insight
insight
announcement
press
Author:
April Ford
Someone in Brisbane needs a commercial lawyer, a paediatric dentist, or a bookkeeper who understands property trust accounts. A few years ago they would open Google, scan a page of results and clicked on four or five options. Plenty now open ChatGPT instead, describe the situation in a sentence or two, and get three names back.
OpenAI reported ChatGPT had passed 900 million weekly active users in February 2026. Not all of them are shortlisting suppliers, but enough are that it matters. And unlike a search results page, there is no scrolling past the fold. You are either in the answer or you are not mentioned at all.
For service businesses this is more consequential than it looks. When someone buys a product they can compare specs. When they choose a professional service they are outsourcing a judgement call, which is exactly the kind of question people bring to an assistant.
The work involved is mostly unglamorous. Here are seven things that help.
Someone in Brisbane needs a commercial lawyer, a paediatric dentist, or a bookkeeper who understands property trust accounts. A few years ago they would open Google, scan a page of results and clicked on four or five options. Plenty now open ChatGPT instead, describe the situation in a sentence or two, and get three names back.
OpenAI reported ChatGPT had passed 900 million weekly active users in February 2026. Not all of them are shortlisting suppliers, but enough are that it matters. And unlike a search results page, there is no scrolling past the fold. You are either in the answer or you are not mentioned at all.
For service businesses this is more consequential than it looks. When someone buys a product they can compare specs. When they choose a professional service they are outsourcing a judgement call, which is exactly the kind of question people bring to an assistant.
The work involved is mostly unglamorous. Here are seven things that help.
Start here, because a surprising number of Australian sites fail at the first step.
OpenAI runs three separate crawlers doing different jobs:
When AI scraping became a talking point a couple of years ago, plenty of businesses asked their web developer to block the OpenAI bot. Some of those blocks were applied broadly, which removed the site from ChatGPT search results along with the training data.
Open your robots.txt file and look. If OAI-SearchBot is disallowed, ChatGPT cannot retrieve your pages and no amount of content work will change that. You can block GPTBot to stay out of training while still allowing OAI-SearchBot to index you. They are separate decisions.
An assistant does not send someone to your page to read it. It reads the page and restates the useful part.
That changes what a good page looks like. A services page that opens with two paragraphs about your commitment to excellence before it explains what you actually do gives the model very little to work with.
In practice:
Vagueness about pricing and process is common in professional services and it actively hurts here. A model asked to recommend an accountant for a small business will favour the firm whose site says what it does for small businesses over the one promising tailored solutions for clients of all sizes.
This is the part most businesses underestimate.
When an assistant recommends suppliers, it draws heavily on third-party sources: review platforms, industry directories, professional association listings, local news and comparison articles. Your website confirms you exist. Everywhere else establishes whether you are any good.
For an Australian service business that usually means:
If a competitor appears in AI answers and you do not, the difference is more often here than on their homepage.
Reviews deserve their own point, because they carry more weight than most businesses expect and because very few collect them with any consistency.
Assistants lean on reviews for the same reason people do. They are one of the few signals about a business that the business did not write itself. Volume counts, recency counts, and so does what the reviews actually say.
That last part is where most businesses lose ground. A review reading "great service, highly recommend" gives an assistant nothing to match a question against. One that mentions the specific service, the suburb and the problem solved is far more useful, because it lines up with how people phrase what they are looking for.
You cannot script what clients write, but you can influence it through when and how you ask:
Consistency beats intensity here. A steady trickle of recent reviews is worth more than thirty gathered in a fortnight two years ago, and it reads as considerably more credible.
Assistants assemble a picture of your business from fragments scattered across the web. Contradictions make that harder, and a model unsure about what you do is less likely to recommend you with confidence.
Check that your business name, service descriptions, practitioner names, locations and opening hours match across your website, Google Business Profile, directory listings, association registers and LinkedIn. Multi-location service businesses get this wrong constantly, usually because a suburb changed or a practitioner moved and only two of nine listings were updated.
If your content restates advice the model could generate itself, there is no reason to cite you.
What gets quoted tends to have something specific attached:
That last one is underused. A firm that publishes a clear explanation of what a new state regulation means for local businesses is far more citable than one publishing another article on the importance of good advice.
This is more measurable than most people assume.
