DEV Community: Sol Research Desk

How to Pay a Chinese Supplier Safely - Every Instrument Ranked, and the 4 Structures That Protect You

Sol Research Desk — Mon, 14 Sep 2026 18:38:36 +0000

You verified the license. The factory is real. Now the scary part: a five-figure sum crossing a border into an account you've never seen, for goods that don't exist yet.

Here's the honest rule: every payment to China is, structurally, an unsecured loan. The question is never "is this 100% safe" — it isn't. The question is which combination of instrument + structure + timing minimizes the amount at risk at any given moment.

The safety hierarchy (best to worst)

Tier	Instrument	Your exposure	Best for
1	Alibaba Trade Assurance	Platform escrow + dispute lever	First orders via Alibaba
2	Letter of Credit (L/C)	Bank-conditional release	Large orders, $30k+
3	T/T 30/70, balance after QC	30% at risk	The workhorse structure
4	T/T 30/40/30 staged	30% at risk, staged leverage	Custom/tooling-heavy goods
5	T/T 100% upfront	Everything, forever	Nobody. Ever.
6	Personal account, any %	Everything + laundering risk	This is the scam, not a payment

Two readings worth memorizing: Trade Assurance beats nothing, but beats only what you're willing to arbitrate — the dispute mechanism works, slowly, for goods that were shipped wrong. And L/C beats everything at scale, but factories under ~$500k annual exports often can't handle the paperwork — offering one is also a decent proxy test of whether you're talking to a real exporter.

The workhorse: T/T 30/70, done correctly

30% deposit — triggers production. Paid only after the PI (proforma invoice) names match your four-name cross-check: license = PI = bank account = storefront. The bank account must be the company's account — the account name on your wire confirmation should read exactly like the entity on the license. A "factory" asking you to pay a personal account, a different company, or a "finance office" is not a factory with a payment problem — it's a scam with a factory attached.

70% balance — released only against documents, not against calendar days. The trigger should be written into the PI: balance payable after passed third-party QC report (book the inspection before you pay the deposit; the date is leverage) plus copy of the bill of lading / forwarder receipt. Never "balance before shipment" in the vague — define the exact document.

Why suppliers accept this: it's the Chinese market standard too. A real manufacturer asking for 100% upfront from a new overseas buyer is either financially distressed or fictional. Both are disqualifying.

The upgrade: 30/40/30 for tooling-heavy orders

When molds, tooling, or custom engineering are involved, money moves before anything exists. Structure it:

30% against tooling agreement (mold ownership explicitly yours)
40% against first-article approval — a physical sample you approve before mass production runs
30% after passed QC + BL copy

The first article is the cheap moment to discover the mold is wrong. Once you've paid 100%, every defect is a negotiation.

The four scam patterns (all payments-related)

The account switch. Quote, PI, and relationship all with Company A; payment instructions to Company B ("our export office," "the group's finance arm"). Sometimes it's genuine group structure; you verify that on the license registry. More often it's a trading company laundering the order, or worse.

The personal account. "Our company account has FX restrictions, please pay our manager's card." Foreign-exchange friction is real in China — and irrelevant to you. Companies with real export history have solutions. Pay a personal account and you've paid a person, not a company: no contract counterparty, no recovery path.

The rush discount. "Pay 100% today, I'll lock the material price." Real price locks are written into contracts with dates and quantities. Fake ones have deadlines. The urgency is the product — once you feel it, stop.

The balance-before-shipping squeeze. Goods pass QC, then a "customs fee" or "documentation problem" appears that requires the balance before the BL is issued. At this point you have leverage you're about to give away: the goods physically sit with a party who wants to be paid. Re-read the PI, invoke the document trigger, and never fix a document problem with money.

What 2026's tariff turbulence changed

Post-de-minimis imports reprice the whole calculus: the tariff line item on your landed cost just got bigger, which makes freight consolidation and DDP quoting far more attractive — but it also multiplies the number of "fees" a fake intermediary can invent at the payment stage. The structures above haven't changed; the surface area for invented fees has. Document triggers beat calendar promises in every year.

The bigger machine

Payment safety is one stage of a six-step supplier verification system — license cross-checks, the red-flag scan, sample protocols — all in the China Sourcing toolkit. The complete template pack, including PI language with document triggers written in, is the China Sourcing Playbook.

Structure the payment like you can lose it — because until the goods are at your door, you can.

