I tested 12 products in 30 days. Here is what actually worked.

Between February 28 and March 30 I ran twelve product tests on instxnt. One $50 budget per product, Meta and TikTok ads, single-product storefronts. The point was not to make money — it was to figure out which kinds of products show signal early and which do not. Here is what happened, with the numbers.

The 12 products, ranked by outcome

1. Magnetic phone mount for car vents. $24 retail, $4.20 from CJ. Won. 19 purchases, $1.74 CPA on Meta. Scaled to $200/day for the rest of the month.
2. LED strip with app control. $32 retail, $9 from CJ. Marginal win. 8 purchases at $5.30 CPA — break-even, then better with retargeting. Kept running.
3. Stainless steel insulated water bottle. $28 retail. Marginal. 4 purchases, mostly from a TikTok creator I sent the link to. Killed Meta, kept the TikTok handoff.
4. Silicone non-slip oven mitts. $18 retail. Two purchases. Killed.
5. Air freshener clip for cars. $12 retail. Two purchases. Killed.
6. Foldable pet bowl set. $22 retail. One purchase. Killed.
7. Reusable produce bags (10-pack). $19 retail. Zero purchases at $50 spent. Killed.
8. Magnetic measuring spoons. $16 retail. Zero. Killed.
9. Stretchy silicone food covers. $14 retail. Zero. Killed.
10. Self-stirring mug. $24 retail. Zero. Killed (TikTok had clearly already burned this one out two years ago).
11. Cat grooming glove. $20 retail. Zero. Killed. Ad creative was bad — I take the blame on this one.
12. "Smart" notebook with reusable pages. $34 retail. Zero. Killed. The price point was wrong for the audience I targeted.

What I noticed

Winners had a sub-3-second "why I want this" moment

The magnetic phone mount and the LED strip both had ad creative where you understood the product and the use case before three seconds. The car-vent mount: hand sticks phone to vent, phone stays. Done. The LED strip: bedroom looks dark, phone tap, bedroom looks like a Netflix show.

The duds had ad creative that needed explanation. The smart notebook ad spent six seconds showing someone writing in a notebook before revealing the trick. By then the buyer was gone. I should have known better. The good news is bad ad creative is fixable; bad product–audience fit is not.

$50 was enough to detect failures, not enough to optimize wins

Of the three products that worked, $50 told me they had a pulse. None of them were truly optimized at that budget. The phone mount, for example, jumped from $1.74 CPA to about $0.95 once I ran a $300 second-round test with new audiences. Another seven days at higher budget and the CPA settled around $1.10 — which is where Meta's algorithm finds the cheap conversions.

The lesson: $50 is a screen, not a verdict. Spend $50 on every candidate. Of the ones that survive, spend $300 to verify. Then scale the verified ones.

Free-tier instxnt was the right call for screening, Pro paid off after

For the screening round I stayed on the free plan. 3% per sale plus Stripe — the math works out to about 6% all-in on each sale. On twelve test storefronts that is $0 in monthly cost even when most of them sold nothing.

After the magnetic phone mount validated, I upgraded that storefront's account to Pro at $19/mo. The 3% platform fee disappears and a custom domain becomes available. At $200/day in ad spend converting at $1.74 CPA, the 3% on the free plan would have cost about $20/day — Pro paid for itself in about a day. If you are at any meaningful sustained volume, Pro wins. If you are still screening, Free wins.

Meta beat TikTok for me, but not by much

I expected TikTok to crush Meta because the audience is younger and the ads feel more native. In practice Meta won 9 out of 12 head-to-heads on CPA, which surprised me. I think the reason is iOS tracking — Meta's Conversions API is mature, TikTok's is catching up but not there yet. The Meta numbers I see are closer to truth than the TikTok numbers.

If you are starting fresh and want one platform to learn deeply, my pick is Meta with the Conversions API enabled. The setup guide covers the pixel + CAPI part.

What I'd do differently next month

1. Pre-screen on TikTok organic before paying for ads. Three of the four products that worked already had viral organic posts I could find. Three of the eight duds had nothing. That is a free signal I should have used.
2. Standardize the storefront copy template. I tweaked copy for each product and the results were noisy because of it. Better experiment: same template, same structure, only the product changes. Then the variable you are actually testing is the product.
3. Kill at $30, not $50. Half my zero-conversion duds showed zero at $20 spent. I burned another $30 hoping for a turn that does not come. Defining a kill rule before launch and sticking to it would have saved $150-200 across the month.
4. Run two creatives per test, not one.Cat grooming glove had a bad creative; the product might have worked. I will never know. Two creatives per test costs another $25 per product but is the difference between "product is bad" and "creative is bad".

The full month, in numbers

• Total ad spend: $623 (12 tests × ~$50, give or take).
• Total revenue from tests: $748 (most from the phone mount).
• Net of product COGS and shipping: ~$110.
• Net after Stripe and instxnt 3%: ~$48.
• Hours spent: ~22, mostly on creative iteration.
• Net per hour, ignoring scale-up phase: $2.18. Bad.

On its own the screening round was barely profitable. The point of the screening round though was not the revenue — it was the phone mount. That product is now doing $4-6k/month in revenue at decent margin, found by spending $50 to discover it. The other eleven $50 bills were tuition.

If you are running this experiment yourself, the playbook is in the $50 validation post and the storefront flow is on the dropshippers landing. Send me your numbers when you finish a round; I want to see how my pattern compares.