Somewhere in your job history are your most persuasive marketing assets, and they are currently invisible. Every system you have installed is a story with numbers in it: a real address, a real bill before and after, real panels on a real roof that a neighbour can drive past. Installers sell on these stories constantly in person, then publish websites that say quality installations, competitive prices, and wonder why the website does not sell the way they do.
The case for fixing this stopped being a content nicety in 2026. When we measured how AI assistants answer Australian commercial solar queries, installer websites took 97 per cent of ChatGPT’s citations, and the pages being cited were overwhelmingly project and capability pages, because in commercial solar they are often the only substantive content that exists. Only 23 installer domains nationally were being named at all. Case studies are no longer just conversion assets. They are the raw material AI answers are built from, in the thinnest, most winnable field in solar marketing. This guide covers the structure that works, the numbers to publish, and how to produce them without adding a job to anyone’s week.
Why case studies outperform every other page type
Three audiences read a case study, and it is the rare format that serves all three at once.
Buyers read them for proof and permission. A homeowner in Frankston looking at a Frankston install with a bill number attached is doing the exact trust arithmetic that closes deals, seeing themselves in a completed project. This is the considered, comparison heavy behaviour we mapped in the Australian solar customer journey, and a case study meets it better than any service page.
Search engines read them as suburb and system type coverage you cannot get any other way. Ten genuine project pages across your service area rank for the long tail of suburb plus solar searches without a single templated location page, the honest version of the local strategy in local SEO for solar installers.
And AI systems read them as quotable, checkable specifics. When an assistant explains why it recommends an installer, its stated reasons are facts pulled from the installer’s own pages, system sizes, locations, outcomes, timeframes. A case study is a page made entirely of such facts. The general principle, that assistants quote what businesses publish, runs through everything we measured in the 500 homepage study, and case studies are the highest density fact format a trade business can produce.
The structure: eight parts, one page, real numbers
A headline with the specifics in it. The suburb, the system, the headline outcome. 13.2 kilowatts in Craigieburn: a $412 quarterly bill down to $38. That is a title a buyer clicks, a search engine parses and an AI quotes, and it beats Residential Project by every measure that exists.
The situation. Three or four sentences on who and why, anonymised to the customer’s comfort. The bill pain, the roof, the motivation. This is where the reader recognises themselves, so write it in the customer’s terms, not yours.
The design decision. Why this size, this layout, this inverter, this brand. This section does the quiet work of demonstrating expertise, the difference between we installed panels and we split the array east west because the household’s usage peaks at breakfast and dinner. It is also where honest trade offs live, and honesty here is what separates your page from a brochure.
The install itself. Timeframe from deposit to grid connection, any complications and how they were handled, who did the work. Naming that your own employed crew installed it, if true, is a differentiator worth stating plainly.
The numbers block. The heart of the page, as a simple list: system size, panel count and brand, inverter, battery if fitted, install date, generation since, bill before and after, payback estimate. Publish real figures with the customer’s permission, ranged where commercial confidence requires it. Every number here is a fact an assistant can lift into an answer with your name attached, and the absence of numbers is why most installer content is unquotable.
What the customer says. Two or three sentences in their words, with a first name and suburb. Solar has no testimonial prohibition, use that freedom, and link the quote to your review profile where the fuller version lives, which compounds the review strategy from our review guide.
The photographs. Real ones: the roof, the switchboard, the app screenshot showing generation. Stock photography in a case study reads as fiction and undoes the whole page.
The bridge. One paragraph: if your situation looks like this, here is what an assessment involves, and a contact path. Restraint sells here; the page has already done the persuading.
The commercial case study is its own weapon
Everything above applies double for commercial work, with the numbers reframed: demand charges, daytime load matching, payback against operating cost, and the compliance and disruption story, installed over two weekends without interrupting trade, that commercial buyers actually fear. The strategic context bears repeating: commercial solar AI answers draw almost entirely from installer websites because so few installers publish real commercial content, ChatGPT pulls 5.5 sources per commercial answer against 2.6 residential, and the national field being cited is 23 domains thin. One genuine commercial case study with real numbers does not just support your commercial page, in the current landscape it can be the reason an assistant names you to a commercial buyer, as we found in our commercial solar study.
The production system: one hour per job, at handover
Case studies die in installers’ businesses because they are treated as writing projects for later. The fix is to make capture part of the job close, when everything is fresh and the customer is happiest.
