How to get your products recommended by ChatGPT and AI search

How to get your products recommended by ChatGPT and AI search

In the last week alone, my team and I have helped three clients whose Meta ad accounts or profiles were shut down. For reasons not even Meta can explain.

Three. In one week.

If you run an eCommerce brand, you’ll know exactly what that means. For most brands we work with, Meta still produces the bulk of their traffic and sales. So when an account disappears overnight, it isn’t an inconvenience. It’s devastating for the business. And it’s incredibly stressful for the people behind it – the founder refreshing her inbox at midnight, waiting on an appeal that nobody seems to be reading.

If I had a dollar for every time someone’s asked me whether they can grow their eCommerce brand without Meta ads, I’d be very, very rich.

The honest answer? Not quite. Not yet.

But the work we’ve been doing to grow sales from organic search – including getting recommended by AI – has us heading in the right direction. And today I want to show you that using my own brand, Indagare. Because with my own brand, I can show you everything.

Here’s what happened.

A year ago, I switched off my Meta ads. And my Google traffic went with them.

This winter, I paused them again. This time, Google held. New customers held.

This month, sales from Google are up 147% on the same time last year. I’m getting full-price orders from the US and the UK – markets I’ve never advertised in. Not once.

And ChatGPT? It barely shows up in my analytics at all.

So how do those things fit together? That’s what this episode is about. How to get your products recommended by ChatGPT, Google and AI search. Why you probably won’t see it happening in your dashboard. And why, with ad costs where they are right now, this has become the difference between growing profitably and paying for every single sale.

What’s happened to search

Let’s start with Google. Because something has changed there that most founders I speak to haven’t fully clocked yet.

AI Overviews – those AI-written summaries at the top of your search results – arrived in Australia around October 2024. Then on the eighth of October last year, Google started rolling out AI Mode here. That’s the full conversational version, where you ask a long, complicated question and Google does the research for you.

And here’s what that’s done to clicks.

Two in every three Google searches now end without anyone clicking anything. Not on a website. Not on an ad. Nothing. That’s from SparkToro’s research using US data from the first four months of this year – 68% of searches ended with no click at all. Two years ago it was about 60%.

Of the clicks that are left, about a quarter stay inside Google. YouTube, Maps, AI Mode. Google’s own properties.

So only around a quarter of all Google searches now send someone to a website Google doesn’t own. Two years ago, it was more than a third.

And when an AI Overview appears at the top of the page? Ahrefs found the number one result gets about 58% fewer clicks. Pew Research looked at real people’s searches and found they clicked a result 8% of the time when there was an AI summary – and 15% of the time when there wasn’t. Nearly half the clicks, gone.

I should say this is US data, because that’s where the best research is. I haven’t seen an equivalent Australian study. But Google’s AI search works the same way here.

And in fairness, Google disputes all of this. Its position is that the volume of clicks to websites is broadly steady, and that the clicks AI sends are higher quality.

Hold on to that second part, because I actually think it’s true. And it’s the most important idea in this episode.

Now, here’s where it gets personal.

October last year – the month AI Mode arrived in Australia – was the month Indagare’s Google traffic hit the floor. Around the same time, Meta’s Andromeda update rolled out properly and destabilised a lot of ad accounts. Mine included. So badly that I turned my ads off completely for four months to step back and rethink the strategy.

Can I tell you which of those things caused what? No. Nobody can. It was a year where everything moved at once – the platforms, the algorithms, and the way people search and decide what to buy.

What I can tell you is that it was the moment I stopped treating organic as something that would just happen in the background. And started building it deliberately. Everywhere.

Why fewer clicks can mean more sales

So let’s think about how your customer actually shops now.

Say she’s looking for a gift for her mum. She types something like, “What’s a good natural face oil for mature, sensitive skin? It’s a gift.”

The AI goes away and does the work. It reads product pages. It reads reviews. It reads articles and forum threads. It compares ingredients and prices. Then it comes back with three or four brands, and a reason for each one.

She hasn’t clicked on anything yet.

And when she does click, she isn’t browsing anymore. She’s not comparing ten tabs. She’s already chosen – or she’s very close. She’s coming to check you out, and probably to buy.

The click didn’t disappear. The browsing did.

That’s what I’m seeing in my own numbers. Orders that come through Google are now worth about half as much again as they were a year ago. People are arriving and buying more.

I can’t prove that every one of those customers came through an AI answer first. I want to be honest about that. But it’s exactly the pattern you’d expect if more of the shopping is happening before the click.

And this is where it gets interesting for premium brands.

