1 spent, every year, with no capital tied up, is financially different from a capital project with a 3-year payback. Make sure your internal financial review is using the right framework. Some CFOs default to capital project IRR models for anything energy-related, which puts energy software subscriptions at a structural disadvantage compared to how they should actually be evaluated.
The Numbers a Vendor Should Show You Before You Commit
Here is a short checklist of what any reputable energy optimization vendor should be able to provide before you sign a contract:
- A weather-normalized baseline methodology, with an explicit description of the normalization approach
- Savings attribution that separates schedule optimization, demand charge reduction, and pre-conditioning contributions (not a single blended number)
- Average measured savings across comparable facilities — building type, climate zone, and tariff structure should all match your profile
- The specific BMS integration method they use and what telemetry they require
- Pilot structure with defined success criteria agreed in writing before the pilot starts
If any of those are missing, the vendor is asking you to take on the measurement risk. That is not necessarily a dealbreaker, but it should be reflected in the contract terms — specifically in how annual savings commitments, if any, are structured.
We built Voltpathio's pilot process around weather-normalized measurement from day one, because we found early on that customers who could not verify the savings number were the ones who churned. The easiest way to keep a customer long-term is to make sure the math checks out.