Does AI visibility work for a new business with no reviews and no backlinks?

The size gap, measured

Recognisable brands do better, and our own data says by how much. We tagged 36 of the 300 businesses in the study as anchor brands, meaning names a general Australian audience would plausibly recognise, and left the other 264 untagged.

What the named non-anchors in our sample had in common

Not their websites, which is the most useful thing in the data. The 28 non-anchor businesses named by both engines averaged 79.5 out of 100 for on-site AI readiness, and the 54 non-anchors never named by either averaged 61.0. That looks like a website story until you look at the whole sample: across all 137 businesses with complete two-engine records, site readiness correlates with mean per-engine visibility at Pearson r = 0.116. Readiness behaves like a floor, not a lever.

The single clearest thing separating the two groups was crawler access. Twenty-seven of the 28 named non-anchors allowed AI crawlers in robots.txt. Among the 54 never-named non-anchors, 40 did, so 14 were blocking the systems they wanted to be named by. An llms.txt file separated almost nothing: 8 of the 28 had one, and 11 of the 54.

The study's own conclusion says it in four words: fix the site, then earn the mentions. The full method and the per-industry splits are in the study.

The order of operations for a business with nothing yet

Do the free and structural things first, in this order, and do not pay anybody until you reach step five.

1. Unblock the AI crawlers. One file, no agency, no cost. In our scan 13% of all 300 businesses were blocking AI crawlers outright in robots.txt. Check yours before anything else, because everything downstream is wasted while it is set.
2. Make your facts unambiguous and identical everywhere. Legal name, trading name, address, phone, service area, services, and prices if you publish them. Identical on your site, your Google Business Profile and every directory. Contradictory listings make it harder for any system to state anything about you confidently.
3. Get listed where your category is already listed. Google Business Profile, the directories your industry actually uses, association and regulator registers. These are the third-party pages the engines retrieve, and being absent from them is why an engine has nothing to cite.
4. Earn real reviews at a steady rate, on the platforms that already rank for your buyer questions. Steady beats a burst, and the platform matters more than the count.
5. Only now consider paying anyone, and only against a gap you have measured.

None of the first four require an agency, and steps one and two are usually a single afternoon. Where the information the engines use actually comes from is covered in where ChatGPT gets information about your business.

When to spend nothing on this and fix something else first

If you cannot serve more customers than you have today, spend nothing on this. Demand you cannot fulfil is not an improvement, and it is the most common reason this work is a bad purchase.

Four more situations where the honest answer is not yet:

• Your site has no clear service pages. Nothing downstream compensates for a site that cannot answer the question a buyer arrives with.
• You have almost no reviews. Reviews are upstream of this whole problem. Fix that first and some of this fixes itself.
• You are pre-launch, with no trading name, address or service area settled. There is no entity for a system to resolve yet.
• Someone has offered you a guaranteed AI ranking. Walk away. Nobody controls placement inside a generated answer, including us.

Our longer version of this argument, written by a firm that sells the work and therefore has every reason not to publish it, is on is AI visibility worth it.

A twelve month expectation that is not a promise

We will not put a number or a date on it, because we have not measured one. Our study is a single scan on a single day. Nothing in it tracks how long a business takes to move from never named to named, because we did not re-scan the sample.

What the data does support is a shape:

• The businesses being named have external corroboration. That is what separates them, far more than their own markup.
• Corroboration accrues at the speed the rest of the web moves: reviews arrive, listings get approved, roundups get published and recrawled. That is months, not weeks.
• The one lever with a plausibly fast effect is removing a block, because it stops actively excluding you from live retrieval.

So plan in quarters, judge by the rate across a fixed question set rather than by a screenshot, and treat any vendor timeline stated in weeks as a claim about their sales process rather than about the engines. How to read progress honestly is covered in how to know if the work is actually working.