How to Turn ChatGPT Into Your Best Salesperson

When we found hundreds of purchases connected to ChatGPT, we immediately started asking the wrong question.

How do you get more?

It seemed obvious.

If ChatGPT was already helping customers discover and buy the product without the company deliberately investing in it...

How do you make that number bigger?

But that question assumes the answer lives inside ChatGPT.

What if it actually lives across the internet?

What does AI need to know before it can recommend you?

Imagine someone asks ChatGPT:

“What’s the best bed frame for a 6'4" side sleeper under $1,500?”

That's not really one question.

Before recommending anything, there are a bunch of things an AI might need to understand.

What products actually fit the criteria?

How much do they cost?

What makes one different from another?

What does the company claim?

Do customers agree?

What happens when people actually use the product?

Is there credible information outside the company supporting those claims?

Suddenly, “How do I get ChatGPT to recommend my product?” becomes a much more interesting marketing problem.

Because your website can help answer some of those questions.

But it can't answer all of them equally well.

Your website tells AI what you want it to know

Your website is still the obvious place to start.

It's where you explain what the product does, who it's for, how it works and why someone should choose it.

For us at Blueprint, that means creating useful content around questions marketing leaders are already asking:

How do I prove brand spend is creating pipeline?

Why does Meta say one thing while GA4 says another?

How do I know where my next marketing dollar should go?

That's part of what we've been building with our own content.

Not because we've discovered some secret format that makes AI recommend Blueprint.

Because if someone asks one of those questions, we want a genuinely useful Blueprint answer available to be found.

But there's an obvious limitation to anything you publish yourself.

You're the one saying it.

If a bed-frame company says its frame is incredibly sturdy, AI has learned something about how the company positions the product.

It hasn't necessarily learned whether the frame is actually sturdy.

For that, it may need another kind of evidence.

Your product data answers a completely different question

Go back to the original recommendation:

“What's the best bed frame for a 6'4" side sleeper under $1,500?”

Before AI can decide whether one product is better than another, it needs to know whether the product actually qualifies.

Does it come in the right size?

What does it cost?

Is it available?

What materials is it made from?

What features does it have?

That's why product data matters differently from content.

A brilliant article about your bed frame doesn't help much if the system can't confidently determine whether the king-size version costs $1,200 or $1,800.

OpenAI says ChatGPT shopping results can consider structured metadata from first- and third-party providers, including things like price and product descriptions, along with other third-party content. Its shopping research can also use merchant product data, publicly available product information and other retail sources.

OpenAI has gone even further by expanding its commerce infrastructure around product discovery. Merchants can provide structured feeds containing current product information such as descriptions, pricing, inventory and media so ChatGPT can better understand and surface relevant products.

So if content helps explain why someone might want your product...

Product information helps establish whether your product actually fits what they're asking for.

Those aren't competing strategies.

They're different pieces of the answer.

Then comes the information you don't control

The website establishes the fundamentals.

But what about what customers are saying?

What about YouTube?

Reddit?

Independent reviews?

Comparisons?

The important insight isn't that you need to go create an account on every platform you can find.

It's that these sources can tell AI something your own website can't tell it as convincingly.

Suppose your product page says:

“The most comfortable premium bed frame in America.”

Now imagine customers repeatedly describe it as incredibly sturdy...

But complain that assembly is a nightmare.

That's richer information.

The company gave AI a claim.

Customers gave it experience.

And the two don't even have to agree.

Because if AI is helping someone decide what to buy, a world in which every available source simply repeats the manufacturer's marketing language isn't particularly useful.

Real customers can introduce tradeoffs.

They can explain what surprised them.

They can reveal who the product seems to work especially well for.

They can describe problems that only become obvious after six months of ownership.

They can also provide evidence that the benefit you keep talking about is something people actually experience.

Your website can make the promise. Your customers can show whether the promise survives contact with reality.

That distinction is easy to miss if you're thinking about AI discovery as just another form of SEO.

Some questions are easier to show than explain

Now imagine the same bed-frame company has dozens of videos across the internet.

One person films the entire assembly.

Another compares it side by side with a cheaper alternative.

Someone else shows what it looks like in a small apartment.

A reviewer comes back six months later and explains what held up and what didn't.

Those sources contain information that's difficult to communicate through another paragraph on a product page.

You can claim something is easy to assemble.

A video can show someone assembling it.

You can publish the dimensions.

A video can show how much space it actually occupies in a normal bedroom.

You can say the frame doesn't move.

Someone can climb onto it and demonstrate what happens.

Again, the point isn't:

You need YouTube because AI likes YouTube.

The more useful idea is that different formats can contain different kinds of evidence.

And increasingly, that information doesn't necessarily live in isolation from search and AI discovery. Google, for example, now lets publishers track how content from YouTube, Instagram, TikTok and X is being discovered through Google Search and Discover.

Your website may be the most controlled description of your product.

It doesn't have to be the only description AI can encounter.

The strongest evidence may be the evidence you didn't create

Suppose your website says your product is the safest option in the category.

Customers say they love it.

Creators demonstrate it.

That's all useful information.

But what happens when an independent publication compares several products and reaches the same conclusion?

Now AI has something else.

Corroboration.

You didn't write the comparison.

You didn't decide which competitors were included.

You don't completely control the conclusion.

That's precisely why it can carry a different kind of informational value.

And it changes the way we think about things marketers have traditionally separated into different buckets.

A customer review might have been considered social proof.

A YouTube demonstration might have been considered content.

An editorial comparison might have been considered PR.

A product feed might have belonged to ecommerce.

A useful article on your website might have belonged to SEO.

