A homeowner who buys solar panels in March often first visited your website the previous October. They read a page about batteries, left, came back in December to look at your warranty page, read something on a forum, came back again in February with a spreadsheet of three quotes, and finally filled in your survey form on a Sunday night.
Solar is a long, considered purchase. It involves the roof, a large sum of money, and a decision the household will live with for twenty years or more. Pressuring someone in October to "book your free survey now" does not speed that up. It just annoys a person who is still working out whether solar makes sense at all.
An AI receptionist for solar installers is useful precisely because it does not get bored of early questions. It can answer the same "do panels work on cloudy days?" question for the hundredth time, gather suitability details when the visitor is ready, and never, ever promise a savings figure.
Match the conversation to the stage
Visitors arrive at very different points. The assistant's job is to recognise roughly where they are and respond in kind.
| Stage | What they ask | What helps them | What to capture |
|---|---|---|---|
| Early research | Do panels work in winter? Do I need a battery? | Plain explanations from your content | Optional email for a guide or follow-up |
| Checking suitability | Is my roof OK? I have a chimney on the south side | Explaining what matters and why | Roof direction, shading, roof type, ownership |
| Comparing options | What is the difference between these panels? What warranty do you give? | Your products, warranties and process | Energy use, budget band, timeline |
| Ready to quote | Can someone come and look? | A clear survey request path | Full contact details, address, preferred days |
The mistake many installers make is treating everyone as the bottom row. The early-research visitor who gets an honest, useful answer and no pressure is the one who remembers your name in February.
The guide on how solar installers qualify leads with a chatbot covers the bottom two rows in more depth. This post is mostly about the top two.
The savings question, handled properly
Every solar conversation reaches "how much will I save?" eventually. It is a fair question. It is also the one where an assistant can do real damage, because a number given in a chat becomes a number the customer remembers and quotes back to you.
Write a firm rule: no savings figures, no payback periods, no export income estimates, no percentages. Then give the assistant something good to say instead.
Visitor: How much will I save a year with a 4kW system?
Assistant: It depends on a few things we would need to check, so I will not guess a figure. The main ones are which way your roof faces and how much shade it gets, how much electricity you use and when, whether you add a battery, and your tariff for buying and selling power. Our surveyors work those out for your home and show the assumptions behind any estimate. If you tell me your roof direction and roughly what your annual electricity bill is, I can pass that on so the estimate starts from real numbers.
That answer respects the question, explains the variables and moves the conversation towards useful data. It also protects you. For more on the phrases to keep out of your assistant entirely, see what an AI receptionist must never promise.
Suitability details worth gathering early
When a visitor moves into the second stage, a few answers tell your team a lot about whether a survey will be worth the drive:
- Roof orientation. Roughly which way the main roof slopes face. Most people can tell you from where the sun rises and sets.
- Shading. Nearby trees, taller buildings or chimneys that shade the roof for part of the day.
- Roof type and condition. Tiles, slate, flat roof, and whether it is due for replacement soon.
- Ownership and permissions. Owner-occupier, landlord, leaseholder, and whether the property is listed or in a conservation area, which may need checking with the local planning authority.
- Electricity use. An approximate annual bill or kWh figure, and whether they work from home or have an electric car.
- Timeline. Just researching, within six months, or as soon as possible.
The assistant should explain why it is asking. "Shading matters because even part of a panel in shade can reduce what the whole system produces" is more persuasive than a bare form field.
Keeping months-long leads warm
This is where the long cycle becomes an advantage, if you set it up.
Here is an example to adapt. Say your site captures 50 solar leads a month, and your team only has time to properly call the 15 who say they want a quote soon.
- Switch on lead capture and ask each visitor about timeline in plain words.
- Use a webhook to send every lead to your CRM with that timeline attached, so "within six months" and "just researching" go into different follow-up lists.
- Call the ready-now group within a working day.
- Send the researchers useful, non-pushy updates: how surveys work, what to check on their roof, what a quote should include.
- Review the assistant's unanswered questions each month and add content for the gaps. Research-stage questions are your best guide to what your website is missing.
In this example, the 35 leads not ready now are no longer lost. They are in a list you speak to occasionally, which is how solar actually gets sold. The post on assistants for B2B and long sales cycles covers the same principle for other slow-decision businesses.
A word on grants and schemes
Funding schemes, tariffs and rules change. If you want the assistant to mention them, keep a single dated page with the current position and a pointer to the official source, and update it when things change. Out-of-date funding claims are worse than none, and the assistant can only be as current as the content you give it.
Where to begin
Pull the questions your sales team answered by email last quarter and sort them into the four stages above. Write short, honest answers for the first two stages, and add your savings rule word for word. Load your product and warranty sheets as documents.
Then test it with the hardest question you get, the savings one, asked five different ways. When it refuses a number every time and still sounds helpful, put it live. The solar installers industry page has a demo built this way.