I had a mildly terrifying thought while setting up my first ad on ChatGPT this week. I was looking at the option to upload a custom audience, cuz that’s awesome, but then it hit me …
What exactly does “audience” mean when the platform is also the thing people tell their tale of woe to … or talk about their customers with?
Because this definitely isn’t Google. Google knows what you searched.
And it’s not LinkedIn … which knows where you work, what your job title is, and that you’ve evidently been a “visionary thought leader” since 2016.
It’s not Meta either. They’ve spent years jengaing together your interests, clicks, behavior, and demographics.
But ChatGPT? People TALK to ChatGPT. It’s their biz bestie, family counselor, tax advisor, and memaw all rolled into one Yass Queen super fan.
Y’all be out there …
- Asking it why your website isn’t converting
- Telling it you’re worried about money
- Asking it to compare CRMs.
- Asking it to write your “you owe me money” emails
And now advertisers can upload customer lists.
Oh. That hits a little different now … doesn’t it?
Let’s say I know Jane.
Actually, that’s generous.
Let’s say Jane downloaded one of my freebies three years ago.
What do I actually know about her?
Well… her name is Jane, and her email address is jane@example.com … annnd we haven’t talked since like ever. Cool. I know nothing about Jane 😭
Fast forward three years and I’m uploading my contact list to GPT’s Custom Audience.
If OpenAI can match the measly amount of info I have on Jane to an eligible ChatGPT user, Jane wins a seat in the audience I’m advertising to. Wild right?!?
Picture Jane at her computer telling her GPT bestie Harold all about the struggles she’s having with clients, money, etc
Ok, slow your roll …
Just because she’s on my list doesn’t mean I suddenly get Jane’s ChatGPT history. Or her conversations, memories, etc. There’s no handy little report saying:
“FYI, Jane hates her job, wants to become a consultant, and spent 47 minutes Tuesday night trying to figure out Keap.”
Thank God.
But here’s where it does get a little bit fascinating …
OpenAI’s documentation says that when Personalized Ads is enabled, ad selection can consider not only the current conversation but selected signals from past chats and memory.
Read that again.
Because that changes the advertising equation. A lot.
- It uses my uploaded list to determine WHO is eligible to see my ad.
- It can use what it knows from Jane’s past chats and memory to decide WHEN my ad is relevant to her.
And I think that second part is hella important.
We’ve spent years targeting people. Traditional advertising audiences are based on …
- Age
- Location
- Job title
- Industry
- Interests
- Search terms
- Website visits
- Past purchases
- Income proxies
- Lookalike audiences
We use all those elements to try to answer some version of … who is this person, and are they the kind of person who might buy my wares?
But that’s actually the wrong question …
What we really should be asking is … Does this person need this thing right now?
Those are VERY different questions.
Jane was probably the perfect demographic match for my consulting offer for the last three years, but maybe she didn’t need me. (As if! For real, everyone needs me)
Then Tuesday happens, and Jane goes down the GPT bestie rabbit hole …
“I’m thinking about leaving corporate.”
“Could I make a living consulting?”
“Help me figure out what I could charge.”
“What do I need on my website?”
“Should I use a CRM?”
“How do consultants find their first clients?”
Jane didn’t magically become a different demographic, but now I’ve got her because her decision state has changed dramatically.
AND that’s the part I can’t stop thinking about. Apparently I took the red pill when I clicked “Custom Audience.”
The next great advertising target isn’t a person. It’s a moment.
“I’m getting leads, but nobody is buying.”
Moment.
“I’m making money, but I’m working constantly.”
Different moment.
Same 48-year-old female consultant in Spring Valley.
Completely different ad.
Because each conversation tells us something demographics never could:
What problem is she trying to solve right now?
And potentially:
How far along is she in solving it?
That makes me think we need a whole new set of questions when creating GPT ads.
Instead of:
Who is my target audience?
I want to know:
What conversation is my perfect customer having with ChatGPT five minutes before they need what I sell?
Seriously. If you market anything, sit with that question for a minute.
Holy Moley … that produces a different marketing strategy!
If I’m selling CRM implementation, I don’t just need an ad for “small business owners.”
I need the ad that makes sense when someone is trying to figure out why leads keep disappearing.
I need another for the person drowning in manual follow-up.
Another for someone researching Keap versus HubSpot.
Another for the business owner who just realized nobody contacted the 237 leads sitting in a spreadsheet.
Those aren’t personas.
They’re stressors … and that makes them decision moments.
And yes, there’s a creepy side.
Because someone will always find a way to make it so.
Any technology capable of understanding intent this well deserves caution.
OpenAI says advertisers don’t receive people’s conversations, memories, or personal ChatGPT data. It also restricts sensitive targeting and puts additional controls around personalized advertising.
But bro marketing (I could literally stop there) has been weaponizing scarcity and insecurity since approximately five minutes after someone invented the Buy Now button.
So we’re going to have to figure out what ethical marketing looks like when the platform serving the ad may understand deeply personal context about Jane’s life, even though the advertiser doesn’t.
Because there is a line somewhere between:
“Show my bookkeeping ad when Jane is trying to figure out how to clean up her business finances.”
and
“GPT … aka Harold … knows Jane is scared, vulnerable, and desperate. Sell her something.”
Where that line gets drawn matters. A lot.
But from a marketing strategy standpoint?
I think we’re looking at the beginning of a significant change.
Search advertising was built around what did they search for?
Social advertising became who are they?
AI advertising is the horizon of what are they trying to accomplish right now?
Which means the smart money is on the marketers who understand their customers’ problems well enough to recognize the exact conversation happening immediately before somebody needs them.
That’s a whole different advertising game, and I’m here for it.
Is AI using what it knows about your current situation to show you ads genuinely useful … or does the fact that it knows why the ad is relevant make the whole thing feel creepy?
About the Author, Michelle Bell
Michelle Bell is a Revenue Systems Architect and Business Literacy Expert who has spent more than 20 years helping entrepreneurs figure out how their businesses actually work — and where the money gets stuck.
With a background spanning finance, regulatory affairs, marketing automation, and business operations, Michelle has a knack for spotting the connection between seemingly unrelated problems and turning complicated business messes into systems that make sense.
Today, her work focuses on business literacy for a post-AI world: helping business owners develop the judgment, systems thinking, and strategic decision-making skills that can’t simply be outsourced to an algorithm.
She’s also deeply curious about what happens when technology changes the rules of business — which is usually where her best rabbit holes begin.