AI for Solar Installation: Sales Estimating and Project Scoping

The residential solar industry has a conversion problem. According to EnergySage and SEIA data, the average solar installer converts only 10 to 15% of qualified leads into signed contracts. The single biggest reason is speed: most companies take 3 to 7 days to deliver a proposal after initial contact. By the time a homeowner gets a detailed quote with system size, production estimates, and financing options, they have either lost interest, signed with a faster competitor, or talked themselves out of the purchase entirely. Every additional day between lead capture and proposal delivery drops close rates by roughly 7 to 10%.

The underlying bottleneck is the manual work required to produce an accurate proposal. A sales rep needs to assess the roof (orientation, pitch, available area, obstructions), determine the optimal panel layout, model shading from trees and nearby structures, calculate expected energy production, pull the customer's utility rate schedule, run financial scenarios across cash purchase, loan, and lease options, and package everything into a polished document. That process involves at least 3 to 4 different software tools, often a site visit, and 45 to 90 minutes of skilled labor per proposal. For a company generating 200 leads per month, that is a massive operational drag.

AI is collapsing this entire workflow into something that takes minutes. Satellite imagery analysis replaces manual roof measurement. Machine learning models handle panel placement and shading. Automated engines pull utility rates and generate financial projections instantly. The result is that a homeowner can go from filling out a web form to reviewing a detailed, accurate proposal in under 10 minutes, with no human involvement. Companies that have adopted AI-driven sales workflows report close rate improvements of 25 to 40% and cost-per-acquisition reductions of 30 to 50%. This is not a marginal improvement. It is a structural competitive advantage that will separate the winners from everyone else in residential solar over the next three years.

Every solar proposal starts with the roof. You need to know the total usable area, the pitch and azimuth of each roof face, and the location of obstructions like vents, chimneys, skylights, and HVAC equipment. Traditionally, this required either a physical site visit ($150 to $300 per visit when you factor in drive time, technician wages, and vehicle costs) or manual analysis of aerial photos by a trained designer. Both approaches are slow and expensive, especially when you consider that 40 to 60% of leads never convert, so all that upfront work on non-buyers is pure waste.

Google's Solar API (Project Sunroof) changed the game when it became commercially available. The API provides roof segment geometry, annual solar irradiance values, and shading data for over 320 million rooftops across the United States, parts of Europe, and other markets. It uses a combination of aerial imagery and LIDAR-derived 3D models to calculate hour-by-hour solar flux on each roof segment, accounting for terrain, nearby buildings, and vegetation. For solar installers, this means you can programmatically retrieve a building's solar potential before a customer even finishes filling out your lead form. The API returns roof pitch, azimuth, usable area, and monthly/annual energy generation potential for multiple panel configurations.

EagleView (formerly Pictometry) takes a different approach. Their TrueDesign product uses high-resolution oblique aerial imagery captured by their proprietary fleet of aircraft. The oblique angles give you perspective views of each roof face, which are significantly more useful than straight-down (nadir) satellite images for identifying obstructions and measuring pitch. EagleView reports are typically ordered per-property at $15 to $40 each, and delivery takes 1 to 3 hours for addresses in their coverage area. Many large installers (Sunrun, SunPower, Tesla Energy) have bulk agreements that bring per-report costs under $10.

Aurora Solar combines satellite imagery with AI-powered LIDAR analysis in their design platform. Their SmartRoof technology automatically detects roof edges, planes, and obstructions from satellite imagery, builds a 3D model, and calculates shading hour-by-hour using a digital surface model (DSM) derived from LIDAR data where available. Aurora claims their automated roof modeling is accurate to within 3% of manual on-site measurements. For installers, the practical benefit is that a complete roof assessment that used to take 30 to 60 minutes of manual work now happens in under 60 seconds.

