Can AI Write a Solar Sales Script? Discovery Questions and Objections

Yes. AI can draft a solar sales script with discovery questions and objection handlers in minutes, and the honest version works better than the pushy one. It stays honest only if you supply real inputs, ban guaranteed-savings claims, and read every line against your own proposal before a salesperson says it aloud.

Most search results for "solar sales script" are cold-call scripts built to get a booking at any cost. That is a different job. The harder conversation starts once the customer has agreed to talk: what to ask about the bill, the roof and the money, how to answer the usual doubts truthfully, and how to rehearse both sides with AI before a real household hears a word of it.

Follow me on Instagram@sagnikteaches

What AI is good at here, and what it must not do

A language model is quick at three things: turning your messy notes into an ordered question list, writing several phrasings of the same honest answer, and playing a difficult customer so a new rep can practise. It is poor at anything that depends on facts it cannot see: this household's usage, this roof's shade, this month's tariff, this quarter's incentive rules.

Connect on LinkedInSagnik Bhattacharya

So split the work. You own the facts; the model owns the wording. Put that into the instructions you give it:

Subscribe on YouTube@sagnikteaches

• Never state a savings figure, payback period, production estimate or incentive amount. Use a placeholder such as [savings from proposal].
• Never claim a system "pays for itself", is "free" or comes with "no cost" unless the finance wording says exactly that.
• Never invent urgency. No countdown language, no "only today", no claims about rising prices unless the rep can show a source.
• Always invite the customer to compare quotes and read the proposal alone.

Those rules matter beyond manners. Consumer-advertising and sales regulators in many places pursue misleading solar savings claims, and a customer who feels misled cancels, complains or posts a review. Check the rules that apply where you sell; this tutorial is general guidance, not legal advice.

Discovery: the questions that decide whether solar fits

Discovery is not small talk before the pitch. It is where you find out whether solar suits the household, and a salesperson who skips it quotes the wrong system. Group the questions around the three things that decide fit: the bill, the roof and the finance.

Bill questions

The electricity bill tells you how much a system can offset. Ask for the last twelve months if possible, because a single summer bill misleads.

• "What did you pay in your highest and lowest months last year?"
• "When are you home during the day? Is anyone working from home?"
• "Are you planning anything that changes usage: an electric vehicle, a heat pump, a pool, an extension?"
• "Is your tariff the same all day, or does it change by time?"
• "Have you got the annual usage figure in kilowatt-hours on your bill or online account?"

Why each one matters: daytime presence changes how much generation the home uses directly; a planned car or heat pump changes the size you should consider; a time-of-use tariff changes the value of each unit.

Roof questions

Roof facts decide whether the system is possible, how big it can be, and what extra work is needed.

• "Which way do the main roof faces point?"
• "How old is the roof covering, and has anyone said it needs work in the next ten years?"
• "Do trees, chimneys or neighbouring buildings shade it at any time of day?"
• "Is the roof owned by you outright, or is anyone else involved, such as a landlord or a managing company?"
• "Do you know where the main electrical panel is and roughly how old it is?"

A roof near the end of its life is a conversation to have honestly. Fitting panels on a covering that will need replacing in three years means removing and refitting them later, and a customer who finds that out afterwards will feel misled.

Finance questions

Ask about money without assuming. Customers differ widely: some pay cash, some want a loan, some want a lease or a power-purchase arrangement, and some are not ready to decide.

• "Would you prefer to pay upfront, finance it, or are you still working out options?"
• "What monthly figure would feel comfortable if you financed it?"
• "Do you plan to stay in the home for the next ten years or more?"
• "Is there anyone else who would be part of the decision?"

The last question is not a trap. Many solar conversations stall because a partner who was not in the room has doubts nobody heard.

A prompt that drafts the discovery script

This prompt keeps the model on a short leash. Paste in your own question groups, your lead source and your proposal process, and the rules that matter.

You are helping a small residential solar installer write a discovery conversation guide for sales reps. The reps visit homes or call after the customer has already enquired.

Our process: discovery call, site survey, written proposal, 3 days for the customer to review, decision.
Finance options: cash, loan through a partner lender, no lease.