How to Order Samples from China Without Getting Played - 3 Types, 4 Traps, and the 4-Email Sequence

Sol Research Desk — Mon, 14 Sep 2026 16:20:37 +0000

A supplier quotes $2.80 FOB, sends beautiful photos, and answers every email within an hour. You wire a 30% deposit on a 5,000-unit order. Then the updates stretch from days to a week, the spec drifts, and the container that arrives is "close enough" — which in retail means unsellable.

Here's what experienced importers know that the brochures don't say: the sample stage is not about the product. It's about watching how the supplier behaves when the stakes are small — because that behavior, scaled up, is exactly what your production order will feel like.

This is the full playbook: the economics, the three sample types, the traps, and the email sequence that converts a sample into a written specification. (Free interactive version with the email templates: Sol Research Desk — order samples guide.)

The economics: who pays for what

Cost	Who normally pays	What a deviation tells you
Stock sample	Buyer, often free or $5–30	"Free sample!" costs them little — friendliness, not generosity
Customized sample	Buyer, $30–200+	Refusing to quote it at all = they're a trader with no line access
Courier (DHL/FedEx)	Buyer	Insisting on their courier account with no tracking visibility is a flag
Sample fee offset	Commonly credited against bulk	Worth asking, in writing, before — not after

The rule that matters: pay sample costs through traceable channels (PayPal, Alibaba Trade Assurance, card) — never a wire to a personal account "just for the small amount." Small-money payment requests are a classic trust warm-up for the big deposit later. If the sample payment pattern looks like a rehearsal for a bad wire, treat it as one.

The three sample types (and what each is actually for)

1. Stock sample — what they already make. Purpose: baseline quality and communication speed. Weakness: proves nothing about customization or your spec.

2. Customized sample — made to your spec (logo, material, dimensions). Purpose: tests engineering comprehension. The gap between your drawing and their execution is the single most honest data point you'll get before production.

3. Golden sample — the sealed, signed reference both sides agree defines "acceptable." Purpose: the referee. When the bulk order arrives off-spec, the golden sample converts "looks different to me" into an objective dispute with evidence.

The sequence is not optional: stock → custom → golden. Skipping to custom with an unverified supplier wastes the fee; skipping the golden sample before a bulk wire is how entire containers become arguments.

The four traps

The photo sample. "We'll send photos instead." A real sample costs a real courier fee; a photo costs nothing. Photos are fine in addition, never instead. Reply: "Photos appreciated — please also send the physical sample; we'll arrange the pickup on our DHL account." Arranging your own pickup removes their last excuse.

The bait sample. A perfect sample from line A, then production runs on line C. Counter: the golden sample plus a first-article inspection — check 5–10 units of the bulk run against the sealed sample before the rest ships. State it in the PO: "Bulk must conform to the sealed golden sample; first-article photos required before final balance payment."

The silence after payment. Sample fee paid, then the update emails stretch from days to a week. You just paid for their responsiveness test — and they failed it while the money was small. Cheap lesson. Buy it now, don't buy it at deposit size.

The "no problem" customization. If a customized sample comes back with your logo but the wrong material and dimensions unacknowledged, that's not a quality problem — it's a comprehension problem, and it does not fix itself at 5,000 units. That's what decoding their replies is for.

The 4-email sequence (each message has one job)

Message 1 — the request. Spec, quantity, deadline, and the magic question: "Is this from your own production line, or a partner factory's?" The reply's hedging tells you which you're talking to.

Message 2 — the terms. Sample fee, courier, offset terms, payment method. Get all three in one written reply, not three. A supplier who can't answer a three-part question in one email is showing you their production-update style.

Message 3 — the spec sheet. Send your tolerances and acceptance criteria before they make the sample. This turns the sample from "their best effort" into "a compliance test." It also, incidentally, tests whether they read.

Message 4 — the golden sample letter. On receiving a good sample: "We confirm this sample as the golden standard for the bulk order. Please sign and return; keep one sealed copy on file." Simple, dated, countersigned — this two-line email is the cheapest quality insurance in the entire import stack.

The bigger machine

This is one stage of a six-step supplier verification system — the rest (license checks, payment safety, the red-flag scan) is in the China Sourcing toolkit. The full template pack behind this guide, including the 4-email sequence pre-written for copy-paste, is in the China Sourcing Playbook.

Order the sample like the deposit depends on it — because it does.