The accuracy question is important. AI-based roof assessments are not perfect. Trees grow, homeowners add structures, and satellite imagery can be 6 to 18 months old. The best practice is to use AI for the initial proposal (the "speed to lead" moment) and then validate with a site survey before finalizing the contract. This hybrid approach gives you the speed advantage for lead conversion while maintaining the accuracy needed for installation. Some companies are now using drone imagery captured by the sales rep during the first visit to update the 3D model in real time, combining the best of both approaches.

Once you have the roof model, the next step is designing the actual solar array. This means deciding how many panels to install, where to place them, what equipment to use, and how to handle shading. Manual system design by a trained solar engineer takes 20 to 45 minutes per project. AI reduces this to seconds, and in many cases produces better designs than humans because it can evaluate millions of layout permutations that no person would have time to test.

Panel placement optimization is fundamentally a constrained optimization problem. The AI must maximize energy production while respecting fire code setback requirements (typically 18 inches from ridgeline and 36 inches from eaves in IFC-compliant jurisdictions), maintain required spacing for maintenance access, avoid obstructions with appropriate buffers, and comply with local building codes that vary by municipality. Aurora Solar, Scanifly, and OpenSolar all offer AI-powered auto-design features that solve this optimization in real time. Aurora's system evaluates hundreds of possible layouts per roof segment and selects the configuration that maximizes annual kWh production per dollar of installed cost.

Shading analysis is where AI delivers its most significant accuracy improvements. Traditional shade analysis used simple geometric projections based on obstruction height and compass bearing. Modern AI-based tools use hour-by-hour sun path calculations combined with 3D models of the surrounding environment (trees, buildings, terrain) to compute shade impact on each individual panel position for every hour of the year. This granular approach matters enormously because even partial shading on a single panel in a string inverter system can reduce the output of the entire string by 20 to 40%. AI-powered shade analysis helps designers make informed decisions about panel placement and equipment selection (microinverters vs. string inverters with optimizers) based on actual shading conditions rather than rules of thumb.

Equipment selection is another area where AI adds value. The model needs to match panel wattage and electrical characteristics with inverter specifications, determine stringing configurations that stay within inverter voltage and current limits across temperature extremes, and optimize the balance between system cost and performance. A 7.5 kW residential system might use twenty 375W panels with a single string inverter, or it might use eighteen 420W panels with microinverters. The "right" answer depends on shading conditions, roof geometry, local utility net metering rules, and the customer's budget. AI design engines evaluate these tradeoffs programmatically and recommend the configuration that maximizes the customer's financial return.

The cost of AI design tools varies considerably. Aurora Solar charges $150 to $300 per user per month depending on the plan. OpenSolar offers a free tier with premium features at $200+ per month. Scanifly, which focuses on drone-based 3D modeling, runs $199 to $499 per month. For a mid-sized installer producing 50 to 100 proposals per month, the per-proposal cost of AI design software is $2 to $6, compared to $30 to $75 for manual design labor. The math is not even close.

A solar proposal is a sales document, not a technical document. The homeowner does not care about string sizing or inverter clipping ratios. They care about three things: how much will it cost, how much will I save, and how do I pay for it. AI-powered proposal engines take the technical design outputs and translate them into clear, compelling financial stories that close deals.

Instant proposal generation is now table stakes for competitive solar companies. Platforms like Aurora Solar, Enerflo, and SolarNexus can generate a branded, customer-ready proposal in under 30 seconds once the system design is complete. The proposal includes a 3D rendering of the panels on the customer's actual roof (which is psychologically powerful for conversion), estimated monthly and annual energy production, a 25-year savings projection, and side-by-side comparisons of financing options. The entire flow from lead form submission to proposal delivery can be fully automated, meaning the customer receives their proposal via email or SMS within minutes of expressing interest.

Financing integration is critical because 70 to 80% of residential solar is financed rather than purchased outright. The major solar loan providers (Mosaic, GoodLeap, Sunlight Financial, Dividend Finance) all offer API integrations that allow proposal tools to pull real-time rates and generate accurate monthly payment estimates without the sales rep manually entering data into a separate loan portal. A typical residential solar loan today runs 4.99% to 8.99% APR for 10 to 25 year terms, and the monthly payment needs to be clearly lower than the customer's current electricity bill to make the "solar saves money from day one" pitch work. AI-driven proposals dynamically select the financing product and term that creates the most favorable comparison against the customer's actual utility bill.

