1/ I run an e-commerce team in Shenzhen selling on Amazon, TikTok Shop, Walmart, eBay and DTC. Over the past year AI changed how we make almost everything. Here's what we actually use it for 🧵
@georgeclem My guess: Amazon management sells "protect what you have", TikTok Shop sells "new growth". Brands pay faster for upside than for maintenance.
@ashley_wright This is the part people miss. A video posted Nov 5 has 2 weeks to get pushed and collect comments before peak traffic. A video posted on Black Friday is competing with everyone at once.
@jordantwestecom Agree. 5 creators who actually use the product will out-sell 500 who got a free sample. Love shows up in the first 3 seconds of the video, and buyers can tell.
@maverickecom Volume is the easy part now. The bottleneck is knowing which 5 of the 550 to put spend behind. A simple rule that helps: AI makes the variations, real customer reviews pick the angles.
@NotZainAgain The timing math is the whole game with seasonal: sample shipping + creator posting window means you're really launching 3-4 weeks before the holiday. Most brands plan backwards from the holiday date instead of from the creator's posting deadline.
Ad teardown: David Yurman's holiday campaign "Share the Warmth" (Nov 2).
What sellers can steal:
1. Sell the moment, not the product: friends, snow, a card game.
2. Post behind-the-scenes clips next to the hero ad.
3. Point every ad to a gift guide.
No celebrity budget needed.
TikTok Shop tip for Black Friday:
Don't send creators a script. Send them the product, 3 real customer complaints it solves, and one rule: show it working in the first 3 seconds.
Scripts make every video look the same. Real problems make creators sound like buyers.
A Temu shopper ordered a bookcase this weekend and got a palm-sized dollhouse model. The listing said "1:12 miniature". Nobody reads that.
Amazon lesson: show scale in your images. A hand, a person, a ruler. If buyers can misread the size, some will, and you pay in returns.
@alex_hrtley Good catch. Fees are the silent margin killer in a sale. Add payment fees and return costs, and the real break-even ROAS is even higher than most people think.
1/ I run an e-commerce team in Shenzhen selling on Amazon, TikTok Shop, Walmart, eBay and DTC. Over the past year AI changed how we make almost everything. Here's what we actually use it for 🧵
@juokaz Makes sense for both. On-site agents use data they already own: reviews, returns, purchase history. Off-site agents have to trust whatever the listing says. For sellers, the takeaway is the same either way: clean, accurate product data is now the storefront.
@alex_hrtley Good point. Cheap AI UGC usually breaks after second 5: the hook works, then the middle has no reason to keep watching. What helped us: write the script first, then use AI to make the shots. A good script with AI visuals beats random AI clips every time.
@zmbnski Agree. The other risk is creative fatigue. In BFCM everyone floods the same auction with discount ads, so CPMs rise and the same creative burns out faster. Having a second channel ready is insurance, not just extra reach.
@Enuffles The IP complaint is the one I'd turn into a system. Before adding a new product, we check the brand's trademark and listing history first. Ten minutes of checking up front is a lot cheaper than a frozen listing in Q4. Solid month either way.
@MyAmazonGuy One thing worth adding for Q4: run the FBA vs FBM math again in October and November, not just once a year. Peak storage fees and slower check-in can flip the answer for slow movers, while fast sellers usually still win on FBA because of the Prime badge.
Black Friday deals started this weekend, 4 weeks early.
Question for sellers:
Are you running one long sale, or saving your best offer for the actual week?
We're a lean team in Shenzhen. We test offers early and only scale what works.
What's your plan this year?
Sunday question for founders:
If you could hand ONE job to an AI employee tomorrow, which would it be?
A) Product photos + short videos
B) Listing copy
C) Customer service replies
D) Ad reporting
Reply with a letter. I will share what we automated first and why.
@EcomSucker Exactly. Platform ROAS tells you what each channel claims. Blended MER tells you what the business actually earned. When the two disagree, trust the bank account.
@MoritzBohmbach Ops, honestly. Creative gets the attention, but the biggest time savings came from the boring parts: daily reports, listing updates, first drafts of support replies. That frees our people to spend more time on creative, not less.
@Robin_ecomm Agree. Your best Meta ads already tell you which angles and words buyers respond to. Reuse that language in Shopping titles and search ad copy. You're not starting Google from zero, you're starting with proven messaging.