But from the perspective of an AI trying to answer:

“Which product should I buy?”

Those may all be pieces of evidence helping it reach a conclusion.

AI may be researching your product, not merely finding your webpage

That's the idea we've been thinking about the most.

Traditional search trained marketers to think heavily about the page.

Someone searches for something.

You want your page to rank.

So you optimize the page.

That still matters.

But an AI recommendation can create a different problem.

ChatGPT's shopping research is explicitly designed to conduct multi-step product discovery using merchant data, public product information and other relevant retail sources before returning a smaller set of recommendations tailored to the user's needs and constraints.

So the question becomes less:

“Did I create the perfect webpage?”

And more:

“If AI researched my company today, would the total information available give it enough evidence to recommend me?”

That's a very different audit.

Maybe your website makes a claim nobody else on the internet appears to support.

Maybe customers love a benefit you barely mention.

Maybe your product information is too incomplete for AI to confidently determine whether you're relevant to a specific request.

Maybe videos demonstrate something far more persuasively than your product page explains it.

Maybe independent sources consistently associate your product with a use case you've never considered part of your positioning.

Or maybe the entire internet is reinforcing exactly the conclusion you want a prospective customer to reach.

That's the information environment AI has to work with.

So how do you make ChatGPT your best salesperson?

Probably not by trying to find one secret ChatGPT ranking factor.

And definitely not by publishing hundreds of pages of content written for robots.

Start with a much more human question:

What would someone need to know before they could confidently recommend your product?

They'd need to understand what it is.

They'd need to know who it's right for.

They'd need accurate information about the things that matter to the buyer.

They'd want evidence that your claims hold up.

They'd probably want to know what actual customers experienced.

They might want to see the product in action.

And the more consequential the purchase, the more likely they'd want some kind of independent validation before confidently saying:

“This is the one I'd choose.”

That's what makes the salesperson metaphor useful.

If you hired the best salesperson in the world but gave them nothing except the copy from your homepage...

How good could they really be?

Give them detailed product knowledge.

Real customer feedback.

Demonstrations.

Independent comparisons.

Credible evidence supporting the claims you're making.

Now they have something to work with.

AI isn't a salesperson you train with a sales script.

But the same principle raises a useful question:

Have you given it enough information to make the case for you?

We're asking ourselves the same question

We don't pretend to have cracked how AI decides which companies to recommend.

We're thinking through it the same way we'd approach any marketing problem where the answer isn't obvious.

Follow the evidence.

Understand the inputs you can influence.

Test what you can control.

Measure what happens downstream.

Our own Blueprint blog is one part of that.

We're taking the questions we hear marketing leaders struggling with and publishing useful answers to them.

But this has made us think beyond the articles themselves.

If someone asks AI about cross-channel measurement...

What does the rest of the internet tell it about Blueprint?

Does the language outside our website reinforce what we're saying on it?

Do customer stories substantiate the claims we're making?

Is there enough independent information for someone researching the category to understand where Blueprint fits?

Our website can tell AI what Blueprint is.

The bigger question is whether the information surrounding Blueprint gives it enough reason to believe us.

And that's probably a useful question for any company to ask.

What would AI conclude about you?

Forget ChatGPT for a minute.

Imagine an incredibly well-informed salesperson researching your company before deciding whether to recommend you to their best customer.

They read your website.

Study the product.

Listen to customers.

Watch demonstrations.

Look for independent opinions.

Compare you with the alternatives.

Then they have to make a recommendation.

What conclusion would all of that evidence lead them to?

More importantly:

Is it the conclusion you want?

That's the audit we'd start with.

Because you can't simply tell ChatGPT to make you its favorite recommendation.

But you can pay attention to the information environment it may have available when making that recommendation.

And then you can start strengthening the parts you actually control.

Find out whether AI is already selling for you

Before you rebuild your entire content strategy around AI...

Find out whether it's already creating customers.

That's what made this question interesting to us in the first place.

Hundreds of purchases connected to ChatGPT were already sitting in the data.

Blueprint helps marketing teams connect the sources influencing customers to the business outcomes that actually matter.

So before you start guessing what ChatGPT wants...

See what it's already doing.

FAQ

How does ChatGPT decide which products to recommend?

There isn't one publicly documented ranking factor that determines which products ChatGPT recommends. OpenAI says shopping results can consider the user's query and context, structured metadata from first- and third-party providers, and other third-party content. Its shopping research can also use merchant product data, publicly available product information and other retail sources.

Can a company pay to rank higher in ChatGPT product recommendations?

OpenAI says the product results in ChatGPT's organic shopping experience are selected independently and are not ads or influenced by OpenAI partnerships. OpenAI separately offers advertising products, but those are distinct from organic product results.

Does my website matter for ChatGPT recommendations?

Yes, but it is not necessarily the only available source of information. OpenAI's shopping research can use merchant product data, public product information and other relevant retail sources, while its shopping results can incorporate structured metadata and third-party content. That means clear, accurate information on your own properties remains important while other publicly available information may add context.

Do customer reviews matter for AI product discovery?

They can. OpenAI says reviews can be among the factors considered for product relevance, and ChatGPT can generate review summaries using reviews from public websites. Reviews can be especially useful because they contain customer experiences, benefits, drawbacks and use cases that may not appear in a manufacturer's own description.

Should marketers create content specifically for AI?

Useful, accessible content can make more information about a company or product available to search and AI systems, but there is no documented secret content format that guarantees recommendations. The stronger principle is to make the product easy to understand, accurately represented and supported by substantive information rather than trying to manufacture content solely for an algorithm.