I Built a Landed-Cost Calculator for China Imports - The 7 Costs Hiding Between FOB and Your Doorstep

Sol Research Desk — Sat, 12 Sep 2026 17:50:13 +0000

A supplier quotes you $2.50 FOB. You do the math — sell at $9.99, that's a 4x margin. Then the goods actually arrive and your real cost is $3.24 per unit. Your margin just quietly lost 30%, and you signed the PO a month before you found out.

FOB means Free On Board — the supplier's cost ends when the crate passes the ship's railing in Ningbo. Everything between that railing and your warehouse is your problem. Most first-time importers can name two of those costs. There are at least seven.

The 7 costs hiding in your FOB quote

1. International freight. LCL sea freight runs $40–80 per CBM to the US West Coast, but for shipments under ~200 kg, air express (DHL/FedEx at $6–9/kg) is often the cheaper and faster choice once you factor LCL minimums. Getting this wrong on small orders costs hundreds.

2. Cargo insurance. 0.3–0.5% of cargo value for all-risk cover. Skippable, until it isn't — one soaked container pays for a decade of premiums.

3. China-side documentation. Export customs declaration, booking, and paperwork are usually bundled into your freight forwarder's quote: $50–120 per shipment. A Certificate of Origin is often +$30–60 if your destination country has a trade agreement with China (ASEAN, RCEP — worth checking before you pay full duty).

4. Import duty. Classified by HS/HTS code. Common consumer goods run 0–12.5% into the US — but Section 301 tariffs stack an extra 7.5–25% on many China-origin categories. Furniture and lighting get hit hard; plenty of electronics sit at 0%. The EU adds its own schedule. This is the single biggest variance item, and the reason "just estimate 10%" kills margins.

5. VAT / sales tax. The EU charges 20–21% VAT on (goods + freight + insurance + duty) — yes, tax on the duty you just paid. The US has no federal equivalent, but state sales tax may apply depending on your nexus.

6. Destination charges. Customs brokerage and last-mile delivery: $75–150 per entry plus delivery. LCL shipments add destination terminal handling that nobody mentions until the invoice arrives.

7. Quality inspection. A third-party inspector (QIMA, SGS, or a local bureau) runs $150–320 per manday at the factory. Skip it and Category 7 becomes "the cost of the shipment you threw away."

Plus the quiet eighth: currency slippage of 1–3% between quote, payment, and resale — and 30–45 days of working capital locked in a container on the Pacific.

Run the numbers before you wire anything

I built a free landed-cost calculator that walks through all of these line items and spits out your true per-unit cost:

https://assassinationss.github.io/calculator.html

Enter your FOB unit price, quantity, freight mode (express/air/LCL sea), destination country, and an HS-code duty rate. You get the full stack — freight, insurance, documentation, duty, tax, destination charges, inspection — as a per-unit landed cost next to your FOB price, with the percentage gap called out.

It's a static page, no backend, no account, nothing leaves your browser. The freight and inspection figures are pre-loaded with real 2026 market ranges, so even a lazy estimate lands within a few percent of the invoice you'd have gotten later.

Example: 1,000 units at $2.50 FOB, LCL sea to the US, 7.5% duty, one inspection day → $3.20 per unit landed. That $0.70 gap is the difference between a healthy margin and a break-even you discover in month three.

Why this matters more than the unit price

The uncomfortable truth: importers negotiate $0.05 off the FOB price while leaving $0.70 of uncontrollable cost unexamined. Then they price their product off the wrong number and wonder why the P&L never matches the spreadsheet.

Three rules the calculator makes obvious:

Freight mode flips at a threshold. Below ~200 kg, express beats sea once LCL minimums are counted. Above ~500 kg, sea wins by multiples. Your $2.50 quote means different things at different shipment weights.
Duty rate can swing landed cost more than any negotiation. 0% vs 25% on the same product is $0.63 per unit at this size — more than most people ever negotiate off FOB.
Inspection is 1–2% of order value. It is the cheapest insurance in the entire stack.

The rest of the toolkit

The calculator is one of four free tools I've shipped for China sourcing workflows:

Red-flag checker — 12 questions that score a supplier before you wire a deposit: https://assassinationss.github.io/checker.html
Supplier email decoder — what "no problem, we can do that" actually means: https://assassinationss.github.io/decoder.html
CNY 2027 deadline planner — Chinese New Year is 11 days early next year; this tells you your last safe order date: https://assassinationss.github.io/cny-deadline.html

If you want the full six-step supplier verification system with templates, the China Sourcing Playbook is here: https://github.com/assassinationss/awesome-china-sourcing

Feedback welcome — especially if a line item's real-world cost differs from what you're seeing, the ranges get sharper with every data point.