In GA4, referrals from AI assistants arrive as their own referral sources and can be segmented out. Look past session counts to what those visitors do, since volumes are small across most industries but tend to arrive further along in the decision. Our guide to Google Analytics 4 and Google Ads covers the reporting setup.
Run a manual check alongside it. Write down ten questions a client in your industry would realistically ask, including the location, put them to several assistants, and record whether you appear, how you are described and who shows up beside you. Repeat monthly.
If your enquiries arrive by phone, which they will for most service businesses, make sure call tracking is in place. Otherwise this channel will look like it does nothing. Or be diciplined enough to ask where they heard about you.
The market has filled with tools promising guaranteed AI visibility. Be sceptical. There is no paid placement for organic recommendations, no submission form, no verified shortcut. Some widely promoted technical fixes, including certain proposed file standards for AI crawlers, have never been confirmed as supported by OpenAI. Anyone offering certainty here is overselling.
The businesses showing up in these answers are mostly the ones with clear content, genuine third-party credibility and consistent information across the web. That is not a new formula. It is being applied by a different reader.
AI assistants are not replacing search, and the traffic they send remains modest for most Australian businesses. What has changed is that a meaningful share of shortlists are now generated by a machine, and very few service businesses have done anything deliberate about whether they appear on them.
The work is largely an extension of good search engine optimisation rather than a separate discipline, which means most of it compounds with what you are already doing.
If you want to know how your business currently appears in AI answers, and what it would take to change that, get in touch with the team at April Ford.
Can I pay to be recommended by ChatGPT?
Not for organic recommendations within answers. OpenAI runs a separate advertising product with clearly labelled sponsored placements, but that is distinct from being cited in the response itself.
Does this matter for a local service business?
Arguably more than for national brands. Location-specific questions such as finding a physiotherapist in a particular suburb are common, and local visibility depends heavily on third-party sources you can influence.
How long does it take to see results?
Changes to your own site can be reflected reasonably quickly once pages are re-crawled. Building third-party presence takes longer, usually months rather than weeks.
Does this replace SEO?
No. The two overlap heavily and strong SEO foundations improve AI visibility. Treat it as an extension of existing search work.
How do I check whether ChatGPT currently recommends us?
Ask it. Put the questions your clients would ask, with your city or suburb included, to several assistants and record what comes back. It costs nothing.
Why does a competitor appear when we do not?
Usually because they have stronger third-party presence through reviews, directory listings and inclusion in comparison content, rather than a better website.
Start here, because a surprising number of Australian sites fail at the first step.
OpenAI runs three separate crawlers doing different jobs:
When AI scraping became a talking point a couple of years ago, plenty of businesses asked their web developer to block the OpenAI bot. Some of those blocks were applied broadly, which removed the site from ChatGPT search results along with the training data.
Open your robots.txt file and look. If OAI-SearchBot is disallowed, ChatGPT cannot retrieve your pages and no amount of content work will change that. You can block GPTBot to stay out of training while still allowing OAI-SearchBot to index you. They are separate decisions.
An assistant does not send someone to your page to read it. It reads the page and restates the useful part.
That changes what a good page looks like. A services page that opens with two paragraphs about your commitment to excellence before it explains what you actually do gives the model very little to work with.
In practice:
Vagueness about pricing and process is common in professional services and it actively hurts here. A model asked to recommend an accountant for a small business will favour the firm whose site says what it does for small businesses over the one promising tailored solutions for clients of all sizes.
This is the part most businesses underestimate.
When an assistant recommends suppliers, it draws heavily on third-party sources: review platforms, industry directories, professional association listings, local news and comparison articles. Your website confirms you exist. Everywhere else establishes whether you are any good.
For an Australian service business that usually means:
If a competitor appears in AI answers and you do not, the difference is more often here than on their homepage.
Reviews deserve their own point, because they carry more weight than most businesses expect and because very few collect them with any consistency.
Assistants lean on reviews for the same reason people do. They are one of the few signals about a business that the business did not write itself. Volume counts, recency counts, and so does what the reviews actually say.
That last part is where most businesses lose ground. A review reading "great service, highly recommend" gives an assistant nothing to match a question against. One that mentions the specific service, the suburb and the problem solved is far more useful, because it lines up with how people phrase what they are looking for.