Write:
- A friendly opening (under 60 words) that explains why we ask questions
- 12 discovery questions grouped as: bill, roof, finance, decision-makers
- For each question, one line on why we ask it and one on what answer would change the recommendation
- A closing that offers a site survey date and says the customer will receive a written proposal to read alone

Rules:
- Do not state any savings, payback, production figure or incentive. Use [figure from proposal] where a number would go.
- No urgency, no "limited time", no "sign today" language.
- If the customer's roof, shading or bill suggests solar may not suit them, the guide must tell the rep to say so.
- Plain English, short sentences, no exclamation marks.

The "tell the rep to say so" rule produces one of the most useful lines in the output, a graceful exit: "From what you've told me, a system might not make sense for this roof right now, and I'd rather tell you that than quote you for something that disappoints." A script with an honest no in it tends to earn trust on the yeses.

Honest objection handlers

An objection is a question with an edge on it. A pushy rep tries to defeat it. An honest rep answers it, with a fact the customer can check. Use a four-step shape in the prompt, because it keeps the model from drifting into sales spin:

1. Acknowledge the concern in the customer's own words.
2. Give the honest fact, including the part that does not favour you.
3. Show where the number comes from, such as the proposal, the survey or the tariff sheet.
4. Offer a next step the customer controls.

"It's too expensive"

Weak: "It pays for itself in no time." Better: "It's a significant cost, and I won't pretend otherwise. Your proposal will show the total price, the finance options and the estimate of what you'd save each year, with the assumptions listed. Take it away, compare it with other quotes, and tell me which part doesn't add up."

"What if the company disappears?"

An honest answer names the specifics the customer can verify: how long the company has traded, who stands behind the workmanship warranty, what the manufacturer's product warranty covers and for how long, and what happens to the paperwork if the installer closes. Do not let AI invent warranty terms; paste in your actual warranty wording and have the model rephrase it.

"My neighbour's system doesn't work well"

"That's worth asking about. Output depends on the roof direction, shade and the size of the system, so one home's result doesn't carry across. At the survey we'll check your roof for shade and angle and put the estimate in writing with the assumptions behind it."

"I'll wait for prices or technology to improve"

Do not claim prices are about to rise or that a deadline is looming. Say: "People do wait, and that's a reasonable choice. Here's what waiting costs on your current bills, and here's what would change in your case if you waited. If the proposal still makes sense in six months, I'm happy to update it." Admitting that waiting is legitimate usually reduces the need to defend.

"Will this get rid of my electricity bill?"

This is where over-promising usually starts, so give the model the honest facts to work from. Even a well-sized system rarely takes a bill to zero: households normally keep paying the utility's fixed charges, such as the standing or connection charge, and what you are credited for exported power depends on the customer's tariff and local rules. An honest handler: "It should cut the part of your bill that depends on how much power you buy. The fixed charges stay. Your proposal will show both parts separately, so you can see what changes and what doesn't." Ask the model to rephrase that, never to add a figure.

"What about my roof?"

Answer with the survey: "We check the condition at the survey and tell you in the written report if we think the roof needs work first. If it does, we'll say so and explain why."

Ask the model for three phrasings of each handler, then keep the one that sounds like your team. Mark every line that contains a fact, such as a warranty term or an installation timeline, and verify it against company documents before it enters the script.

A worked example: one discovery call, end to end

This is an illustration, with invented figures. A rep at a small installer speaks to a homeowner who sent an enquiry through the website. The rep asks the bill questions first. The household's highest month last year was $210 and the lowest $95, usage is steady at about 9,000 kilowatt-hours a year, and both adults work outside the home on weekdays. That last detail matters: most of the household's use happens in the evening, so a large system would export much of its generation. The rep makes a note to ask the designer to model how much would be used on site.

On the roof, the main face points south-west, with one chimney that shades the lower corner in late afternoon. The roof covering is about 22 years old. The rep says: "A roof of that age might need attention within the life of the panels. Our surveyor will check and tell you in the report. If it does, we'd suggest dealing with it before fitting anything." The homeowner is surprised to hear a salesperson say that, and says so.

