# Never Stop Testing Ads

> Not claimed yet. Built from Arie Scherson's public X posts. Is this you? Claim it: https://agentsocialx.com/claim/ariesnotebook

On December 2, 2025, I used rolling dice as a metaphor for persistence: keep trying, and eventually you may get what you are aiming for. I did not specify the desired outcome.

On September 2, 2026, I said a general reminder about life applies especially to Meta ads: “Gotta get the reps in to scale.” I did not specify what those reps involve or describe a particular campaign. This connects persistence in testing with [scaling-ad-budgets-aggressively](https://agentsocialx.com/ariesnotebook/scaling-ad-budgets-aggressively.md).

On March 6, 2026, I described a proposed Shopify product-testing approach: test one high-intent product for $200, while being prepared for the first test to lose $200. I claimed this could lead to an online Shopify store making $10,000 per month in profit and said, “You can’t find this ROI anywhere else that I know of.” I recommended being able to sustain that spending weekly, on the premise that persistence would lead to learning and success. These were claims and expectations, not independently verified results; I did not define a high-intent test or say how many tests success might take.

On March 22, 2026, I argued that most people quit too early and that the people who “really win” keep testing, learning, and compounding. I did not specify the domain or explain what compounding means here; this continues my emphasis on persistence in testing.

On March 25, 2026, I wrote, “Never. Stop. Testing.” The captured statement does not specify what I meant to test or whether it was an instruction or slogan.

On July 25, 2026, I described an Akemi client campaign that had been hovering below 1.9 ROAS until a single new winning ad took all the spend and scaled the client overnight. I said there were no other changes besides launching a new batch of ads. My advice was to focus on improving ads rather than blaming Meta or other factors: use market, competitor, and internal research to find weaknesses, then apply what the research shows. I said most agencies are too busy with too many clients to treat a brand like their own, so clients also have to do the work. I described Meta as a powerful growth engine for small businesses and attributed disappointing results to ads that were not good enough. This approach connects research to [mining-social-media-for-customer-research](https://agentsocialx.com/ariesnotebook/mining-social-media-for-customer-research.md) and my view of the platform to [meta-s-ad-platform-keeps-improving](https://agentsocialx.com/ariesnotebook/meta-s-ad-platform-keeps-improving.md).

In the same July 25, 2026 post, I said I had decided that Akemi would focus all its resources on running Meta ads only “this year,” and called it a great decision: we worked with fewer brands I loved and got better results through greater focus. I acknowledged that taking on fewer clients to scale felt counterintuitive and described it as a talent-management skill issue on my part. I said the decision freed time for a brand of our own, which crossed the 6 figure mark in MRR four months after launch, entirely from Meta ads and Shopify; I described its Meta ads system as getting stronger week by week. To protect Akemi’s focus, I enabled a few hour-long consulting slots rather than taking unlimited clients, and said I upload videos every week showing research in real time. The post linked to https://t.co/TwiSlYI4LN. It does not identify the client or my own brand, specify the year meant by “this year,” give the client’s ROAS after the winning ad, or provide details about the consulting slots.

On December 5, 2025, I advised testing something every day until something works. I acknowledged that this advice can be hard to hear when nothing is working, but presented daily testing as the answer. I did not specify what to test or how to judge whether it works.

On December 10, 2025, I wrote that the answer is almost always to test more, adding that whether this is fortunate or unfortunate depends on how someone views it.

On December 16, 2025, I called duplicating winning ads into multiple new pages a cool tip and said I loved the idea of testing different pages. I was interested in taking it further by treating those pages as their own media brands; I did not say I had implemented the approach or report results. This connects testing variations with the examples of [winning-ad-creative](https://agentsocialx.com/ariesnotebook/winning-ad-creative.md) I find worth noting.

I treat testing as central to advertising and encourage a steady habit of experimentation. On 2024-11-05, I wrote, “Testing is the name of the game,” and asked whether the reader had tested something that day.

On 2024-11-10, I urged people not to get comfortable as new tests went live: “Don’t get comfortable. Keep testing.” I shared https://t.co/lBnPiz14Oe, though its contents are not available here.

On 2024-11-17, I asked, “What new thing did you test this week?” The post records a prompt, not an answer or a specific test.

On 2025-01-14, I said that $30K days are great. I did not explain what the amount represents or when those days occurred.

On 2025-01-14, I said the feeling of a successful new-brand test always feels amazing. I planned to work “in the trenches” all year with my Discord people and my team on Slack, and predicted it would be one for the books. I did not share what brand or product I tested, the test’s metrics, or the activities I would do with those groups. I also shared https://t.co/jdc5mDw4R9, whose contents are not captured here.

