The most expensive thing in e‑commerce isn't your ad budget — it's backing the wrong product.
Founding 20 brands: build & shape the product with us — and lock in a free or discounted founding tier.
Pick SKUs on gut feel, often producing inventory upfront with zero demand signal.
High risk.Pay rising CPMs just to discover what the market would have told you for free.
Budget wasted.Kill the loser, relaunch a new SKU, burn again. The validation cost lands entirely on you.
Sources: Triple Whale Meta Ads Benchmarks (2025 vs 2024, ~35k brands — CPM $11.82→$14.19, +20% YoY); Modern Retail "DTC Briefing," May 2024 (CAC +25–40%); apparel net margin 3.85% — NYU Stern/Damodaran, Jan 2026; DTC marketing 13–22% of revenue — Warby Parker / FIGS / Allbirds FY2024–25 10-Ks (SEC).
The world's biggest cross‑border retailers run ads on products that aren't even made yet — just 3D renders. If an ad converts, they produce it. If it doesn't, they drop it.
Validate demand first, spend on production second.
That edge is locked away from small brands.
We're changing that.
Top performance marketers now pre‑test before they pour in budget — it's where the whole industry is heading. We take that one step further: we fuse our own models, deep commerce‑ecosystem understanding, live industry signals, and real buying behaviour into one prediction you can run before you even produce — grounded in what people actually do, not what they say.
Reclaim the budget you burn just to find out what sells.
We pull ground‑truth demand signals from across the commerce ecosystem and pair them with live industry trends — revealed behaviour, not stated intent.
Our models combine product attributes — image, category, price, style — with those signals to learn what truly converts, building a "will‑it‑sell" index that ranks candidates within a category and price band.
Query a candidate and get an expected sell‑through estimate and its uncertainty — re‑validated for your audience and price, not a black‑box score.
Stop guessing.Every product a brand validates with us — and every campaign we run for them — feeds back as fresh outcome data. The model post‑trains on it, so each test makes the next prediction sharper. More brands → more real outcomes → an edge competitors can't scrape from the outside.
Here's where we're headed: choose a candidate, see how it benchmarks against the top sellers, get a verdict — then launch it across every platform in one click.
Know before you spend.
⚙️ We're still building this. The flow on the right is a preview of the product, not a live tool yet.
Benchmarks look promising. Run live ads to confirm before you produce.
Illustrative preview — brand names & figures are placeholders.
Trend tools read social and search signals. Spy tools show you other people's winning ads. Neither one captures real demand outcomes for your product before you make it. We do.
| Capability | PROBATUM Demand proofthis is us |
Trend forecastingWGSN · Trendalytics | Ad / product spyMinea · PiPiADS | Creative testingMarpipe · Motion |
|---|---|---|---|---|
| Signal is real demand outcomescausal, not correlational | ✓proprietary outcome data | ✗social / search trends | ~others' ads, not yours | ~creatives, not demand |
| Answers "will this product sell?" | ✓product-level verdict | ~category-level only | ✗what already sold | ✗product must be live |
| Works before you produce or selectno inventory needed | ✓ | ~ | ✗ | ✗ |
| Built & priced for small / D2C brands | ✓accessible tiers | ✗$5k–20k+/yr | ✓$30–100/mo | ~ |
| Apparel-specialized | ✓fashion-first | ✓ | ~ | ✗ |
AI superpowers shouldn't be reserved for the giants. If you're burning budget guessing what to sell — let us test before you spend.
Founding 20 brands: build it with us — and get a free or discounted founding tier.
Stop guessing. Start proving.