Lease and PPA (Power Purchase Agreement) modeling adds another layer of complexity. With a PPA, the customer pays a per-kWh rate for the solar electricity rather than buying the system. The proposal needs to show year-over-year savings compared to the customer's projected utility rate, accounting for the PPA escalator (typically 1 to 3% annually) and the utility's historical rate increase trajectory (3 to 5% annually in most U.S. markets). AI models that incorporate utility rate forecast data from sources like Genability (now part of Arcadia) or UtilityAPI produce significantly more accurate and credible savings projections than static spreadsheet models.

The proposals that convert at the highest rates share a few common traits. They are visually clean and mobile-optimized (over 60% of homeowners first view proposals on their phone). They include a personalized video or animation showing the panels on the customer's roof. They present no more than three financing options to avoid decision paralysis. And they include a clear, one-click e-signature workflow that lets the customer commit immediately while motivation is high. Companies like those building custom solar energy applications are increasingly embedding these proposal experiences directly into their own branded apps rather than relying on third-party proposal platforms.

The credibility of every solar proposal rests on the accuracy of its energy production estimate. If you tell a homeowner their system will produce 10,500 kWh per year and it actually produces 8,800 kWh, you have a customer satisfaction disaster, a potential warranty claim, and a negative review that poisons your lead pipeline. Conversely, if you underestimate production to be "safe," your savings projections look less compelling and you lose deals to competitors with more aggressive (but potentially dishonest) estimates. AI-based production models thread this needle by delivering accuracy within 3 to 5% of actual production, compared to 8 to 15% error rates from simpler tools.

Irradiance modeling is the starting point. The amount of solar energy hitting a surface depends on latitude, time of year, panel tilt and orientation, and atmospheric conditions. The NREL PVWatts calculator, which has been the industry standard for years, uses Typical Meteorological Year (TMY) data to estimate irradiance. The problem with TMY data is that it represents historical averages, not actual recent conditions. If a region has experienced increasing cloud cover or wildfire smoke over the past decade, TMY data will overestimate production. AI models trained on 10+ years of satellite-derived irradiance data from sources like SolarAnywhere (Clean Power Research) or Solcast capture these trends and produce location-specific forecasts that reflect current climate patterns rather than 30-year averages.

Weather data integration goes beyond irradiance. Temperature directly affects panel efficiency. Crystalline silicon panels lose roughly 0.3 to 0.5% of their rated output for every degree Celsius above 25C. In Phoenix, Arizona, where rooftop temperatures regularly exceed 65C in summer, a panel rated at 400W might only produce 320W during peak afternoon hours. AI production models that incorporate hourly temperature profiles from NOAA or Weather Company data produce meaningfully different annual estimates than models that use simple monthly averages. Snow cover, soiling (dust and pollen accumulation), and humidity also affect output, and sophisticated models account for all of these.

Panel degradation modeling is essential for long-term savings projections. Solar panels degrade over time, typically losing 0.4 to 0.7% of their output per year. But degradation is not linear or uniform. Panels degrade faster in their first year (light-induced degradation, or LID), then settle into a steadier decline. Higher temperatures accelerate degradation. Cheaper panels from tier-2 manufacturers may degrade 50% faster than premium panels from companies like REC, LG (before their exit), or SunPower. AI models that use manufacturer-specific degradation curves and adjust for local climate conditions produce 25-year production forecasts that are significantly more accurate than the "0.5% per year flat rate" assumption most installers use.

