Which Australian industry has the worst AI visibility?

The six industries, ranked by how often nobody was named

Tradies finish last on both of the measures in our dataset that can carry a ranking, and on the never-named rate they finish last by a distance: 76% of 21, against 47% of 19 for the industry immediately above them. Fifty businesses were scanned per industry. The never-named column is counted over each industry's businesses with complete records from both engines, because a blank Gemini row cannot be counted either way, so read every percentage against the complete-records column beside it.

Why tradies finish last, on the evidence we have

Not because their websites are worse. Trade websites in our sample averaged 71.5 out of 100 for on-site AI readiness, which is below the national 74.9 but nowhere near far enough below to explain a never-named rate of 76% of 21 against a national 41% of 137. Only 12% of the 50 trade sites block AI crawlers, the equal lowest blocking rate in the study, so the engines could read almost all of them.

What our data can say is that readiness and visibility have come apart, and that they have come apart furthest here. Across the 137 businesses with complete two-engine records, on-site readiness correlates with the per-engine visibility scores at Pearson r = 0.116. Tidy sites are not what these engines are choosing on, in any of the six industries.

What our data cannot say is exactly what the trades are missing instead. We measured outcomes, not causes, on one day. Anyone telling you precisely why an entire industry is invisible, ours included, is reasoning past the evidence. The published six causes are ranked by how often we found each across the sample, which is as close as this dataset gets to a cause.

The result that surprises people: the best websites finished second last

Accountants have the best-prepared websites in the entire study and the second-worst visibility in it. Accounting websites averaged 80.8 out of 100 for on-site AI readiness, the highest of the six industries and well clear of the national 74.9. Only 14% of the 50 firms were missing structured data, the lowest rate we recorded. Their mean AI visibility score is 34 out of 100, and 47% of the 19 accountants with complete two-engine records were never named by either engine.

That pairing is the most useful thing in the industry table. It removes the comfortable explanation. If website quality decided this, accountants would sit at the top of the ranking rather than one place off the bottom. It is also the evidence behind whether AI visibility is just SEO with a new name, because the part classic SEO optimises hardest is the part that did not decide the outcome here.

What "worst" can and cannot mean in this data

It means worst of the six industries we measured, on one day, on one set of questions, and nothing wider than that. The question as people usually ask it assumes there is a settled league table of Australian industries somewhere. There is not one, ours included, and a page that hands you a ranking without saying where it stops is selling you a confidence the measurement does not have.

• One scan week. Every row is dated 30 July 2026. These engines change their answers, so this is a snapshot, not a trend.
• Each question asked once. Five shared buyer questions per industry and city cohort on ChatGPT, plus one grounded direct question per business on Gemini. No repetition runs, so a business one answer short of a mention on the day stays unmentioned here. Why that matters is set out in why AI answers change every time you ask.
• Unequal denominators. Complete two-engine records range from 19 accountants to 29 dentists. A percentage over 19 rows moves by more than five points every time one business changes state, so treat gaps of a few points between the middle four industries as noise.
• Search-based sampling. Businesses were drawn from the top organic results for buyer-style queries, plus recognisable national anchor brands. That skews the sample toward businesses that already rank well in classic search, which strengthens the finding rather than weakening it: these are drawn from the visible end of the market.
• Six industries is not every industry. Nothing here rules out an unmeasured industry sitting below tradies.

The full method, including how the scores are built and where they stop being able to tell you things, is on the technology page and in the study's own methodology section.

Six industries is not every industry

Six is the whole sample, so nothing on this page rules out an unmeasured industry sitting below tradies. What we scanned was 50 businesses each across tradies, accountants, cosmetic and skin clinics, dentists, law firms and e-commerce. Hospitality, real estate, allied health, automotive and every other category were not scanned at all, and this page cannot tell you where any of them would land.

We are not going to guess at them either, not a rank, not a range and not a direction of travel, because a number nobody can check against raw data is worth nothing. If we scan further industries, they will arrive the way these six did: every row published as CSV and JSON under CC BY 4.0, every percentage carrying its denominator, and the caveats stated in the same place as the findings.

What an industry average tells you about your own business

Less than people expect, because the spread inside every industry is wider than the gap between industries. The worst industry in the study still contains its leader: the top tradie score was 71 out of 100, held by Fallon Solutions of Brisbane, against an industry mean of 29.8. The best industry still contains businesses that were never named at all.

So an industry figure is useful for one thing, which is calibrating your expectations before you look. It cannot tell you which of the three states your own business is in. Two questions asked in a fresh chat will, and the method is in how to check what ChatGPT says about your business.