What the assistant is actually doing when someone asks
It is not consulting a directory of contractors. It is retrieving passages from pages it can read, plus whatever it holds about your business as an entity, and assembling a recommendation from them.
Which means three things have to be true: it can read your site at all, it can tell what you do and where, and there is something quotable to lift. Most contractor websites fail the first one, which makes the other two moot.
The practical version of that: if your site builds its pages in the browser, the crawlers behind ChatGPT, Claude and Perplexity receive an empty page. Your Google ranking will not tell you, because Googlebot renders JavaScript and those crawlers do not.
The short versionReadable, identifiable, quotable. In that order, because the first one gates the rest.
Service area has to be written down, not implied
Almost every local query has a place in it. If your service area exists only in a footer, a map embed, or the assumption that everyone knows where you are based, there is nothing for a machine to extract.
Name the towns and neighbourhoods you actually serve, in text, on a page that exists for that purpose. Add the structured data that states the same thing unambiguously. If you serve a radius, say the radius and name the places in it.
The trap to avoid is generating a near-identical page for every town within fifty miles. That is a doorway-page pattern, it is a known quality problem, and it costs more than it earns. One honest, specific page beats forty thin ones.
The short versionWrite the service area as text on a real page. Do not spin up one thin page per town.
Answer the questions people actually ask a machine
People ask assistants things they would be slightly embarrassed to ask a contractor: whether a twelve-year-old furnace is worth repairing, what a fair price is for a callout, whether a noise means something expensive, whether they can wait until Monday.
These are the questions your site almost certainly does not answer, because your site is written to establish credibility rather than to be useful before someone calls. They are also exactly the shape an answer engine retrieves: a question as a heading with a complete answer beneath it.
Answering them honestly, including when the answer is "that is usually a cheap fix, get a second opinion if someone quotes you a replacement", is what gets a business cited as a source rather than listed as an option.
The short versionAnswer the pre-purchase questions, including the ones where the honest answer costs you a sale.
Evidence, and the specific way trades get this wrong
The KDD 2024 research found that citations to sources, quotations from credible sources, and statistics each lifted generative visibility by roughly 30 to 40%. Keyword stuffing performed at or below baseline.
The trades version of getting this wrong is repeating industry statistics from vendor blogs, which circulate as a closed loop with no study at the end of the chain. We found this while writing about missed calls: several widely quoted figures trace back to nothing measurable, and one commonly cited dollar value turns out to be a worked example that got repeated until it read as data.
A statistic you cannot trace is worse than no statistic, because it sits in the exact spot where you are claiming expertise. Where you have no sourceable number, describe the mechanism instead. Your own trade experience is evidence that a content mill cannot fake.
The short versionCite real sources. And where none exists, use the specific trade detail that only someone who does this work would know.
The things that count and are not on your website
Assistants draw heavily on third-party sources when answering local questions. Your Google Business Profile, review platforms, trade directories and local press all feed what a model can say about you, and consistency across them matters more than polish on any one.
The unglamorous version of this work: the same business name, address, phone number and category everywhere, categories chosen accurately rather than aspirationally, and reviews earned rather than bought.
What we would not advise is astroturfing the community sources. It gets detected, and the cost when it does is larger than the benefit while it works.
The short versionConsistent details across every third-party listing. It is boring and it is load-bearing.
Why the small operator wins here
In classic search, a national competitor's decade of backlinks is a wall you cannot climb quickly. In generative search, a precise, well-evidenced, specific answer from a two-van operation can be cited ahead of a vague page from a franchise, and the research says smaller sites gain the most.
There is a second advantage that is harder to copy. The technician who has spent fifteen years on emergency callouts can write the answer to "is this noise serious" in a way that a marketing agency writing for forty contractors cannot. That specificity is precisely what gets quoted.
The window is open because the sector has not noticed yet. That will not last, and the sites that are readable and quotable when the assistants are consulted more heavily are the ones that will already be in the answers.
The short versionStructure and specific trade knowledge beat budget in this channel. That is unusual, and it is temporary.
Where to start, in order
Check whether the crawlers can read your site without JavaScript. This takes about a minute and it decides whether anything else is worth doing.
Then fix what is blocking them, write the service area properly, and answer ten real pre-purchase questions as headings with complete answers. Then tidy the third-party listings.
That order is deliberate. Content on an unreadable site is invisible content, and it is the most common way this money gets wasted.
The short versionRendering, then robots, then service area, then answers, then listings. Skipping to content is the expensive mistake.
