A Marpipe alternative for creative testing on Meta

Marpipe sells catalog advertising. Its homepage describes itself as "catalog advertising superpowers for the agentic era" and promises full control over Dynamic Product Ads. The product menu has five items: Catalog Creative, Product Level Video, Generative Catalogs, SKU Optimization and Feed Management (marpipe.com, checked October 19, 2026).

Catalog Creative is a design canvas for feed-driven ads. Templates pull text, images and prices from the product feed, and if/then rules change the design by category, price or inventory, for example showing a sale badge when the discount is over 20%. Product Level Video turns each SKU image into a short templated video; the page says "we don't need to use AI for this" (Product Level Video). Generative Catalogs applies image models to a feed for lifestyle backgrounds and upscaling, with video listed as coming soon (Generative Catalogs).

SKU Optimization is free and does something advrt doesn't: it removes the bottom 30% of products by monthly sales, or high-return SKUs, from catalog ads (SKU Optimization). Feed Management sends tailored feeds to any channel that accepts a feed URL (Feed Management).

Marpipe launched as a multivariate creative testing tool. Its September 2020 launch post described a freemium self-serve product that uploaded creative into Meta Ads Manager and reported per-asset results within about a week (Marpipe blog, 2020). An August 2022 post explained the test structure: "each ad variant is in its own ad set," so Marpipe could "control for an even spend across all variants" (Marpipe blog, 2022).

In October 2026 we couldn't find that product on Marpipe's current site. The old multivariate testing URL returns a 404, the help center's categories are all about catalog ads, and the one place "creative testing" appears is the Catalog Creative page: "This is creative testing for catalog ads," with a line about testing unlimited variants at once. That page doesn't describe ad sets, budgets or how a winner is chosen.

That matters if you searched for a Marpipe alternative because you liked the old idea: equal spend per variant so every creative gets a fair shot. advrt uses the same structure for concept ads, one ad set per creative with an equal daily budget, and adds a published rule for when to stop. Many pages that rank for this search still describe Marpipe as a testing tool, so check its current site before you compare.

advrt builds one Meta campaign per test with one ad set per creative, gives every ad set the same daily budget, and creates it all paused. Nothing spends until you confirm the budget and switch it on. Targeting is by country and ages 18 and up, so the creative is the main thing that differs between ad sets.

Before launch advrt records your ad account's last 30 days as the baseline. If the account had at least 10 purchases in that window, each ad is judged on purchases against the account's cost per acquisition. With fewer purchases it falls back to click-through rate and needs at least 1,000 impressions before it can call anything. Results sync daily.

Catalog ads and concept ads do different jobs. A Dynamic Product Ad shows the product a shopper viewed or a similar one, usually to people who already know the store. A concept ad has to stop someone who has never heard of you, so the angle, hook and visual matter more than the price badge.

If most of your Meta spend goes to Advantage+ catalog or retargeting campaigns across hundreds of SKUs, Marpipe's per-SKU design rules and feed tools are aimed at that. Its Startup plan covers 500 SKUs and 3 live designs on one channel, and Enterprise removes the channel and design limits (pricing page, October 2026).

If your growth depends on prospecting with a few hero products, the bottleneck is usually new concepts and a fair way to test them. That is what advrt is built for: import a product from a URL, Shopify (up to 250 products per run) or a CSV, get 3 to 6 angles, make the ads, and test them.

Marpipe publishes its prices. As of October 2026 the Catalog Creative plans are Free ($0, design previews and un-enriched DPA), Startup at $199/mo (1 channel, 1 market, 3 live designs, 500 SKU cap, daily sync) and Enterprise at $999/mo billed monthly, or $999/mo plus $0.50 per SKU on yearly billing. Extra SKUs are sold in blocks of 200 for $100, and seats are unlimited (Marpipe pricing).

advrt has three plans: Free (5 ad generations a month, 1 brand), Basic at $19.99/mo (100 generations, up to 5 brands, live tests, publishing) and Pro at $49/mo (unlimited generations, up to 50 brands). Neither tool's price includes your ad spend.

For a small store testing concepts, the gap is $199 against $19.99 a month. For a store with 5,000 SKUs running catalog ads on four channels, advrt simply doesn't cover the job, so price isn't the deciding factor.

We computed this table from the verdict code advrt runs (lib/loop/judge.ts), not from a benchmark. It shows how much one creative has to spend before advrt can call each outcome, at three account CPAs. Multiply by the number of creatives for the whole test.

Two things stand out. First, a dud is cheap to find: with zero purchases advrt calls a loser at about 2.3× CPA, so at a $40 CPA that's $92 per bad creative. Second, an early winner has to be dramatic. To be called after 3 purchases, an ad has to beat your CPA by more than half. A creative that's only 20% better will usually run to the stop point and come back inconclusive, which is the honest answer at small budgets.

For a 5-creative test at a $40 CPA, the worst case is 5 × $200 = $1,000 if nothing is decided. Marpipe's 2022 approach had the same equal-spend idea, but we couldn't find a published stopping rule to compare against.