Trades & Construction AI Visibility Study: 100 Attempts

Home owners now ask search and AI tools to suggest builders, electricians, plumbers, gasfitters and roofers before visiting a contractor's website. The answer can shape the shortlist, the evidence a buyer sees and whether they request a quote.

We tested 25 Australian trades and construction questions across ChatGPT, Google Search, Microsoft Copilot and Claude. The 100 attempts covered residential building, electrical, plumbing and gas, and roofing in Melbourne, Sydney, Brisbane, Perth and Adelaide, plus national questions about licensing, insurance, safety and written quotes.

Download the full AI Visibility in Australian Trades & Construction white paper (PDF)

Inspect the frozen prompts, methodology, privacy-safe coded observations and recalculated summary.

What we tested

We wrote all 25 questions before collection and retained a fixed denominator of 100 attempted observations: 25 on each surface. Seventy-five attempts produced a material answer we could review. Twenty-five technical capture failures remain in the record and are excluded from answer-quality percentages; no inaccessible answer was backfilled.

We reviewed Presence, Evidence, Accuracy and Journey. This is not a ranking of contractors, and an appearance does not prove licence status, insurance, availability, workmanship or suitability.

What the study found

Licence advice often stopped before the evidence

Among 75 usable answers, 66 told the buyer how to verify licence or registration status. Only 22 linked an applicable state licence or registration source.

The distinction matters. An ABN confirms business identity, a white card confirms general construction induction, and an insurance certificate addresses cover. None substitutes for the licence class and scope required for the particular work and jurisdiction.

Business websites remained central to discovery

All 56 usable local discovery answers named at least one business or contractor, and 45 linked directly to a business website. Commercial directories or marketplaces appeared in 18 of the 75 usable answers.

Directories can support a shortlist, but the contractor's own service, project, licence and contact pages remain where scope, identity, evidence and the next step should be resolved.

Written-quote guidance was stronger than availability evidence

Of the 75 usable answers, 67 qualified insurance or applicable warranty evidence, 66 addressed project or licence-scope fit, 65 supported a sufficiently detailed written quote, and 58 treated availability or timing as something to confirm.

A buyer can therefore receive a convincing shortlist before knowing who will perform the work, whether the contractor has capacity or whether an advertised response time applies to the requested job.

Useful guidance and verifiable evidence were separate tests

We called an answer safe and actionable when it preserved choice, gave a sensible next step, included at least four relevant safeguards and told the buyer to verify the exact licence or registration. Fifty-four of 75 usable answers met that standard, while only 23 of 75 linked any Australian government or regulator source.

The technical exclusions make platform-level comparisons uneven, so these results should not be treated as a permanent ranking. The stronger cross-surface conclusion is that practical instructions were common but direct regulator evidence was not.

Seven practical actions for builders and trades businesses

Build pages for real jobs. Explain the exact work, service area, project type, inclusions, exclusions and enquiry path.
Put licence proof beside the claim. Show the licence holder and class, then link the applicable official register.
Separate identity from competency. Make the trading entity and ABN clear without presenting them as a trade credential.
Explain insurance and warranty evidence. Use jurisdiction- and project-specific language and identify what the buyer should request.
Show relevant project proof. Describe comparable constraints, materials and scope without promising identical outcomes.
Make quotes easier to compare. Cover allowances, exclusions, dates, payment stages and variations.
Measure the full journey. Track Presence, Evidence, Accuracy and Journey alongside calls, forms, quote requests and won jobs.

These actions belong inside a wider trades and construction SEO and AI visibility programme.

What this study cannot tell us

This was an observational study completed on 8 September 2026. Results can change by account, location, interface and product experiment. The five cities and four service categories do not represent every trade or construction journey.

The study does not measure traffic, workmanship, safety performance, customer outcomes, market share or permanent rankings. Twenty-five observations were invalid because of fixed-window or browser-capture failures. No inaccessible result was reconstructed.

Integral Media sells SEO and AI-visibility services to trades and construction businesses. We disclose the frozen questions, coding rules, answer counts, exclusions, primary-source set and commercial interest so readers can judge the findings in context.