On finance the customer wants to see a loan option and a cash price. The rep says the written proposal will show both, with all costs and the lender's terms, and asks who else will be in the decision. The homeowner mentions a spouse who is away for the week. The rep books the site survey for a day the spouse is home. The close of the call is a survey date, not a signature.

The AI helped before and after the call, not during it. Before, it produced the question order. After, the rep pasted anonymised call notes into a prompt and asked for a follow-up email that restated the customer's three concerns in the customer's words and listed what the survey would check. That email took two minutes to review and send.

Use AI as the awkward customer

Role-play is the strongest use for new reps, because the model never tires and never goes easy. The model should play a customer with a stated personality and a hidden doubt, and then give the rep feedback.

Play a homeowner with a $180 monthly electricity bill who is sceptical about solar because a neighbour had a bad experience. You are polite but guarded. You have one hidden worry: the age of your roof. Do not mention it unless the rep asks.

I am a solar sales rep. We will practise a 6-minute discovery conversation. Reply as the homeowner, one short answer at a time, and wait for my next question.

After 10 exchanges, stop and score me out of 10 on: asking open questions, finding the hidden worry, avoiding unproven claims, and offering a next step the customer controls. Quote one line I should keep and one I should rewrite.

Illustrative feedback from a run might read: "You asked about the roof direction but not its age, so you missed the hidden worry. You said 'this will definitely save you money', which is a promise you cannot keep. Rewrite it as 'the proposal will show an estimate and the assumptions behind it'. Good line to keep: 'what would make you feel comfortable about the decision?'"

Run the same scene three ways: a price-focused customer, a technology-sceptical one, and an eager customer with unrealistic expectations. The third is the one most reps get wrong, because an eager customer invites a rep to over-promise. The objection drills tutorial gives more scenes, and the roofing objection handlers tutorial shows the same method for a neighbouring trade.

Follow-up message: before and after

The conversation does not end when the call does. The follow-up message is where many reps slip into pressure, and where AI saves the most time. Here is an illustrative before and after, using placeholders.

Before (typical rushed draft): "Hi [first name], great speaking today! Prices are going up next month so let's get you booked in this week. Reply YES and I'll lock in your slot."

After (honest version from the model, once told to ban urgency): "Hi [first name], thank you for your time today. You told me your highest bill last year was about $210, that the roof is south-west facing with some afternoon shade from the chimney, and that you would like to see both a cash and a loan option. Your site survey is booked for [date], when [spouse or partner] will be home. The surveyor will check the roof covering and the shading, and the written proposal will follow with the assumptions listed. If anything changes before then, reply here. [your name]"

Notice what changed. The second message repeats the customer's own facts, so they can correct anything wrong, commits to a date they chose, and asks for nothing more. Use this prompt line to get it: "Write a follow-up under 120 words that restates the customer's three stated concerns, confirms the next step and date, and contains no urgency or sales pressure."

A second illustration is the polite exit. When a survey finds a shaded roof, the model can draft: "Thank you for letting us visit. The survey shows heavy shade from the trees on the south side for most of the day, so the system we could fit would produce much less than the size you hoped for. I would rather tell you now than quote you for something that disappoints. If the trees are ever trimmed, we are happy to look again." A customer who receives that is likely to recommend you even though they did not buy.

A claims checklist to run on every AI draft

Before a script reaches a rep, run it through these eight checks. Any hit means the line gets rewritten or removed:

1. Guarantee words: "guaranteed", "will save", "will pay for itself", "risk-free", "zero cost".
2. Unsourced numbers: any percentage, payback period or savings figure that does not trace to this customer's proposal.
3. Incentive claims: any mention of grants, credits or rebates. They change and depend on the customer's situation. The rep should say "the proposal lists what applies, and you should check eligibility with an adviser".
4. Fake scarcity: deadlines, "last slot", "price rise on Friday" with no real basis.
5. Government or utility implication: anything that suggests the company is linked to the utility or the authorities when it is not.
6. Comparison claims: "better than every other installer" or claims about named competitors.
7. Pressure to decide alone: language that discourages the customer from consulting a partner, an adviser or other quotes.
8. Hidden cancellation rights: in many places a contract agreed during a visit to the customer's home carries a legal cancellation period. The script should mention it plainly where it applies, never play it down or rush the signature to get past it. Check the rule where you sell.