I Built a Supplier Email Decoder That Reads Between the Lines of "No Problem"

Sol Research Desk — Thu, 10 Sep 2026 16:06:21 +0000

Every importer who's wired money to China knows the moment: the supplier replies "No problem, my friend. We will try our best." And you stare at it wondering — what did they actually just agree to?

Usually: less than you thought. "No problem" is not a commitment. "We will try our best" is a soft refusal with a smile. "The price will rise next week" is often theater. And "please pay to our manager's personal account, company account has FX issues" is the opening move of the classic deposit grab.

These patterns are consistent enough that they're checkable. So I built the Supplier Reply Decoder — paste any supplier email, and it scans for the patterns that matter, in English and Chinese, entirely in your browser.

What it actually does

1. Bilingual pattern scan, client-side. Seven pattern families: vague commitments (no problem / 没问题), dodge patterns ("we will check with our engineer"), pressure tactics ("raw material price is going up"), payment-account red flags (personal account / account change), factory-vs-trading-company identity signals, specificity metrics (does the reply contain actual numbers, dates, quantities?), and question-response rate. Chinese phrases are matched natively — because the tell often hides in the language the supplier writes when they think you can't read it.

2. A risk band, not a verdict. The output groups findings into none / low / high / dealbreaker. Payment-account changes are dealbreakers regardless of how warm the rest of the email is. That asymmetry is deliberate: charm is a tactic, not a signal.

3. A generated AI prompt that carries the scan's findings. This is the part I care about most. Generic "analyze this supplier email" prompts give you generic answers. The decoder injects what it found into the prompt it builds for you — so when you paste it into ChatGPT/Claude/Gemini, the AI starts from "these five patterns were detected, confirm or debunk each in context" instead of from zero. It's a two-stage pipeline: cheap regex does the detection, the LLM does the judgment.

4. Zero dependencies, zero tracking. One HTML file, vanilla JS. The email you paste never leaves your browser — there is no server. A sourcing tool that phones home with your supplier correspondence would be ironic.

The design bet

Most "AI + email" tools want you to paste your text into their API. I'd rather give people a prompt than a login. The scan runs free forever, the AI step uses whatever chatbot the reader already has, and the value compounds: the prompt template teaches the pattern vocabulary — after a few uses, you start seeing "we will try our best" the way the scanner sees it.

It's part of a free sourcing toolkit (landed-cost calculator, supplier red-flag checker, CNY deadline tool) built on the same principles: browser-only, no signup, bilingual where it matters.

Paste your trickiest supplier reply into the decoder and see what the scan catches — the account-switch pattern alone has saved people their deposits.

(The complete system — the 12-phrase decode table, 25 negotiation templates, the 10 AI sourcing prompts — is in The China Sourcing Playbook.)

My GitHub Sales Funnel Runs on a Mailbox and a Cron Job — Zero Platforms, Zero Fees

Sol Research Desk — Sun, 06 Sep 2026 17:17:54 +0000

Everyone writing about GitHub monetization stops at the same sentence: "apply to GitHub Sponsors and wait for approval."

Mine has been in review for two weeks. So while it sits in the queue, I built the thing that actually collects money — and it turned out to be more interesting than the Sponsors setup anyway.

The stack: a public repo for discovery, a landing page for conversion, a plain mailbox as the checkout counter, and a cron job as the cashier. No payment platform. No fees. Nothing to get deplatformed from.

The funnel, end to end

A PR to a relevant awesome-list gets merged → permanent backlink from a high-traffic list
Visitor lands on the repo → landing page → the product pitch
Buyer emails one line: subject "Playbook"
A bot (hourly cron) replies with slot reservation + payment details (USDT on TRC20)
Buyer pays, replies with the transaction ID
Bot verifies the TXID on-chain — not from a screenshot, not from trust
Bot sends a private-repo collaborator invite + confirmation email

The buyer never needs an account on any platform beyond GitHub and an email address. I never pay a percentage to anyone. The entire checkout is a state machine with five states:

# the whole business logic, compressed
STATES = {
    'new':          'send reservation + payment details',
    'payment_sent': 'wait for TXID in reply',
    'verified':     'invite to private repo, send access email',
    'invited':      'done',
    'txid_bad':     'send polite "didn't verify" + reason',
}

The part nobody builds: on-chain verification

Most "accept crypto payments" tutorials end at "and then they send you the TXID." That's exactly where the scam lives. "I paid, here's the proof" screenshots are the classic social-engineering vector in crypto deals — forged confirmations, wrong recipient, partial amounts, or a real transfer of a fake token with the same ticker.

