AI Marketing for Solar Companies

No industry has a worse generative search problem than solar right now, and it is not subtle. The federal residential credit under Section 25D ended for expenditures after 2025, but the internet is saturated with content describing it as current, and AI tools trained and grounded on that material repeat it confidently. The result is a specific and expensive sales problem: a homeowner asks whether solar is worth it, receives an answer built on a thirty percent credit and a payback period calculated with it, and arrives at your consultation with expectations no honest proposal can meet. Your salesperson then spends the appointment explaining that the internet is wrong, which is a terrible position from which to sell a thirty thousand dollar system.

1The Expired Credit Problem

• The training data problem is structural. Years of content described the credit as available and a few months of content describes it as gone, which is an unbalanced corpus.
• Grounded search results do not fix it reliably. Because the pages that rank on incentive queries are frequently the stale ones.
• The error is consequential. A homeowner planning around a credit worth thousands is going to be disappointed by every proposal they receive.
• Publishing the correction is the fix. Clear, dated, jurisdiction-named content stating the current position, which is what these systems can draw on.
• Cover what replaced it. State and utility programmes, and how third-party ownership is treated, described accurately and with a pointer to official sources.
• Do not become the tax authority. Describe programmes and direct people to the programme itself and their own advisor.
• This is a window. The correction advantage lasts until competitors update, which most will do slowly.

2Payback and Savings Answers

• Payback answers inherit the credit error. A period calculated with a thirty percent reduction is substantially shorter than reality, which sets an expectation your proposal cannot meet.
• National averages ignore the variables that matter. Local rate structure, consumption, roof orientation, and shading determine the outcome and none appear in a chat question.
• Rate escalation assumptions are usually invisible. And they are where optimistic projections are built.
• Publish the arithmetic rather than a number. Content showing how payback is calculated, with assumptions named, is more useful and more citable than a figure.
• Give ranges tied to conditions. If consumption is this and the rate structure is that, then the range is this, which these systems handle well.
• Brief your sales team on what prospects believe. A salesperson expecting the wrong number can address it calmly rather than appearing to contradict a trusted source.
• Keep your own published figures current. A stale example on your site becomes the wrong answer these tools then repeat with your name on it.

3Net Metering and State Variation

• Rules differ by state and by utility. Which general answers flatten into a description of net metering that may not apply anywhere near your customer.
• Successor tariffs are poorly represented. Where net metering was replaced with a different export compensation structure, generic answers frequently describe the old regime.
• Grandfathering is widely misunderstood. Existing customers under prior rules versus new ones is a distinction these tools routinely miss.
• Export value changes system design. Which is why an out-of-date answer produces expectations about system size and battery need that do not match your proposal.
• Name the utility and the state explicitly. The strongest extraction signal available on a topic this locally variable.
• Date everything and review quarterly. Utility tariff changes are frequent and consequential.
• Cover time-of-use plainly. Because it determines value more than total production does in many markets and is rarely explained well.

4Lease Versus Buy Advice

This question has become more important as the market shifts toward third-party ownership, and it is one where generic advice can genuinely mislead.

• The right answer changed this year. With the residential credit gone, the comparison between owning and third-party arrangements shifted materially, and older content reflects the old calculus.
• Tax treatment differs by structure. Which is central to the comparison and is precisely the kind of detail that gets flattened in a summary. Point people to their own tax advisor.
• Escalators are frequently omitted. A payment that rises annually changes a long-term comparison substantially.
• Resale implications rarely appear. Transferring an agreement at sale is a real consideration these answers usually skip.
• Publish a genuine comparison. Conditional, with the variables named, which is both more accurate and more citable.
• Say when each suits whom. Including cases favouring the model you earn less on, which is what makes the content credible enough to be cited.
• Avoid promotional framing. Comparison content that always concludes in favour of what you sell is recognized and discounted.