Your customer has always been a researcher. She reads the ingredient list. She reads the reviews. She asks her friends. She doesn’t buy a $90 serum on impulse.

Now she does a lot of that research inside an AI answer. And either you’re in that answer, or you’re not.

Here’s the uncomfortable bit. The AI doesn’t care how beautiful your Instagram grid is. It isn’t looking at your brand photography. It’s reading what the internet says about you – your product pages, your reviews, what other people have written about you. If the internet doesn’t say much about you, the AI doesn’t either.

Why you can’t see it working

Now – if AI is recommending your brand, you’d think you’d see it in your analytics. A nice line in Shopify that says “ChatGPT”. Lots of sales.

Most founders look, see almost nothing, and decide AI search doesn’t matter yet.

I think that’s the wrong conclusion. And here’s why.

Picture the journey. She asks ChatGPT for a recommendation. Your brand gets named. But she doesn’t click the link. Instead, she opens Google and types in your brand name, to check you’re legitimate. She reads a couple of reviews. Then she clicks through to your website and buys.

What does Shopify say? Google organic.

Or she just types your web address straight in. Then Shopify says “direct”.

The AI made the recommendation. Google got the credit.

The same thing happens with Meta, by the way. When someone clicks an ad inside the Instagram or Facebook app, the app often doesn’t pass on where they came from. So a sale your ad created can show up as “direct” too.

Your dashboard tells you the last thing that happened. Not the thing that mattered.

There’s an Australian study that backs this up, in a way. An agency called Optimising looked at more than a hundred Australian businesses. They found that eCommerce visitors arriving directly from AI tools converted at roughly half the rate of organic search visitors. That data runs to September last year, so it’s a little dated.

If AI worked mainly by sending clicks, that would be disappointing. But I don’t think that’s how it works. AI’s job isn’t to send you clicks. Its job is to influence the decision. The sale comes in through another door.

So – what decides which brands get named? Two things have changed this year that you need to know about.

First. In July, ChatGPT’s shopping feature changed how it picks products – at least in the US. A company called Profound tracked about 1.75 million shopping prompts. On the tenth of July, the share of recommendations coming from merchant product feeds jumped from about 8% to about 62% – almost overnight. Most of the brands they were tracking lost a third or more of their visibility within days.

OpenAI hasn’t confirmed that change. But the direction is clear. The structured product data you give these platforms – your product feed – matters a lot more than it did.

Second. A study earlier this year found that 83% of the products appearing in ChatGPT’s shopping carousels also appeared in Google Shopping’s top 40 organic listings. In other words, your Google Merchant Center feed may be doing as much to get you into ChatGPT as anything else you’re doing.

And here in Australia, Google launched its own AI checkout in June. Shoppers can now buy single items directly inside Google Search, AI Mode and the Gemini app. The launch retailers were Bunnings, Kogan, THE ICONIC, Petbarn and Adore Beauty.

If you sell skincare, a major beauty retailer is already inside that checkout.

So the plumbing matters. Your feed, your product data, your Merchant Center. But here’s the thing.

The plumbing gets you considered. It doesn’t get you chosen.

What gets you chosen is what the rest of the internet says about you. Your reviews. Mentions on sites you don’t own. Content that answers the questions your customers are actually asking. And a brand that people already search for by name.

Why this matters now: the cost of paid

Let’s talk about money. Because this isn’t just an interesting shift in technology. It’s a profit problem.

Meta ads are getting more expensive. Triple Whale tracks data across tens of thousands of eCommerce brands. In the twelve months to July this year, the cost to reach a thousand people on Meta went up about 13%.

And we’re heading into the most expensive weeks of the year for paid ads. Last November, the median cost to reach a thousand people was about 21% higher than in October, according to Superads.

Meanwhile, the median return on Meta ad spend across about 35,000 eCommerce brands was roughly 1.86 to one last year. Less than two dollars back for every dollar spent.

Now think about what that means once you take out the cost of the product, shipping, payment fees and packaging. For a lot of brands, the first order from a paid customer is breaking even at best. The whole model depends on that customer coming back.

Every sale that arrives organically is a sale you didn’t buy at auction prices.

It doesn’t show up in your Meta dashboard. But it shows up in your business. It lifts your marketing efficiency ratio – your total revenue divided by your total ad spend. That’s the number that tells you whether you’re actually making money.

So organic doesn’t replace your ads. It’s what makes your ads affordable.

And remember those three clients from the start of this episode. Organic isn’t only about profit. It’s insurance. When an ad account disappears overnight, organic is what keeps the lights on.

I’ll be honest with you. I would love to run my brand without Meta ads, but I’m not there yet.