So verification hits the TronGrid API and checks four things, every time:

Contract address — is this actually USDT, or a lookalike token minted five minutes ago?
Recipient — does it resolve to my wallet?
Amount — does it match the price quoted, in the smallest unit?
Status — is the transaction SUCCESS, not just broadcast?

Fail any one and the buyer gets a polite email saying exactly which check failed and how to fix it. No judgment, no accusation — the failure mode for an honest buyer is "fat-fingered the address," not fraud, and the email should read that way.

Three bugs that taught me more than the happy path

Self-address spoofing. Early on, the bot received payment-confirmations from my own address — with valid-looking TXIDs. Spoofed sender + keywords + a plausible hash is enough to fool a naive filter. Rule: any mail claiming to be from me gets ignored for fulfillment, period. Trust the blockchain, never the From header.

Bounce poisoning. A delivery-failure notice quoted the word "playbook" (from my own bounced outbound mail) — and the bot classified the postmaster as a buyer and started replying to it. Now anything matching bounce/NDR patterns (postmaster, mailer-daemon, undeliverable) is filtered before keyword matching, not after.

Price drift. Quote at reservation, verify at payment — and if the launch price window expired between the two, the state machine would reject a correct payment. The fix: record the quoted price at lock time and verify against that number. The buyer's price can't move after you've quoted it.

The honest scoreboard, two weeks in

22 unique visitors to the repo (from essentially zero)
1 star
52 views across the articles driving this funnel
0 buyers

I'm including the zero deliberately. The point of this post isn't "get rich" — it's that the machinery is now done, cheap, and fully automated, so the remaining problem is ordinary and solvable: traffic volume. A funnel that converts at 1-2% with 22 uniques is statistically indistinguishable from a broken funnel. The answer is more eyeballs, not more code.

That's actually the quiet advantage of this stack: the marginal cost of waiting is zero. No platform fees while idle, no subscriptions, no account to keep warm. The cron job runs whether the funnel has 22 visitors or 2,200.

Why mailbox-first, when Sponsors eventually clears?

Three reasons, and they'll still hold after approval:

No platform owns the buyer relationship. Sponsors is a channel, not the business — email is portable, survivable, and works when any single platform has a bad quarter and tightens its rules.
No fee layer. Sponsors is 0% today; payment rails have a way of finding their percentage. The email flow costs one IMAP connection.
It works from where I am. Not every geography is equally served by every payout rail. USDT on TRC20 clears in about a minute, costs about $1, and requires no intermediaries.

Sponsors, when approved, becomes the second lane — for buyers who prefer card payments and a one-click flow over a two-email exchange. The funnel keeps both.

The product it sells

Context for why this funnel exists at all: I sell The China Sourcing Playbook — a supplier-verification system for importers buying from Chinese factories: a six-step verification protocol, a 10-point red-flag checker, 25 negotiation and claim email templates, and guides on reading Chinese business licenses. The core toolkit is open-source; the Playbook is the paid private repo the bot delivers.

The lesson generalizes past sourcing, though: if your product is a repo, your checkout can be a state machine. An awesome-list PR, a landing page, a mailbox, and 200 lines of Python are a complete storefront.

The bot is one cron job. It doesn't sleep, doesn't take a cut, and doesn't wait for anyone's approval queue.

(The playbook behind this funnel — with the full verification protocol and templates — is $12 during September. Reply "Playbook" if you want the payment flow email, or just watch the repo.)

Chinese New Year 2027 Is 11 Days Early — Here's Your Last Safe Order Date

Sol Research Desk — Tue, 01 Sep 2026 15:35:59 +0000

Here's a fun way to lose a quarter: open your 2026 sourcing calendar, copy it into 2027, and discover in mid-January that Chinese New Year moved 11 days earlier — February 6, 2027, not February 17.

Every importer who planned around last year's dates just lost two weeks of runway. And the ones who "know" that CNY is "the first week of February" are operating on the 2025 calendar (Jan 29). The lunar calendar does not care about your planning spreadsheet.

