Ai Fixes Catalog Ad Scale · Arie Scherson

I have shared Marpipe as a way to address the challenge of making product ads across huge catalogs. In a September 30, 2025 post, I repeated @ariesnotebook’s scale example: regular ads can showcase 4–6 products max, Shein has 1.3 million SKUs, and Zara adds 36,000 new products annually. Those figures and the argument that individual ads cannot be created for thousands of products were @ariesnotebook’s, not my independent findings.

I described Marpipe as turning catalog ads into designed, dynamic experiences, like Canva for product catalogs at infinite scale. The promoted capabilities included one-click background removal across a catalog, dynamic pricing with strikethrough discounts, and auto-updating review counts. I did not explain how these features work or substantiate the “infinite scale” characterization.

In another September 30 post, I described AI analyzing product data to generate unique ad copy for every SKU, dynamic color extraction matching borders to product colors, and multi-treatment testing without resetting campaign learnings. I said the product works on Meta, TikTok, Snapchat, Pinterest, and anywhere catalog ads run. These product-feature descriptions appeared in a reply to @ariesnotebook; the capture does not provide implementation details for each platform.

I reported Marpipe performance figures for Twillory (127% performance improvement in 10 days), Ridge Wallet (53% ROAS increase), and Simple Modern (52% ROAS boost and 6.6x ad spend scale). I also reported that Marpipe-enriched catalog ads average 8.18 ROAS versus 4.0 for regular image/video ads. The posts give no measurement periods or comparison methodology for most figures, do not explain how the average ROAS was calculated or which campaigns it covers, and do not clarify what “6.6x ad spend scale” means. These are reported claims, not independently substantiated results.

I also promoted a free AI media buying agent described as monitoring ad performance inside Slack. To request it, I told readers to RT and comment “MARPIPE” on the main launch post and said I would send it. The post does not identify the agent’s product name or main launch post, or explain eligibility or delivery timing.

On September 30, 2025, I recommended Marpipe to people serious about scaling ecommerce ads. I said I was happy to partner with them for ShopiLab members, agency clients, and my community, and called it a brilliant product. The post did not specify the partnership’s terms, timing, or scope beyond those audiences. In the same reply, I promoted a free AI media buying agent described as monitoring ad performance inside Slack; I told readers to RT and comment “MARPIPE” on the main launch post to receive it, but did not say whether I or anyone else obtained or used it.

On December 10, 2025, I said Christmas lights were up on a linked platform for the holiday season. The post does not name the platform. It replied to @ariesnotebook, who called portable laptop monitors an unstoppable trend within an unspecified market; that was their assessment, not mine.

On March 31, 2026, I said AI was making ecommerce much more efficient and described Marpipe as one powerful example. I said most brands overlook catalog ads, even though they are among the highest performers based on Meta’s internal data; I gave no supporting details or performance figures. This view connects to my discussion of why ugly catalog ads dominate spend. The post quoted @danpantelo’s announcement of Generative Catalogs, which he described as redesigning a product catalog and DPA in minutes. His post also claimed that ecommerce brands have hundreds of SKUs and spend 50%+ of ad spend on DPA; those were his claims, not figures I independently verified.