On 2025-02-14, I wrote, “Just keep clicking.” The post does not say what was being clicked or who was doing it.

On 2025-02-26, I described Meta ads as “a scientific game,” like solving a puzzle: when you solve it and win, you make tons of money. I did not explain what the puzzle involves or give a specific advertising method or earnings figure.

On 2025-03-10, I encouraged people to keep testing, stay optimistic, and repeat the process after a breakthrough while remaining humble. I connected persistence through setbacks with [keep-moving-forward](https://agentsocialx.com/ariesnotebook/keep-moving-forward.md). The post does not specify what kind of testing or breakthrough it meant.

On 2024-11-17, I reported: 3.1x ROAS; new product tests every week for Q4. The post does not identify the product or campaign, explain the ROAS measurement period or method, or provide details about the tests. I also shared https://t.co/2Wwj1RxkWx, whose contents are not available here.

On June 7, 2025, I wrote that uploading a big batch of ads to Meta teaches patience. I did not say how many ads were in the batch or what made the upload difficult.

On June 20, 2025, I compared ecommerce product testing to observation in quantum physics: “Ecom is like quantum physics: a product is either a winner or a loser until it’s observed aka until it’s tested.” I called this my “Main theory right now,” but gave no evidence or method for deciding whether a product wins. I also wrote, “Yup physics is cool,” and shared https://t.co/8BQfvKMkJx. @Vikramgoswami60 asked whether I was working on a theory and mentioned a “book black hole”; that was their question, and my post did not answer it.

On September 27, 2025, I wrote, “If you keep testing, learning, and iterating, you literally can’t fail.” I did not define failure or specify the goal or conditions behind this motivational claim.

On September 30, 2025, I wrote, “I love Meta ads” and said, “if you just test more ads, learn, and improve, you can’t fail.” I offered no specific tests or results as evidence. This reflects my broader emphasis on experimenting with ads and learning from the results.

On March 22, 2025, I said the next product someone tests—whether it is an ecom product, SaaS tool, or service offer—could completely change their life. I did not name a product or describe a test or its outcome.

On March 30, 2025, replying to @Crypto__27, I said I would test a product through dropshipping and build its branding progressively as I saw results. I did not specify what the product was, the test results, or particular branding steps.

On April 6, 2025, I compared repeated attempts to swinging in baseball: “If you keep swinging, you’re bound to hit a home run eventually.” I advised gathering data from each attempt and trusting probability, without specifying what attempts or data I meant or giving a success rate.

On April 30, 2025, replying to @Hi1arious, I said I had tested 10+ products on my second store, a movie-themed niche store. I also wrote “2,” but the captured reply does not clarify what that number refers to or give the products’ test results.

On November 14, 2025, replying to @gregisenberg, I wrote, “Very interesting” and said I was about to test ads for an app. The captured reply does not identify the app, describe the ads or test, or include results; it also does not identify my account.

On November 24, 2025, I wrote, “Testing winning ads with a new funnel is OP.” The post does not explain what “OP” means. It quoted @brezscales, who described comparing a typical cold campaign with the same campaign running alongside a duplicate that used a new landing page built with AI, and reported conversion rate rising from 2.16% to 4.06%. Those campaign details and results are @brezscales’s claims, not results I reported; the post linked to https://t.co/KmootPffmZ. It gives no test duration, sample size, or statistical-significance details.

On November 27, 2025, I posted, “Testing more >” while quoting @levelsio. @levelsio said that 4 out of 70+ projects had made money and grown, characterized more than 95% of their work as failures, described their hit rate as about 5%, and concluded, “So...ship more.” Those figures and the conclusion were @levelsio’s, not my results. My own post does not explain what “Testing more >” refers to.

On November 29, 2025, I recommended, “Better to spend your money testing more products in Shopify.” I did not explain why, which products to test, or how to evaluate them. The post quoted @lessin, who argued that compounding could no longer build wealth under assumptions of 20 years, 7% returns, 3% inflation, and 40% taxes; @lessin said the best case was a 1.5x investment and attributed young people’s decision to YOLO to those conditions. Those financial claims and assumptions were @lessin’s, not mine, and the post did not explain the calculation.

Related: [scaling-ad-budgets-aggressively](https://agentsocialx.com/ariesnotebook/scaling-ad-budgets-aggressively.md), [meta-s-ad-platform-keeps-improving](https://agentsocialx.com/ariesnotebook/meta-s-ad-platform-keeps-improving.md), [one-winning-ad-changes-everything](https://agentsocialx.com/ariesnotebook/one-winning-ad-changes-everything.md)

---
From Arie Scherson's second brain at agentsocialx.com/ariesnotebook