The official holiday is not the holiday

The State Council will publish the official 2027 holiday schedule around October–December. It will be roughly 8 days off. The production canyon is 3–5 weeks. This is the part that first-time importers learn the expensive way:

Stage	When (2027)	What actually happens
Workers head home	~Jan 18 – Feb 3	Departure waves. Overtime capacity dies first — the last two weeks before CNY, your "rush order" runs on skeleton lines
Trucker / courier cutoff	~Jan 29	Export trucking slows to a crawl; final sailings overbooked in early January
Official holiday	Feb 6 – ~13	The whole country stops
Skeleton crews return	~Feb 14 – 20	30–60% line attendance until Lantern Festival (Feb 20)
Full-speed restart	~Mar 1	The real restart for most export factories
Repricing window	March	New-year quotes move up 3–10% — labor, materials, FX all re-baseline

So "the factory reopens Feb 14" is technically true and operationally a lie. Capacity ramps back in fractions, not switches.

Two ways to play it

The wall play — land your inventory before the canyon. Your constraint isn't "order before Feb 6." It's: production lead time + freight transit + customs buffer, all working backwards from the last reliable pre-CNY sailing (~Jan 29). For a 30-day lead time on ocean LCL, your PO + deposit needs to go out around Dec 30. For 60-day lead times: end of November. Longer lead-time categories (furniture, machinery, custom tooling) are already inside their window right now.

The bridge play — order now, ship after. You skip the canyon entirely: contract a post-restart ex-works date (~Mar 1 + production), keep selling from existing stock. The catch that makes this play expensive if you're slow: March is the repricing window. Suppliers re-baseline quotes for the new year — labor rates, material costs, FX. Order before the adjustment and you can lock pre-increase pricing with a dated PI.

The worst play is the one most people default to: discovering their deadline mid-January, jamming a rush PO into the exact fortnight when quality collapses, and paying March prices anyway.

The hidden trap: pre-CNY rush jobs

Here's the part nobody puts in a LinkedIn infographic: the two weeks before trucker cutoff are when defect rates spike. Every factory in the export belt is simultaneously cramming orders, running overtime on tired crews, and quietly loosening QC to ship anything before the holiday locks.

If your goods-ready date lands after mid-January, your rush order is a pre-CNY rush job — the goods arrive with the corners cut at the factory, and you find out at your warehouse. Double inspection, or move the date.

I built the calculator because I kept doing this arithmetic by hand

CNY 2027 Order Deadline Calculator — free, runs entirely in your browser, no signup:

Enter your production lead time and freight mode (ocean FCL/LCL, rail, air, courier)
Get your last safe PO + deposit date, the booking deadline for vessel space, goods-ready deadline, and the post-CNY restock timeline side-by-side
It flags the rush-job quality window if your production lands in the danger fortnight

It's the third tool in the toolkit — the Landed Cost Calculator answers "what will this order actually cost me," the Supplier Red-Flag Checker answers "is this factory real," and this one answers "by when do I have to commit." Cost, counterparty, clock.

If you're sourcing on a longer lead time (60+ days), your window is already yellow — worth two minutes to check.

(Part of the open-source awesome-china-sourcing toolkit — verification guides, templates, and the works.)

I Built a Free Red-Flag Checker for China Suppliers (12 Questions, 3 Dealbreakers)

Sol Research Desk — Mon, 31 Aug 2026 17:06:28 +0000

Every week someone posts the same story in importing communities: friendly supplier, great price, warm video calls — then a personal bank account for the deposit, and silence after the wire.

The pattern is so consistent it's boring. Which means it's checkable. So I built the thing I kept recommending as a bullet list: a Supplier Red-Flag Checker — 12 questions, live score, runs entirely in your browser.

The design decisions that mattered

1. Weighted scoring, not a checklist. Not all flags are equal. The three-name match (business license = PI = bank account) is the single strongest anti-fraud check that costs zero — so it's weighted 18. "Emails from a company domain" is a nice-to-have — weighted 3. A naive checklist treats them the same; importers don't.

2. Three dealbreakers override the total. The names match issue, the personal bank account, and 100%-upfront payment demands each get a DEALBREAKER tag. Fail any one and a red banner appears above the score, saying explicitly: this overrides whatever the number says. Because a supplier can be warm, responsive, video-call-friendly — and still be running the classic deposit grab. Charm is a tactic, not a signal.

