A site visit, a designer, an engineer and a project manager. Now one person and an address.
Planning a solar installation took four separate teams and up to a month of handoffs before anyone could answer the only question that mattered: is this worth the investment? We built a planner that reads the real site from satellite data and answers it in a single session.
- Teams needed to plan aninstallation
.svg?2026-09-01T09%3A27%3A19.312Z)
.svg?2026-09-01T09%3A27%3A37.428Z)
Teams needed to plan an
installation
- From site to a costed, defensibleproposal
.svg?2026-09-01T10%3A56%3A25.341Z)
.svg?2026-09-01T10%3A54%3A12.520Z)
From site to a costed, defensible
proposal
- Real site modelled from map data
Real site modelled from map data
- Payback and investment case,produced with the layout
.svg?2026-09-01T11%3A14%3A05.200Z)
Payback and investment case,
produced with the layout
A multinational solar manufacturer, selling through its own team and a network.
A multinational manufacturer of solar modules, selling both directly and through a network of EPC contractors and distribution partners who design and install the systems on the ground.
Two very different sales motions sit inside that. Commercial and utility-scale projects go through their own planning team and end in an investment decision made by someone in a boardroom. Residential projects go through installers standing in a customer's driveway.
Same product. Two completely different conversations and both of them stalled at the same place.
Four departments, and a decision waiting on all of them.
Solar is unusual among the products we work with, because performance depends entirely on where it is installed. Roof geometry, orientation, the building next door, the tree that will shade the array every afternoon in October all of it changes what a given panel position will actually produce.
That makes the specification a modelling problem rather than a selection problem. And the modelling was done by four different people, in sequence.Every handoff was a wait. Every wait was a week. And by the time an investment case reached the people who could approve it, the market had frequently moved.
Someone visited the site.
A physical survey, scheduled, travelled to and written up.
A designer rebuilt it in Google Earth.
The site was reconstructed by hand and a panel layout mocked up on top of it.
An engineer calculated the capacity.
Yield worked out from the layout, allowing for shading and orientation as best as could be judged.
A project manager assembled the numbers.
Investment case, payback, return pulled together into something management could actually decide on.
Nobody in that chain was slow. The chain was slow.
We didn't model the products.
We modelled the parts.
The instinctive fix is to make each stage faster better templates for the designer, a spreadsheet for the engineer. That leaves four handoffs in place, which is where the time actually goes.
So we started from the observation that the site already exists in public data. Satellite imagery, map data and building footprints describe the roof, the orientation, the neighbouring structures and the tree canopy everything the survey team was travelling to establish. From an address, the software reconstructs the site and models what each panel position will genuinely produce, accounting for the shadow the building next door casts in the afternoon and the tree that will be in leaf by June.
One person, one address, one session and every stage that used to be a separate department is now a step in the same tool.
The site visit was gathering information that was already available. It just wasn't in a form anyone could compute with.
One model. Layout, yield, cost and the investment case.
That last output is what separates this from a design tool. A layout is an engineering artefact. A payback period is a decision.
- Address
- Site model
- Shading analysis
- Panel layout
- Yield
- BOM
- Quote
- ROI
The panel layout.
Positions determined against the real geometry of the roof or the field, not an idealised rectangle.
The yield estimate.
What the array will actually produce in that location, with shading from surrounding buildings and vegetation accounted for rather than assumed away.
The bill of materials.
Down to panels, inverters and mounting hardware the complete parts list, from the same model that produced the layout.
The quote and the customer proposal.
Costed and presentable, produced with the design rather than after it.
And the answer management was waiting for.
Return, payback period and investment case the question the other four teams were only ever assembling an answer to.
The engineering travelled from one deployment to the other. The commercial consequence did not: the field planner removed a bottleneck inside the company, while the EPC product removed one across an entire network of partners.
Then the same problem appeared in the driveway.
The field planner solved the commercial and utility side, where the customer is an organisation and the output is an investment case. Residential is a different sale entirely. The customer is a homeowner, the decision is made in one or two conversations, and the person in front of them is not an employee of the manufacturer they are an EPC contractor or an installation partner. So we built a separate product for the distribution network, on the same underlying engine. An installer enters the customer's address. The software models that specific house the roof, its orientation, the neighbours' buildings, the trees in the garden and produces a shaded, costed, ROI-backed proposal while the installer is still on site.
The manufacturer keeps control of the products, the rules and the pricing. The channel gets to close in one visit.
When the site decides the product, the site has to be in the software.
Most configuration problems are internal a product range, a set of rules, a pricing structure. This one is not. Half the variables live outside the business entirely, in the physical world where the product will be installed. That is a different class of problem, and it is why no off-the-shelf configurator solves it. There was nothing to buy that could look at a location and reason about what a building next door would do to a panel in October. If your product's performance depends on where it ends up solar, but also facades, HVAC, acoustics, structural systems then a specification tool that ignores the site is only doing half the job.
Does your product depend on where it's installed?
If specifying it correctly means somebody visiting the site first, that visit can often be replaced by data that already exists.
Thirty minutes. Show us how you plan one installation today.