Using randomized split-model validation, Recurve’s recent study of ComEd’s energy efficiency programs proves conclusively that customer targeting accurately predicts results from future customers and can significantly improve customer savings, cost-effectiveness, and grid impacts.
This is Part 2 of a blog series about Recurve’s recent study of four ComEd energy efficiency programs. Read Part 1 here.
To make efficiency and other demand-side solutions effective, finding those customers who are most likely to see energy bill savings and deliver the greatest value for the grid is critical.
The study “Utilizing Smart Meter Data to Improve Program Cost-Effectiveness and Customer Outcomes” tested the ability of Recurve’s customer targeting methods to identify and isolate the highest and lowest savers to optimize program effectiveness.
The study proves conclusively that customer targeting based on historical data predicts results from future customers. Customer targeting was effective for different customer cohorts, including income-eligible/qualified and market-rate residential and commercial customers.
In other words, by studying past customers, program managers can predict which potential future customers are most likely to save money and energy and then prioritize program funds for those customers who will benefit the most from engagement. This prioritization can significantly improve program results including:
- Program cost effectiveness
- Maximum customer savings
- Highest grid value delivered
The study results showed clearly that by applying targeting parameters to historical customer data, it's possible to predict which customers will save the most energy and benefit from the program intervention.
Using smart meter (AMI) or monthly traditional meter data, Recurve analyzes the usage characteristics of customers who saved the most from past program participation in order to identify and prioritize those future customers who are most likely to save from a specific energy efficiency program or intervention.
It is then possible to identify the customers in the population who are statistically likely to have much better outcomes (or those customers who are likely to have poor outcomes and are better candidates for something else). Targeting the right customers can make programs more cost-effective by increasing savings and benefits and ensuring customers see real impacts on their energy use.
This study proved that targeting profiles developed using historical data is a predictor of future outcomes.
To test whether its targeting methods were actually predictive, Recurve conducted a blind, controlled experiment using data from four ComEd programs (Market Rate Weatherization Program (WX), Low-Income Weatherization Program (WX), Central Air Conditioning Program (CAC) for Early Replacement, DX Tune-Up Program (TUNE)).
The study team divided program participants into two groups. Using targeting parameters derived from the first group (Group A), the team used historical (pre-program) customer data from Group B to predict how the second group would perform after an intervention. The team then compared these predictions with real-world post-program results for Group B and found that the targeting predictions were accurate.
To conduct the analysis, ComEd provided pre-program (baseline) AMI data from 2017-2018 for all participating customers, along with data for half of the program participants (Group A). In line with GRIDmeter methods, the Recurve team also used separate non-participant data to correct for any population-wide (external) changes in energy consumption.
With this baseline and performance data, the Recurve team was able to identify the targeting parameters correlated with the highest-saving customers in Group A. The team then used those parameters as a basis for a hypothetical customer targeting analysis on the remaining 50% of customers (Group B) for which it had not been given performance period data.
Recurve identified a subset of Group B customers that fit the targeting parameters from Group A and predicted that they would be responsible for the highest energy savings. After receiving this list of predicted highest savers, ComEd provided Recurve with the performance period data for Group B to test the predictions.
Recurve’s study of ComEd’s efficiency programs proved conclusively that customer targeting based on pre-program AMI data was an effective method for identifying the highest savings participants in a program. Customer targeting was effective for different customer cohorts, including residential, income-qualified, and commercial customers.
In other words, simply by analyzing the past data of customers, it’s possible to know in advance who in the future will save a lot (and who won’t) for a wide range of efficiency and demand-side programs.
Knowing who the high- and low-savers will allow programs to optimize offerings, focus on those customers most likely to benefit, and ensure the programs are both cost-effective and equitable.
To learn more about Recurve’s approach to identifying the right customers for your program, get in touch.