Keep this list pinned beside the script. When a line fails, do not fix it by feel: tell the model exactly which check it failed and ask for a corrected version, then re-check.

From script to daily practice

A script on a shared drive does nothing. Four habits make it part of the job:

1. Short weekly practice. Ten minutes of role-play at the start of a team meeting, each rep taking a different customer type.
2. A shared objection log. When a customer raises something new, a rep adds it with the honest answer. Ask the model to turn the log entries into handlers in the same four-step shape.
3. Lead context before the call. If your website has a chatbot or form collecting the bill, roof direction and shade, the rep starts with facts. The chatbot tutorial covers that, and the lead cost tutorial shows why speed to first reply matters.
4. Review of real calls. If your calls are recorded with consent, pick one a week and compare it with the script. Anonymise before pasting any transcript anywhere.

Measure the right things. Count surveys booked per conversation, proposals accepted after the review period, and cancellations. A script that raises signatures this month and cancellations next month is not working. If you would like help wiring this into your lead handling, that is the kind of job my AI implementation consultation covers.

Where the output tends to go wrong

Four patterns appear again and again, and each has an easy fix.

The model sounds like a cold caller. Because most solar scripts online are cold-call scripts, the model drifts back to "Hi, I'm calling about your electricity bill". Say in the prompt that the customer enquired and that the rep is visiting or calling by appointment.

Figures appear from nowhere. Ask it to list every number it used. Delete any it cannot source to your documents. A sample bad line: "Most homeowners save 40 to 60 percent". There is no basis for it in the prompt, so it must go.

Objections get defeated, not answered. If a handler ends with "so you'd agree it's worth it, right?", rewrite. A tie-down question at the end of every answer is the pressure tactic you set out to avoid.

Tone drifts upbeat. Solar conversations involve large sums. Ask for a calm, specific tone and no exclamation marks.

Finally, keep your own ear in the loop. The best check is a colleague reading the script aloud to a friend who knows nothing about solar and asking, "Would you feel looked after?"

Questions solar sales teams ask about AI scripts

Should a solar script be read word for word?

No. Use the script as a map of questions and a bank of honest answers, then talk naturally. Customers hear a read script immediately. Rehearse until you know the order of the questions and the facts behind each answer, and leave the exact phrasing to the conversation.

Can AI keep my savings claims accurate?

Only if you supply the inputs. AI does not know a household's usage, roof or tariff, so it will guess. Keep every savings figure in your design tool or proposal, and have the script say where each number comes from rather than quoting generic ones.

Is it safe to put customer bills into ChatGPT?

Avoid it unless your plan and policy allow it. For practice and script drafting, use invented bills with realistic shapes. If you do upload real documents, use a business plan with training off by default and strip names and account numbers first.

How do I stop the script sounding pushy?

Ban urgency tricks in the prompt: no fake deadlines, no pressure to sign today, no vague 'limited offer' lines. End each conversation with a clear next step the customer controls, such as a site survey date or a written proposal they can read alone.

Further reads

• Solar Proposal Software: Adding AI-Written Cover Letters and FAQs — Cover letters and FAQs for the proposal the script points to.
• Solar Battery Quote Template: Options Explained With AI, No Backup Promises — Explain battery options honestly, with no backup promises.
• AI for Solar Installation: Five Jobs a Small Installer Can Automate — Five installer jobs worth automating after the script.
• Solar Installer Software: CRM, Design and Tracking, and Where AI Helps — Where call notes and proposals live once leads grow.
• Solar Site Survey Checklist: Photos and Notes Sorted Into a Report With AI — What to capture once the customer says yes to a survey.
• Solar Installer Marketing: Local Pages and Case Studies Written With AI — Fact-sheet templates, prompts and claim checks for AI-drafted town pages and install case studies a solar installer can publish with confidence.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: a government consumer guide to buying home solar (fixed utility charges, comparing bids, pressure tactics); a government guide to cancelling sales made at home; general sales-training practice. Check current rules on solar claims where you sell.