At handover, three asks in five minutes: permission to feature the project, suburb level only, approve before publish. The photos, which the crew takes anyway, plus one of the app screen. And the quote, captured in the same conversation as the review ask. Then a fixed template, the eight sections above as a form, filled by whoever quotes the job, because they already know the design story. Sixty to ninety days later, pull the generation figure and the first real bill from the customer, drop the numbers in, publish. One page per month is twelve suburb anchored, fact dense pages a year, which is more citable content than most competitors will publish this decade.
Two quality rules. Never inflate: the numbers will be checked against physics by savvier buyers and cross referenced by machines, and one caught exaggeration poisons every page. And never template the prose: the numbers block repeats by design, the story must not, because ten identical pages with swapped suburbs are exactly the pattern search engines learned to discount years ago.
The mistakes that waste the format
Having read a lot of installer project pages while researching how AI cites them, we can catalogue the failure modes precisely, and avoiding them matters as much as following the structure.
The gallery page pretending to be a case study is the most common. Twelve photographs, a suburb name, no numbers, no story. It proves you can install panels, which no buyer doubted, and gives engines and assistants nothing to quote, because there are no claims on the page at all. If the choice is five gallery pages or one real case study, the single real one wins every time.
The victory lap is the second failure: a page written as self congratulation, another successful installation by our expert team, where every sentence is about the installer and none are about the customer’s situation or numbers. Buyers read past it the way they read past any ad, and the absence of specifics again leaves nothing extractable. The test is mechanical: count the concrete facts a stranger could check. Under five, rewrite it.
The confidentiality overcorrection kills otherwise good pages: installers assume customers will refuse permission and never ask, or strip every number to be safe. In practice, asked at handover with suburb level anonymity offered, most residential customers agree, and commercial customers frequently agree to ranged figures because the case study flatters their own decision. The permission conversation is a five minute script, not an obstacle.
And the orphan problem wastes more published case studies than anything else: pages that exist but are linked from nowhere except a projects index nobody visits. A case study earns its keep when the relevant service page cites it as evidence, when the suburb pages nearby reference it, and when its numbers appear in your structured data. Unlinked, it is a filing cabinet; linked, it is a proof network, and the difference is an hour of internal linking across the whole collection.
None of these mistakes require skill to avoid. They require only the decision that the page’s job is to carry checkable facts to a nervous buyer, and the discipline to delete anything on it that does not serve that job. Read your last project page with that test in hand and the rewrite usually takes an hour.
Where they live and how they work together
Give them a projects section with clean URLs naming suburb and system, and internally link them hard: every service page cites its three most relevant projects, every case study links its service page and its suburb neighbours. Feed the numbers into your schema and llms.txt so the machine readable layer matches the page layer. Then let the collection compound: after a year you have a body of published, dated, verifiable outcomes that functions as a proof archive no advertising budget can buy, the direct citation asset we argued for in the SolarQuotes analysis, built from work you were doing anyway.
Frequently asked questions
What should a solar case study include?
Eight parts: a specific headline with suburb and outcome, the customer situation, the design reasoning, the install story, a numbers block covering system size, brands, dates, generation and bills before and after, a short customer quote with first name and suburb, real photographs, and a restrained call to action. The numbers block is the part search engines and AI assistants actually quote.
Do I need customer permission to publish one?
Yes, and it is easiest gathered at handover: feature the project at suburb level, customer approves before publish, name withheld or first name only as they prefer. Built into the job close conversation alongside the review ask, refusal rates are low and the approved pages carry real quotes and real numbers.
How many case studies do I need before they matter?
The first three unlock service page linking and give buyers proof variety. Ten spread across your service area covers the suburb long tail. Every one is durable: a 2024 project page keeps ranking, converting and being cited for years, which is a different asset class from advertising that stops when the budget does.
Why do case studies matter for AI recommendations specifically?
Because assistants justify recommendations with checkable facts from the business’s own site, and case studies are the densest fact format an installer has. In commercial solar the effect is extreme: installer websites take 97 per cent of ChatGPT’s citations and the field is only 23 domains, so one genuine commercial case study can put you in answers with almost no competition.
Should case studies show real prices and savings?
As real as the customer permits, ranged where necessary, never inflated. Specific verified numbers are the entire advantage of the format, and the compliance instincts from solar advertising rules apply: claims you can substantiate, in plain terms. A ranged honest number beats a precise invented one on trust, conversions and risk.
Who should write them?
The person who designed and quoted the job fills the template, because they hold the story, and one editor tidies the prose for consistency. An external writer polishing internal substance works; an external writer inventing substance produces the generic sludge that neither buyers nor AI systems quote. The knowledge must come from inside the business.