You cannot script what clients write, but you can influence it through when and how you ask:
Consistency beats intensity here. A steady trickle of recent reviews is worth more than thirty gathered in a fortnight two years ago, and it reads as considerably more credible.
Assistants assemble a picture of your business from fragments scattered across the web. Contradictions make that harder, and a model unsure about what you do is less likely to recommend you with confidence.
Check that your business name, service descriptions, practitioner names, locations and opening hours match across your website, Google Business Profile, directory listings, association registers and LinkedIn. Multi-location service businesses get this wrong constantly, usually because a suburb changed or a practitioner moved and only two of nine listings were updated.
If your content restates advice the model could generate itself, there is no reason to cite you.
What gets quoted tends to have something specific attached:
That last one is underused. A firm that publishes a clear explanation of what a new state regulation means for local businesses is far more citable than one publishing another article on the importance of good advice.
This is more measurable than most people assume.
In GA4, referrals from AI assistants arrive as their own referral sources and can be segmented out. Look past session counts to what those visitors do, since volumes are small across most industries but tend to arrive further along in the decision. Our guide to Google Analytics 4 and Google Ads covers the reporting setup.
Run a manual check alongside it. Write down ten questions a client in your industry would realistically ask, including the location, put them to several assistants, and record whether you appear, how you are described and who shows up beside you. Repeat monthly.
If your enquiries arrive by phone, which they will for most service businesses, make sure call tracking is in place. Otherwise this channel will look like it does nothing. Or be diciplined enough to ask where they heard about you.
The market has filled with tools promising guaranteed AI visibility. Be sceptical. There is no paid placement for organic recommendations, no submission form, no verified shortcut. Some widely promoted technical fixes, including certain proposed file standards for AI crawlers, have never been confirmed as supported by OpenAI. Anyone offering certainty here is overselling.
The businesses showing up in these answers are mostly the ones with clear content, genuine third-party credibility and consistent information across the web. That is not a new formula. It is being applied by a different reader.
AI assistants are not replacing search, and the traffic they send remains modest for most Australian businesses. What has changed is that a meaningful share of shortlists are now generated by a machine, and very few service businesses have done anything deliberate about whether they appear on them.
The work is largely an extension of good search engine optimisation rather than a separate discipline, which means most of it compounds with what you are already doing.
If you want to know how your business currently appears in AI answers, and what it would take to change that, get in touch with the team at April Ford.
Can I pay to be recommended by ChatGPT?
Not for organic recommendations within answers. OpenAI runs a separate advertising product with clearly labelled sponsored placements, but that is distinct from being cited in the response itself.
Does this matter for a local service business?
Arguably more than for national brands. Location-specific questions such as finding a physiotherapist in a particular suburb are common, and local visibility depends heavily on third-party sources you can influence.
How long does it take to see results?
Changes to your own site can be reflected reasonably quickly once pages are re-crawled. Building third-party presence takes longer, usually months rather than weeks.
Does this replace SEO?
No. The two overlap heavily and strong SEO foundations improve AI visibility. Treat it as an extension of existing search work.
How do I check whether ChatGPT currently recommends us?
Ask it. Put the questions your clients would ask, with your city or suburb included, to several assistants and record what comes back. It costs nothing.
Why does a competitor appear when we do not?
Usually because they have stronger third-party presence through reviews, directory listings and inclusion in comparison content, rather than a better website.
Start here, because a surprising number of Australian sites fail at the first step.
OpenAI runs three separate crawlers doing different jobs:
When AI scraping became a talking point a couple of years ago, plenty of businesses asked their web developer to block the OpenAI bot. Some of those blocks were applied broadly, which removed the site from ChatGPT search results along with the training data.
Open your robots.txt file and look. If OAI-SearchBot is disallowed, ChatGPT cannot retrieve your pages and no amount of content work will change that. You can block GPTBot to stay out of training while still allowing OAI-SearchBot to index you. They are separate decisions.
An assistant does not send someone to your page to read it. It reads the page and restates the useful part.
That changes what a good page looks like. A services page that opens with two paragraphs about your commitment to excellence before it explains what you actually do gives the model very little to work with.
In practice:
Vagueness about pricing and process is common in professional services and it actively hurts here. A model asked to recommend an accountant for a small business will favour the firm whose site says what it does for small businesses over the one promising tailored solutions for clients of all sizes.