For companies building their own production estimation engines, the technical stack typically combines PVLib (an open-source Python library maintained by Sandia National Laboratories) for the core photovoltaic modeling, satellite-derived irradiance data via API from SolarAnywhere or Solcast ($0.01 to $0.05 per location query), weather data from NOAA or commercial providers, and custom machine learning models trained on actual production data from monitored installations. If you are building a renewable energy dashboard, the production estimation engine is one of the most technically challenging and highest-value components to get right.

AI does not just accelerate the proposal itself. It transforms the entire sales operation around it, from lead qualification through permitting and interconnection. Solar companies that integrate AI across the full workflow see compound improvements that are much larger than any single point solution delivers in isolation.

AI lead scoring ranks incoming leads by conversion probability before a sales rep ever touches them. The model ingests data points like the lead's zip code (solar adoption rates vary 10x between neighborhoods), home value (Zillow API or county assessor data), roof age and condition (older roofs need replacement before solar, killing the deal), current utility provider and rate plan, whether the home has an HOA (which can complicate approvals), and the lead source (referrals convert at 3 to 4x the rate of paid digital leads). Companies using AI lead scoring report that their sales teams spend 60 to 70% of their time on leads that actually close, compared to 25 to 35% when leads are assigned randomly or by simple round-robin.

CRM integration ensures that AI-generated proposals, design data, and customer interactions flow seamlessly into the sales pipeline. Most solar companies use either Salesforce, HubSpot, or a solar-specific CRM like Sighten (now part of SunPower), Enerflo, or PVBid. The key integration points are automatic proposal creation triggered by lead status changes, real-time sync of customer utility data and consumption history, automated follow-up sequences based on proposal engagement (did they open it, how long did they view each page, did they click the financing section), and pipeline analytics that predict which deals will close and when. These integrations are typically built using REST APIs and webhook-based event systems.

Permit document automation eliminates one of the most tedious bottlenecks in solar installation. After a contract is signed, the installer must submit permit applications to the local building department. Requirements vary by jurisdiction, but typically include a site plan, electrical one-line diagram, structural attachment details, equipment specifications, and various code compliance checklists. Manually preparing these documents takes 2 to 4 hours per project. AI-powered platforms like SolarAPP+ (developed by NREL in partnership with DOE) and permit automation features within Aurora Solar and OpenSolar generate permit-ready document packages automatically from the system design. SolarAPP+ goes further by enabling instant, automated permit approval for code-compliant residential systems, eliminating the 2 to 6 week wait for manual plan review in participating jurisdictions. Over 400 U.S. jurisdictions have adopted SolarAPP+ as of early 2028.

Utility rate analysis powers the savings calculations that close deals. Every solar proposal needs to answer the question: "How much will I save?" Answering that accurately requires pulling the customer's actual rate schedule (which can be extraordinarily complex, especially in states like California with TOU rates, demand charges, and NEM 3.0 export compensation), modeling their consumption pattern against solar production hour-by-hour, and projecting savings over 25 years accounting for rate increases, panel degradation, and net metering policy changes. Services like UtilityAPI and Arcadia's Genability provide programmable access to rate schedule data for over 3,000 U.S. utilities. AI models that incorporate these data sources produce savings estimates that hold up under scrutiny, which builds trust and accelerates the close.

Let's map the traditional solar sales cycle against an AI-powered one to make the impact concrete. In the traditional workflow, a lead comes in (day 1), a sales rep calls to qualify and schedule a site visit (days 2 to 3), the site visit happens (days 4 to 7), the designer creates the system layout and production estimate (days 7 to 10), the proposal is assembled and sent (days 10 to 12), the customer reviews and negotiates (days 12 to 20), and the contract is signed (day 20+). That is a 3 to 4 week cycle at best, with multiple handoffs and plenty of opportunities for the lead to go cold.

In an AI-powered workflow, the lead fills out a web form with their address and a recent electricity bill (minute 0). The system instantly retrieves satellite imagery, builds a 3D roof model, designs the optimal panel layout, calculates shading-adjusted production, pulls the customer's utility rate, and generates a personalized proposal with multiple financing options (minutes 1 to 3). The proposal is delivered via email and SMS with a personalized video walkthrough (minute 5). The customer reviews, asks questions via an AI-powered chat widget, and e-signs the contract (minutes 5 to 60). A sales rep intervenes only if the customer requests a call or the deal value exceeds a threshold requiring human approval. Total elapsed time from lead to signed contract: under one hour for motivated buyers.

