On August 15, 2026, I urged people to persist in pursuing what they want in life, using Meta’s repeated promotion of AI enhancements in its ad-launching flow as an example.
I claim Meta wants everyone launching ads to use its AI enhancements, and I predict that Meta will continue pushing AI enhancements on people launching ads until users have no choice.
I have seen Meta’s advertising platform improve through machine learning, matching creative to audiences and producing stronger results over time. I value its ability to deliver the right message to the right customer at the right time.
In one seven-day sequence I described, the results were -$100, -$10, $0, +$10, +$100, +$400, and +$1,000. I said this pattern does not always happen, but I have seen it dozens of times across dozens of accounts and businesses.
On November 21, 2024, I replied to @benvoyooooons that getting results would take longer while Meta learned. I did not specify which results we were discussing or what campaign this referred to.
On March 23, 2025, I argued that Meta’s AI and machine-learning algorithm were improving, and that conversion-optimized ads for in-demand products could act as a customer-generating system for any business. I urged people not to be cynical about Meta ads and said it was not too late to use them. I gave no performance evidence or campaign setup details.
On May 25, 2025, I said Meta’s machine learning allows marketing to one person at scale. In a separate post that day, I described Meta ads as enabling 1:1 marketing at scale: feed the system the right ads, and it will bring the business the right customers. I did not define what makes an ad right, explain the system, or report measured outcomes.
I said Meta’s ability to show the right message to the right customer at the right time is one reason we do so much Meta media buying. In my view, with enough creative iteration, Meta will do this matching for us.
As I learn more about AI and machine learning, I increasingly appreciate how powerful Meta ads are.
I described Meta’s machine-learning algorithm as analyzing an unreal amount of user behavior, demos, and interests to predict which audiences are most likely to engage or convert. I said, “You cannot beat this.”
On June 17, 2025, I called something a “New placement” while quoting @unusual_whales, who said WhatsApp, associated with $META, would start showing ads to users. I did not specify what placement I meant or give details about the reported ads’ timing, format, or rollout scope.
On December 30, 2025, I wrote that Meta ads were going to become “so awesome” and shared https://t.co/efD0mH1MHa. I did not explain what developments prompted my optimism.
On December 31, 2025, I said Meta ads were getting better and better despite what everyone had said to the contrary, and added, “Plato’s cave is everywhere.” I did not provide evidence or a timeframe for my assessment, or explain what I meant by the Plato’s cave reference.
On May 4, 2026, I said I was excited to test the Meta connector for Claude and remarked, “What a time to be an entrepreneur man”. I described an intention, not completed testing, and did not say what the connector does or what I learn from trying it. This interest in testing connects to never stop testing ads.
Related: never stop testing ads, managing meta ad accounts, meta s growth and stock opportunity