B2B solar prospecting: tools, players, and real upside

B2B solar prospecting used to be an address-book job. Buy a file, canvas a region, dial. In 2026 the bottleneck is not volume. It is time wasted on roofs that should never have entered the pipe.

A C&I rep who walks into a warehouse without data discovers a roof that is too small, a tenant with no mandate, or a load profile that does not fit. Half a day gone. Across a team, that is a quarter of ghost pipeline.

The upside is not another tool. It is a chain: see the territory, score the sites, reach the right decision-maker, and only send engineering to what deserves it. Sometimes that chain already exists in market software. Sometimes you have to build it. An AI audit is how you decide.

A commercial PV project has several players. If the stack ignores one of them, data gets stuck.

On the sell side, the C&I installer or developer owns the offer. Sales opens the meeting. Engineering validates roof, shading, structure. Marketing feeds the site, assets, and campaigns. Ops or data keeps the CRM clean. Without shared rules, everyone prioritizes differently.

On the buy side, the decision-maker is almost never "the company". It is a site director, a CFO, an asset manager, a landlord, sometimes an energy consultant. Enrichment that returns a generic switchboard number is useless. You need the right role, at the right time.

Software vendors (solar mapping, B2B enrichment, CRM, outreach sequences) and integrators, including an agency like Radianz, sit in the middle. Their job is not to replace everyone. It is to move information between these people, without a parallel spreadsheet.

Four building blocks show up in stacks that last.

The first is reading the territory. Satellite imagery, cadastre, solar potential APIs, sometimes computer vision on roofs and parking lots. That is the work we pushed on panel and surface detection: prioritize sites before anyone drives out. Specialized products exist. Google Solar, C&I GeoAI, cadastral layers. The useful reflex is to test what already covers 80% of the need before training a house model.

The second is enrichment. Company IDs, industry codes, headcount, HQ, contacts. Local registries, B2B databases, LinkedIn. B2B data decays fast: about 2% a month. Without refresh, the CRM becomes an email graveyard.

The third is the system of record. HubSpot, Salesforce, Pipedrive, or a vertical CRM. That is where the A / B / C score must live, not in the most organized rep's spreadsheet.

The fourth is activation. Email sequences, LinkedIn, follow-ups, booking. Native CRM sequences, outreach tools, a calendar like Cal.com. A conversion chatbot on the website can qualify a visitor after hours, as long as the conversation context lands in the same CRM.

These blocks do not talk to each other on their own. The useful work is the thread: scored roof, found contact, created record, sent message, booked meeting, updated status.

These loops are realistic. We have seen versions of them on C&I work.

Scan a territory (logistics zone, retail park, ZIP) and output a list of roofs with estimated surface, shading, and a kWp range. Reps stop opening a map app building by building.

Enrich the owner or occupier, match the right person, and only write if the role fits (facilities, finance, site leadership).

Score automatically: surface, distance from the depot, sector, regulatory signal, owner-occupier. An A gets contacted within 48 hours. A B enters a sequence. A C stays on watch.

Generate the first message from the site, not the company. "Your X m² platform in Y, power range, savings hypothesis." Detailed engineering comes after the hook.

Watch press and tenders on a theme (projects, players, zoning) to be first to call. That is the logic behind our article analysis agent.

Qualify inbound from the website: a visitor asking for a sizing should not fill a seven-field form. A chatbot asks one question at a time, books a slot, and leaves the thread for sales.

None of these loops require you to code everything. Many hold with a CRM, a solar API, an enrichment tool, and a thin glue layer. You build only what does not exist, or what truly differentiates the offer. More patterns sit on the industries page.

Teams that filter before fieldwork see three effects.

Time spent on deals that do not convert drops. If a rep still burns half a day on a C account, the upstream filter is too loose.

Time from qualification to first contact shrinks. On A accounts, aim for under 48 hours. The team that arrives with a numbered range before the competitor changes the conversation.

Meeting rates by segment (logistics, retail, offices) become comparable. Without a shared grid, you are comparing apples and warehouses.

The upside for leadership is not "we have AI". It is a shorter, readable pipe, and engineers who stop drawing dead roofs. For sales, less research, more meetings. For marketing, a site that captures intent instead of dropping it. For ops, one source of truth.

Lead-list vendors often sell the same file to several installers. The edge is not the list. It is activation speed and the quality of the first touch.

Not every team needs a homemade platform. If the process is standard (enrich, score, sequence), a well-configured existing product wins. The common failure is buying the tool and never training: the CRM fills with junk, sequences stay generic, the score means something different for every rep.

The useful sequence looks like this. First, say what you automate and what you leave. Then pick existing software when it covers the need. Train for clean usage: required fields, A/B/C levels, scripts per segment. Build only the missing link (vision, owner matching, monitoring agent, a bridge between two systems).

That is Radianz today: an AI and web agency. We do not stack tools for sport. We decide, we wire, we train, and we industrialize what does not exist yet. If you want to frame this chain on your territory, book a slot.