5Choosing an Installer

• The advice these tools give is generally sound. Check licensing, look at years in business, verify warranty terms, be wary of pressure, and get multiple quotes.
• All of which favours an established local installer. If that is you, the generic answer is describing your positioning.
• Longevity guidance is particularly prominent. Because installer failures are well documented, and these tools reflect that.
• Publish your own thorough version. A guide to evaluating any solar company is exactly the content cited on these queries.
• Include verification instructions. How to check a licence and certification in your state, which is specific and locally anchored.
• Cover the door-to-door situation. How to evaluate a company that knocked, which is a live and frequently asked question.
• Explain what a complete proposal contains. Which makes vague competitor proposals look vague.

6Crawler Access

• Verify the named agents are allowed. GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and Applebot-Extended each check robots.txt separately.
• Solar sites are often built on marketing platforms. Whose defaults were set without this in mind and are worth confirming rather than assuming.
• Check the CDN and firewall layer. Edge rules override robots.txt entirely.
• Google-Extended is separate from search. Blocking it affects AI features only.
• Blocking is actively counterproductive here. Your corrected incentive content is only useful if these systems can reach it, and withholding it leaves the stale answers unchallenged.
• Recheck after site changes. Rebuilds reintroduce blocks routinely.

7Entity Signals for an Installer

1. Licence numbers consistent everywhere. Website, Business Profile, contractor board record, and every directory, matching exactly.
2. Founding year stated repeatedly. Which is both a trust signal and an entity attribute in an industry where longevity is the question.
3. Certifications with the issuing body named. Since certification directories are strong corroborating sources.
4. Service areas and utilities named in plain text. Cities, counties, and the utilities you interconnect with.
5. Services stated explicitly. Installation, battery storage, service and repair, and support for existing systems.
6. Add LocalBusiness and service schema. Plus FAQPage where questions render on the page.
7. Keep details identical across sources. Conflicting records weaken entity recognition directly.

8Content Structured for Extraction

• Lead sections with the question a homeowner would type. Answered immediately below rather than after preamble.
• Date prominently and state the review date. More important in solar than any other category given how fast the position moves.
• Name the state and utility throughout. On anything touching incentives, net metering, or payback.
• Give conditional answers with variables named. Which is both more accurate and more likely to be drawn on than a flat figure.
• Byline to a named person with credentials. Role, certification, and years in the industry.
• Visible FAQ sections with matching schema. The questions your consultants field at every appointment.
• Write passages that survive extraction. If a paragraph only makes sense in sequence, it will be summarized rather than quoted.
• Explicitly state what is no longer available. Content saying a programme ended is what corrects the record.

9Prompt Audits and Sales Enablement

• Build a prompt set across the buying journey. Incentive availability, payback, is-it-worth-it, net metering in your utility, lease versus buy, and installer selection.
• Run monthly, and after any policy change. Across ChatGPT, Perplexity, Gemini, and Google AI Overviews.
• Check incentive accuracy first. The highest error rate and the most direct effect on your appointments.
• Circulate findings to the sales team. This audit is sales enablement in solar more than in any other vertical, because the errors are specific and predictable.
• Build objection handling from it. A consultant who can say what the customer was probably told, and why it is out of date, controls the conversation.
• Record who gets named. Your company, competitors, marketplaces, or nobody.
• Track trends across repeated runs. Since responses vary and one answer is not a signal.

10Measuring AI Visibility

• Incentive accuracy rate in your market. The metric that matters most this year.
• Citation rate across the prompt set. Month over month.
• Referral traffic from AI tools. Modest volume, high intent, visible in analytics.
• Branded search growth. The compounding indicator.
• Prospect incentive expectations logged at appointment. What they believed before arriving, which tells you exactly what to publish.
• Appointments where the credit had to be corrected. A real and currently common sales metric worth tracking.
• Ask at booking. Add an AI option to source capture alongside the usual channels.
• Keep expectations proportionate. A position being established rather than a channel producing installs today.