thanks to ai I made a product that allows you to see which ads are gonna convert better without spending a dollar
It’s pretty immaculate how good it is.
Time to go make (or save) money!!
Comment down below and I’ll give you more credits;)))
@scaling01@zephyr_z9 thank god someone else fucking agrees with me. this shitty ass paper gets attention once every few months and i have to tell people that despite what the paper says simulated survey responses do not match real world purchase behaviors lmfao
If there was some correlation, why do you not see every in-house marketing company on earth trying to do this with their own product. If you had any sort of critical thinking you'd have arrived at this already.
And also, had you ever thought to try it on your own data? if it was true, surely they'd have released the files for you to test it out and see if it's possible. Alas, it's not.
It’s irritating that people still bring up this paper.
If you’ve ever tried mapping this paper out to other industries ie., advertising, you will find there is actually no correlation between the two. Purchase intent is off of *survey based questions* in the paper.
When you’re in a survey and asked would you buy x, you’re far more likely to say yes when there is nothing at stake.
Say the same thing about travel. Would you travel to x with your friends? Hypothetically I would love to go to Cabo, but alas in real world behaviors life gets in the way and suddenly my friend has a kid on the way.
These likert scalings are cool, but map it to real world data and it does not work.
However there is some interesting stuff you can do with llms and ads if you want to know;)
A toothpaste company has quietly killed the entire market research industry and nobody is talking about it.
Colgate published a paper showing you can predict real purchase intent by simply asking LLMs to roleplay customers.
And this is beyond insane.
If you ask an AI, "Rate this product from 1 to 5," it gives safe, middle-of-the-road garbage.
So researchers invented a method called Semantic Similarity Rating (SSR).
Instead of asking the AI for a number, they asked it to roleplay.
They gave the LLM a demographic profile. They showed it a product concept. And they asked it to write down its raw, unfiltered thoughts.
Then, they used a semantic model to translate those written thoughts into a numerical score.
The results are staggering.
Tested against 57 real personal care product surveys and 9,300 human responses, the synthetic consumers hit 90% of human test-retest reliability.
They perfectly mirrored how different age brackets and income levels react to price changes.
And they provided detailed, qualitative feedback that was deeper and more critical than what actual humans wrote.
This destroys the economics of traditional market research.
You don't need to wait a month to see if a product will sell.
You can simulate 1,000 hyper-targeted customer interviews overnight.
You can A/B test pricing across every demographic instantly.
@RubberDucky_AI yea i do sell an ad related product, and i'm stating the facts that survey based questions do not correlate to purchase behavior outcomes
if you sit completely still until the urge to check your phone dies, then wait three more minutes until the urge to fix your life dies too, something much older than your biography wakes up behind your ribs and starts looking out through your eyes like a landlord inspecting a house he forgot he owned
if I was a datacenter, i'd probably call myself an energy innovation center rather than something scary that makes jobs pointless and puts our livelihoods at stake.
but i'm not a datacenter
@collin_ruth89 i think i've received over a 100 plus emails since signing up for their platform back in august about their updates. I saw 0 roi on marketing there. it was also very primitive back then
@FlippyMeister wish i could tell you. I'm pretty sure they write out a strategy doc of that says they will do x now to help convert better after so many days of publishing. Pay for it now, if it shows results good, if not, money back guaranteed?
I figured out a great way to have Codex audit your AI mentions and AI visibility, without paying hundreds (or thousands) for AEO/GEO tools.
First - have an agent come up with a bunch of questions that someone might ask ChatGPT to find / recommend you (theoretically) or your competitors.
Then - you have Codex run a bunch of sub-agents without memory — lead agent prompts a bunch of subagents to use search and answer the questions / give recs.
(i.e. “I’m trying to do X, can you recommend a good Company/Tool/Person for Z?”)
From my testing, it actually gets you basically the same recommendations as ChatGPT.
Next - ask Codex to look at what the agents searched. Not what they recommended yet, start with what they searched.
Important context here from my observations — VERY frequently, GPT (and Claude) actually already search up the name of a brand/individual, without you mentioning it. So for example, if you were to ask GPT to recommend an online bank for startups, the search could be something like “best startup banks Mercury” and I’ve found that this happens really often across categories.
Have it gather that list for you and the frequency.
Then, ask Codex go through the top final recommendations from each agent, see who got recommended and how often. Gather a list of the sites / pages being used as sources, and have agents assess the site’s reputation / credibility.
If it’s a pretty new domain or a random-ass site being used as the source, there may be an opportunity.
Also have Codex pull the titles of those pages (i.e. Top Fintech Platforms for XYZ) and tell you the the nature of the page — i.e. is it a case study? A rankings site? An interview?
From there, you should be able to get a clearer idea of where there may be opportunities for you to optimize for AI search discoverability. Typically it’s about writing content.
And lastly, if you do make any changes, you can have Codex run the same prompts and track any changes in your mentions/visibility.
I ended up having Fable build a desktop app for me to do this and make sense of the data more easily, which I would highly recommend.
>wakes up, sees 5 slack messages missed
>sees nda
>client says, 'please sign nda.'
>read nda
>sign nda
hell yea today is gonna be a good day
it's not a huge number, but even still it's pretty striking how many of the most successful people I know actually sold those cutco knives when they were in high school / college