Quick math up top so the title isn't doing all the work.
Last Tuesday my one-product store did $6,766 in sales in a single day. That sounds incredible until you see the rest of the row in my spreadsheet. Facebook ate $3,140 in ad spend that same day. The product and shipping ran about $1,700. Payment and platform fees took roughly $300 more. What actually landed in my account was around $1,600. Still a great day. Just not a $6,766 one.
Across all of last month the store cleared about $5,400 in real profit on roughly $38,000 in revenue. That $5,400 is the number I pay rent with. I'm Tariq, 29, and I used to manage a phone store in Columbus until corporate cut everyone's hours last winter.
This is not my first store. It's my fifth. The first four lost me about $2,400 in ad spend over three months and never turned a profit. Two of them never got a single sale. I almost quit after the third one died.
One honest warning before anything else. This is the one thing I do that needs money to start. Most side hustles I'd point a friend at cost nothing but time. This one costs ad spend, and you will lose some of it while you learn what works. If you don't have a few hundred dollars you can afford to set on fire, go do a different hustle first and come back when you do.
One thing before the workflow. I spent about five weeks watching dropshipping tutorials, and every single one contradicted the last on targeting, on budget, on which app to use. A structured course cut through that so cleanly it was almost embarrassing how much faster I learned once someone laid the steps out in order. YouTube is fine once you already know what you are looking for. It is a maze when you do not.
One more thing. The single most important move in all of this is the last section, and it is the least flashy one. I'm saving it for the end on purpose. Don't skim past it.
I stopped asking AI to find me a product
The biggest mistake I made early was opening ChatGPT and typing "find me a winning product to sell." You get garbage. The AI doesn't know what's selling this week. It guesses, and you burn money testing its guesses.
So now I do the opposite. I find products that are already selling, then move them somewhere they haven't shown up yet. TikTok Shop is where I look. If you scroll TikTok you've seen the little orange cart tag under videos. Those are TikTok Shop listings, and a lot of them are quietly doing real numbers.
I use a tool called Kalodata to see the data behind them. It lists TikTok Shop products with revenue, trend charts, and the top-selling videos. I set the max revenue filter to around $300,000 so I skip the giant saturated products everyone already runs, and I hunt for things on a fresh uptrend. Last month it was a rechargeable camping fan doing about $289,000 in 30 days, with videos from this month already pulling thousands.
The logic is simple. TikTok proves the product can sell. People on Facebook and Instagram aren't there to shop, they're there to scroll, so a good video catches them mid-scroll and turns it into an impulse buy. If a product sells on TikTok, it usually translates to Facebook, and Facebook is where most of these products haven't been flogged to death yet.
The store builds itself now
Back when I started you'd lose a weekend building a Shopify store by hand. Now I paste the product link into a one-click builder app, it pulls the images, writes the copy, lays out the product page, and hands me a full store in a few minutes. I recolor it to match the product, fix the price and the compare-at price, swap the buttons that don't work, and publish.
It is not perfect out of the box. I always rewrite a few lines and kill anything that reads like a robot wrote it, because shoppers can smell that. But the part that used to eat hours now eats minutes, and that's the point.
The ads are where Claude earns its keep
This is the piece that actually moves money. I run Claude with a custom skill, basically a script that tells it the exact order to do things, hooked up to Higgsfield for the video generation. I paste the product URL, it builds a customer avatar with pain points and objections, writes a few script options, and lets me pick a creator image. Then it generates a UGC-style video of that "person" holding the product, in short chunks I approve one at a time.
I pull the clips into CapCut, stitch them, add a bit of music and some real B-roll, and keep the voice consistent across cuts. The first draft always has weird pauses and sometimes the script name-drops a competitor, so I edit. But I go from nothing to three usable ad variations in an afternoon, which used to mean hiring a creator and waiting two weeks.
I learned the prompt and skill side from a mix of YouTube and a short course on Mindwand that broke prompting into 15-minute daily lessons I actually finished. The format kept me showing up. Skool and Coursera have similar material if those fit you better. Pick one. Finish it.
Running it without lighting all your money on fire
My ad setup is boring on purpose. One campaign, a few adsets, broad targeting, United States only. I don't target "campers" or "moms" anymore. Facebook reads the creative and decides who to show it to better than I can, so I let the video do the targeting and just feed it three to five different hooks.
Here's the part the hype videos skip. Most of your tests lose money. You are not building a machine that prints on day one, you are running cheap experiments to find the one creative and product combo that works, then pouring budget into that one. I kill anything that isn't profitable within a small test budget and move on. The fan was the fifth product I tried. The four before it are the $2,400 tuition I mentioned up top.
I picked the prompting course on Mindwand because the daily-lesson format fit around my shifts. Could have done Skool. Format matters more than brand, so just pick the one you'll actually finish.
The part I saved for last: the product is the whole game
Everything above is the easy 20%. The AI store, the AI video, the ad setup, you can learn all of it in a week. The product choice is the other 80%, and it's where almost everyone loses.
A brilliant ad cannot save a dead product. A proven product survives a mediocre ad. My first four stores didn't fail because my videos were bad. They failed because I picked products I thought were cool instead of products the data already proved people were buying. The fan worked because TikTok had already shown me, in hard numbers, that strangers were paying $38 for it this month.
So spend your time there. Most of my hours go into Kalodata, not CapCut. If you only take one thing from this, take that one.
So what's the real hourly rate
If I average everything, including the three months of losing stores and all the hours editing videos that never sold, I'm at maybe $14 an hour over the life of this. The good months feel like a lot more and the bad months feel like a lot less. That's the honest version the tutorials leave out. It got real once I stopped guessing products and started reading data, and once I accepted that most tests are supposed to lose.
I'm not going to tell you to go quit your job. I'd tell you to find one proven product, build one store, run one cheap test, and see how it feels to lose forty dollars on an ad before you ever see a sale. If that doesn't scare you off, you might be built for this.