Innovative Approaches to Solar Lead Generation Using Novel Datasets

Period of Performance: 06/08/2015 - 03/07/2016

$225K

Phase 1 SBIR

Recipient Firm

ClearGrid Innovations
60 Robertson Avenue
White Plains, NY 10606
Firm POC
Principal Investigator

Abstract

Generating leads for new business lead generation) is one of the most important costs for residential solar marketers. It is currently poorly targeted and inefficient, due to suboptimal information about the probability that target households will adopt solar. ClearGrid and Professors Ken Gillingham of Yale University and Bryan Bollinger of Duke University propose to leverage two complex data-sets to reduce the costs and increase the efficiency of lead generation. Bollinger and Gillinghams research has noted that locations of existing solar panels increase the likelihood that customers with nearby addresses, including those on the same street, adopt solar. They have also hypothesized that the presence and visibility of solar panels from the street in an area influences both the rate of adoption and the strength of such peer effects. We will develop computer vision and GIS-enabled techniques to identify existing panels and their visibility from nearby streets using image algorithms. This will empower policy actors and solar marketers to target their efforts based on the presence and visibility of nearby existing solar installations. We will analyze the importance of the magnitude and direction of changes in utility rates in the recent past in driving solar adoption augmenting prior work which examined the levels of the rates), and generate algorithms that link this to solar adoption. Our examination will include recent increases in electricity rates in New England, in the study area. Solar marketers often harness customer dis- satisfaction towards their utilities. We will rate interruptions in utility service unplanned outages) and customer satisfaction with their utilities, analyze a statistical relationship to solar adoption, and build these into our predictive model. The new datasets will be fused with existing data sets that predict household solar adoption, including levels of electricity rates, tax incentives, policy factors, and roof characteristics, to generate a predictive Household Attractiveness Index using machine learning techniques. ClearGrid will commercialize a Household Attractiveness Index to solar marketers and home-centered brands to help them better target lead generation and solar marketing efforts. They will also provide lead generation as a service, offering superior quality leads themselves. Finally, they will use their imagery algorithms to assist utilities to detect panels as built mitigating the debate about unauthorized panels in certain areas such as Hawaii. Benefits to government will include helping understand what motivates solar adoption better, allowing better targeting of programs and incentives to potential high-impact areas, and making innovative use of outage data via the White Houses open outage data initiative.