Dropshipping Without the TikTok Cheat Code, a Ninety Day Log
Dropshipping 2021 is dead. I say that as someone who ran a store that year, rode the algorithm wave, and watched it collapse over eighteen months. Ad costs rose, shipping windows stretched, and platforms started downranking stores with forty day delivery complaints. Most people quit. A smaller group, me included, quietly rebuilt around a slower, more product first model. This article is the ninety day log of that rebuild, with the numbers that matter and the rules I refuse to break anymore.
The product filter I actually use
I pick one product per store, not a general catalogue. The filter is five criteria and the product has to clear all five. Landed cost under 15 euros (product plus shipping to me). Perceived retail over 45 euros. A real problem the buyer can articulate in one sentence. A visual demo that fits in a fifteen second clip. EU based fulfilment option, even if it costs more. I looked at roughly forty products before picking one that cleared all five; the previous store had failed on criterion five alone.
Shipping is the whole game now
I switched from a China direct agent to a Poland based fulfilment warehouse. The landed cost went up by 2.40 euros per unit. The delivery window collapsed from 32 days average to 13. Refund rate fell from 9.4 percent to 1.8 percent. Net margin after chargebacks and refunds doubled. The lesson was unintuitive; paying more per unit made the store more profitable. The same counterintuitive rule shows up in my retail arbitrage notes; the cheapest sourcing is rarely the most profitable.
The ad structure that worked
I stopped chasing trend audio. Instead I paid one creator 180 euros for a single 28 second clip, fixed camera, natural light, problem first script. I ran it as a TikTok Spark ad with a 20 euro daily budget for seven days, then scaled the winning audience. Day eight onward the ad set held a 2.4 ROAS for thirty six days before decay. I replaced it with a second UGC clip from the same creator and the cycle repeated. Two clips ran the whole quarter.
Revenue, cost and net over ninety days
Days 1 to 30: 2,100 in revenue, 980 in ads, 640 in product and fulfilment, 820 in returns and chargebacks, net minus 340. Days 31 to 60: 3,400 revenue, 920 ads, 1,020 fulfilment, 280 returns, net 1,180. Days 61 to 90: 2,700 revenue, 680 ads, 820 fulfilment, 160 returns, net 1,040. Ninety day total, 8,200 revenue, 2,100 net margin which works out to 25.6 percent, above my store floor of 22. The first month loss is the cost of finding the ad angle; budget for it.
What nearly killed the store
A single supplier stockout in week five. The Polish warehouse ran dry on my hero product for eleven days and I had already paid ads for that week. I pulled the ads on day three of the stockout, which cost me a trending audience and a 40 percent ad efficiency loss when I rebooted. Since then I keep two weeks of buffer stock in the warehouse and a backup supplier pre onboarded. The cost is roughly 300 euros of locked capital. It is cheap insurance.
Put your refund email on the product page
Most dropshippers hide their refund policy behind a generic link. I put a plain text sentence directly on the product page: 14 days refund, one email to this address, no questions. It looks like a conversion killer. It is the opposite. My conversion rate rose by 17 percent after adding it, because buyers trusted the store enough to put in their card details. The refund rate did not rise, because people who want to refund always find a way.
Do not run a general store in 2026. The conversion rate on a general catalogue against a single product landing page is roughly a third lower on identical ad spend in my tests. The old logic, test products inside one store, is a false economy; product insights do not transfer across niches and the brand halo is zero. Pick one product, build a real landing page, let the store die when the product dies.
Frequently asked
Is dropshipping dead in 2026?
No, but the cheap version is. The five dollar ad, one dollar product, forty day shipping playbook is gone; platforms downrank it and customers refund it. A slower model with a real product, fifteen day shipping, and creator sourced ads still clears a 22 percent net margin on my numbers.
How much capital do I need to start?
I ran mine on 1,400 euros. Roughly 300 for store and domain, 200 for sample products, 700 for ad testing and 200 for reserves. Anything under 1,000 gets eaten before you see a signal. If you do not have the reserve for one failed product test, do not start.
Can I still use TikTok ads?
Yes, but treat it like paid search, not organic virality. Creator sourced UGC with a clear problem statement beats trend hijacking. My winning ad was a 28 second fixed frame clip; it had zero trendy audio and it converted three times better than the dance edits I tested first.