This is the part most businesses underestimate.
When an assistant recommends suppliers, it draws heavily on third-party sources: review platforms, industry directories, professional association listings, local news and comparison articles. Your website confirms you exist. Everywhere else establishes whether you are any good.
For an Australian service business that usually means:
If a competitor appears in AI answers and you do not, the difference is more often here than on their homepage.
Reviews deserve their own point, because they carry more weight than most businesses expect and because very few collect them with any consistency.
Assistants lean on reviews for the same reason people do. They are one of the few signals about a business that the business did not write itself. Volume counts, recency counts, and so does what the reviews actually say.
That last part is where most businesses lose ground. A review reading "great service, highly recommend" gives an assistant nothing to match a question against. One that mentions the specific service, the suburb and the problem solved is far more useful, because it lines up with how people phrase what they are looking for.
You cannot script what clients write, but you can influence it through when and how you ask:
Consistency beats intensity here. A steady trickle of recent reviews is worth more than thirty gathered in a fortnight two years ago, and it reads as considerably more credible.
Assistants assemble a picture of your business from fragments scattered across the web. Contradictions make that harder, and a model unsure about what you do is less likely to recommend you with confidence.
Check that your business name, service descriptions, practitioner names, locations and opening hours match across your website, Google Business Profile, directory listings, association registers and LinkedIn. Multi-location service businesses get this wrong constantly, usually because a suburb changed or a practitioner moved and only two of nine listings were updated.
If your content restates advice the model could generate itself, there is no reason to cite you.
What gets quoted tends to have something specific attached:
That last one is underused. A firm that publishes a clear explanation of what a new state regulation means for local businesses is far more citable than one publishing another article on the importance of good advice.
This is more measurable than most people assume.
In GA4, referrals from AI assistants arrive as their own referral sources and can be segmented out. Look past session counts to what those visitors do, since volumes are small across most industries but tend to arrive further along in the decision. Our guide to Google Analytics 4 and Google Ads covers the reporting setup.
Run a manual check alongside it. Write down ten questions a client in your industry would realistically ask, including the location, put them to several assistants, and record whether you appear, how you are described and who shows up beside you. Repeat monthly.
If your enquiries arrive by phone, which they will for most service businesses, make sure call tracking is in place. Otherwise this channel will look like it does nothing. Or be diciplined enough to ask where they heard about you.
The market has filled with tools promising guaranteed AI visibility. Be sceptical. There is no paid placement for organic recommendations, no submission form, no verified shortcut. Some widely promoted technical fixes, including certain proposed file standards for AI crawlers, have never been confirmed as supported by OpenAI. Anyone offering certainty here is overselling.
The businesses showing up in these answers are mostly the ones with clear content, genuine third-party credibility and consistent information across the web. That is not a new formula. It is being applied by a different reader.
AI assistants are not replacing search, and the traffic they send remains modest for most Australian businesses. What has changed is that a meaningful share of shortlists are now generated by a machine, and very few service businesses have done anything deliberate about whether they appear on them.
The work is largely an extension of good search engine optimisation rather than a separate discipline, which means most of it compounds with what you are already doing.
If you want to know how your business currently appears in AI answers, and what it would take to change that, get in touch with the team at April Ford.
Can I pay to be recommended by ChatGPT?
Not for organic recommendations within answers. OpenAI runs a separate advertising product with clearly labelled sponsored placements, but that is distinct from being cited in the response itself.
Does this matter for a local service business?
Arguably more than for national brands. Location-specific questions such as finding a physiotherapist in a particular suburb are common, and local visibility depends heavily on third-party sources you can influence.
How long does it take to see results?
Changes to your own site can be reflected reasonably quickly once pages are re-crawled. Building third-party presence takes longer, usually months rather than weeks.
Does this replace SEO?
No. The two overlap heavily and strong SEO foundations improve AI visibility. Treat it as an extension of existing search work.
How do I check whether ChatGPT currently recommends us?
Ask it. Put the questions your clients would ask, with your city or suburb included, to several assistants and record what comes back. It costs nothing.
Why does a competitor appear when we do not?
Usually because they have stronger third-party presence through reviews, directory listings and inclusion in comparison content, rather than a better website.