This compression has second-order effects that multiply the advantage. Faster proposals mean fewer leads are lost to competitors, so customer acquisition cost drops by 30 to 50%. Automated design eliminates the need for a large team of system designers, reducing overhead by $50K to $80K per designer per year. Fewer site visits (only for contracted projects, not every lead) cut vehicle and labor costs. And higher proposal volume with consistent quality means the company can grow revenue without proportionally growing headcount, which is the definition of operational leverage in a services business.

The companies leading this transformation are a mix of vertically integrated installers and software platforms. Sunrun and Tesla Energy have built proprietary AI sales tools that are core competitive advantages. Aurora Solar, which raised $250M+ in venture funding, has become the dominant independent platform, serving thousands of installers. Younger companies like Scanifly (drone-based 3D modeling), Aerialytic (AI satellite analysis), and Palmetto's technology platform are pushing the boundaries further. For installers building custom technology stacks, the convergence of AI with broader energy management capabilities is creating opportunities to differentiate beyond just the sales process into ongoing customer relationships around energy optimization, battery storage, and EV charging.

If you run a solar installation company or are building software for the solar industry, the practical question is where to invest first. Not every company needs a fully custom AI platform. Most should start by assembling best-in-class tools and integrating them, then build custom capabilities only where off-the-shelf solutions create unacceptable limitations.

Phase 1: Adopt AI design and proposal tools (Month 1 to 2). If you are still doing manual roof assessments and proposals, switching to Aurora Solar or OpenSolar is the single highest-ROI move. Budget $200 to $500 per month per design seat. Train your sales team to use auto-design features rather than manually placing panels. Set up automated proposal delivery so customers get their quote within minutes of the design being completed. This phase alone should increase close rates by 15 to 25% and cut design labor by 70%.

Phase 2: Integrate CRM and lead scoring (Month 2 to 4). Connect your design platform to your CRM using native integrations or middleware like Zapier. Implement basic lead scoring using conversion data from your last 6 to 12 months of proposals. Even a simple scoring model (zip code + home value + lead source) will dramatically improve how your sales team allocates time. Invest $500 to $2,000 per month in CRM tools and integrations.

Phase 3: Automate permitting and utility analysis (Month 3 to 6). Enroll in SolarAPP+ for jurisdictions that support it. Integrate UtilityAPI or Genability for automated rate schedule retrieval. Build or configure automated permit document generation from your design tool. This phase reduces post-sale overhead by 40 to 60% and accelerates time from contract to installation.

Phase 4: Build custom AI capabilities (Month 6+). This is where companies with proprietary data can create lasting competitive advantages. Train custom production estimation models on actual monitored data from your installed base (every system with a monitoring API becomes a training data point). Build predictive lead scoring models that incorporate your specific market dynamics. Develop AI-powered customer communication tools that handle routine questions and scheduling automatically. This phase requires software engineering talent and budgets of $100K to $500K+, but the companies that make this investment become very difficult to compete against.

The solar industry is at an inflection point. Installations are growing 20 to 30% annually in the U.S., but customer acquisition costs have not declined at the same pace because most companies are still running manual, labor-intensive sales processes. AI is the lever that breaks this dynamic. The installers and software companies that aggressively adopt AI across their sales and operations workflows will capture disproportionate market share over the next three to five years. Those that wait will find themselves competing on price in an increasingly commoditized market with structurally higher costs.

If you are ready to build AI-powered solar sales tools, whether that is a custom proposal engine, an intelligent CRM integration, or a full-stack platform that differentiates your installation business, we can help you architect and ship it. Book a free strategy call to discuss your specific requirements and get a realistic roadmap with timeline and budget estimates.