At Indagare, our Meta ads are there purely for brand discovery and audience engagement. They find new people. And when they’re off, I feel it – in new customers, especially.

But something has changed this year. And I want to show you what.

The experiment I didn’t mean to run

I’ve paused my Meta ads twice in the past year. Neither time was a planned experiment. But together, they tell a really useful story.

The first time was last summer. Four months with no ads.

My Google traffic fell with them. New customers flatlined.

And that told me something uncomfortable. A lot of what I’d been calling “organic” wasn’t organic at all. It was my ads working twice. Someone would see an ad, not click, and then search for Indagare by name a few days later. Shopify called that Google organic. But the ad created it. So when the ads stopped, those searches stopped too.

I suspect that’s true for a lot of brands listening right now. Your Google organic number may be propping up your ads more than you think.

So I spent the year building something that didn’t depend on my ads.

Rewriting product pages around real customer questions, publishing blog content, adding in Pinterest and YouTube organic content, cleaning up the Google Merchant Center feed.

Then in August, I pulled the ads back again for a few weeks.

This time, Google held. It stayed at the same level as August last year – when my ads were running at full strength. And new customers held too.

Now, I want to be careful here. It’s a small brand, and it was a short pause. This isn’t a laboratory experiment. But it’s the first time switching the ads off didn’t take everything else down with it.

And then there are the international orders.

I have never advertised Indagare in the US or the UK. Not one dollar. And yet I’m getting full-price orders from both. Someone on the other side of the world found an Australian skincare brand made with prickly pear seed oil, and paid full price plus international shipping.

That didn’t come from my ads. There weren’t any. It came from being findable.

The system that gets you recommended

So what actually works now?

Here’s what I’ve learned. New customers don’t come from one channel. At Indagare, they come in when three things are working at the same time, and I’m not slacking off on any of them.

Meta ads, doing discovery. Finding people who’ve never heard of us, and planting the brand name so they search for it later.

Klaviyo flows, doing the keeping. Welcoming new people properly, and turning a first order into a second and a third.

And Google – paid and organic. So when someone looks us up, or asks a question we can answer, we’re there.

AI recommendations sit on top of all three. They’re not a separate channel you can switch on. They’re what happens when the rest of the system is working.

So here’s what I’d focus on, in this order.

One – fix the plumbing. Make sure your products are in Google Merchant Center with free listings switched on. Check your product feed properly. Are your titles written the way your customer would describe the product, not the way your warehouse does? Are the attributes complete and accurate? Then check in Shopify which AI shopping channels are available for your store, and switch on the ones that make sense. This is boring work. It’s also the work that gets you considered.

Two – answer real questions on your product pages. Who is this for? Who is it not for? How do you use it? How is it different from the alternatives? Write it in the words your customers actually use. If you’re not sure what those words are, go back and listen to episode 339 on customer research.

Three – earn proof from other people. Reviews. Editorial mentions. Podcast interviews. Comparison articles. Community conversations. The AI trusts what other people say about you far more than what you say about yourself. And honestly, so does your customer.

Four – keep your ads doing discovery. If you do run ads, let some of them do the job of creating awareness, not just chasing the last click. Every person who sees your ad and later searches your name is building your organic presence.

Five – measure differently. Stop expecting AI to show up as a neat line in your dashboard. Instead, watch your branded search trend in Google Search Console – are more people searching for you by name? Look at your impressions alongside your clicks. Add a “Where did you first hear about us?” question at checkout or after purchase, and include ChatGPT or AI as an option. And judge your marketing on MER – total revenue against total ad spend – not just the return Meta reports.

None of this is a quick fix. My first six months were the worst numbers I’d seen in years. That’s the stretch where most founders give up.

But here’s what I believe, having watched this play out in my own brand and across our clients. I don’t know a premium brand that’s growing profitably right now any other way.

Why should you start this now?

We’re heading into the busiest shopping weeks of the year. Over the next couple of months, thousands of people are going to ask an AI tool what to buy as a gift for someone they love.

The brands that get named in those answers this November didn’t start preparing this month. They built that position months ago.

And the same will be true next November. Much like the saying, ‘if you want shade, when’s the best time to plant a tree? 20 years ago. When’s the second best time? Today.’ The work starts now.

Your system will have different gaps from mine. Maybe it’s your product feed. Maybe it’s your product pages. Maybe your organic presence is really just your ads in disguise, like mine was.

That’s exactly the kind of problem we work through in a Brand Growth Strategy Session. We look at your whole system – paid, organic, email and AI – and find the gaps that are costing you new customers and profit. There’s a link in the show notes to book yours.