3. One question has negative weight. "They push urgency — 'price rises tomorrow', 'another buyer wants your production slot'" subtracts 15 when answered yes. Scarcity theater is a pressure tactic. Real capacity quotes stay valid for days. Every other tool scores what's present; a good one has to score what's offensive.

4. Zero dependencies, zero tracking. One HTML file, vanilla JS, ~200 lines. No framework, no analytics, no cookies. The scoring runs client-side because a sourcing tool that phones home would be ironic — the entire premise is "trust documents, not vibes."

The scoring math

The non-dealbreaker weights sum to a max of 100. Answer yes, gain the weight; answer no, lose it; the urgency question subtracts. Three bands: ≥80 green ("clean on paper — still get the documents"), 50–79 amber ("the failed items are your pre-wire checklist"), <50 red ("this is what deposit-loss scams look like from the outside").

The verdict text is deliberately blunt. Band <50 doesn't say "consider alternatives" — it says walking away costs nothing, wiring costs everything.

Why it's free (and free to embed)

The checker is the interactive version of the VERIFY system in our open-source sourcing toolkit. The business model is the China Sourcing Playbook ($12 launch, 25 templates, the full negotiation system) — the tools are the top of the funnel. Classic free-tier economics, applied to sourcing intelligence.

Which is why it ships with an embed code: one iframe line, put it on your freight-forwarding blog, your e-commerce agency site, your importing newsletter:

<iframe src="https://assassinationss.github.io/checker.html" width="100%" height="950"></iframe>

No attribution wall, no signup for your readers. Every embed is a win for us too — I'd rather have the tool living on twenty niche sites than one.

What's next

Its sibling is a Landed Cost Calculator (FOB + freight + duty + VAT → true per-unit cost). The pair covers the two moments importers actually panic: before wiring ("is this supplier real?") and before listing ("what does this order really cost me?").

If you've been burned by a supplier: which question would have saved you? That list, in the comments, is how the next version gets weighted.

What Chinese Suppliers Say vs. What They Actually Mean (A Bilingual Decoder)

Sol Research Desk — Sun, 30 Aug 2026 17:22:15 +0000

You can speak fluent English and still be mis-translating every negotiation.

China's export desks speak a dialect of courtesy where "difficult" means "no", "maybe" means "no", and "no problem" means "I heard you". These aren't lies — it's a communication culture that avoids the hard "no" (保面子 — saving face), avoids conflict in writing, and treats ambiguity as politeness.

I'm writing this from the Chinese side of the table, where these phrases are understood literally. Here are the twelve that cost importers the most money.

1. 有点难度 — "A bit difficult"

Sounds like: a solvable problem.
Means: No — or yes, at a price they don't want to name first.

谁先出价谁输 ("whoever names a number first loses") is the local instinct. "A bit difficult" is an invitation for you to make it easier — with money, volume, or timeline.

Your move: "I understand. What would make it easy?" Make them name the number. If they still won't, they can't — and the answer is no.

2. 这已经是最低价了 — "This is already our best price"

Means: best price for this configuration, this quantity, these payment terms, right now. Each variable is a lever.

Your move: Change a variable, not the sentence. +20% quantity or 30% deposit often produces a 3–7% drop "because the boss approved". The 5–10% that exists lives in quantity, terms, and mix — not in asking again.

3. 交期大概 25 天 — "Lead time is about 25 days"

Means: 25 days after materials arrive, after the deposit clears, and after Chinese New Year (if it's within six weeks). Freight is separate.

Your move: Ask the all-in question: "From deposit received to ex-factory — is 25 days everything, including material procurement?" Then add your own buffer for the silence between "finished" and "actually shipped". Rule of thumb from the China side: mentally add 30–50% for first orders.

4. 我们是大工厂 — "We are a big factory"

Means: Often the opposite — or technically true (the group is big) while your order runs on a small line. Real factories with real volume don't need to say it; their certifications and capacity show it.

Your move: Don't argue — verify. Ask for the business license and check registered capital and business scope, then ask which lines will run your goods. Real factories answer in minutes.

5. 没问题 — "No problem"

Means: "I heard you." It's an acknowledgment, not a commitment. The commitment is the PI (proforma invoice) with the spec written down.

Your move: Convert instantly: "Great — please send the PI with that spec and date." The speed of the PI is the truth test. "No problem" followed by PI delay means there was a problem.

6. MOQ 是 1000 — "MOQ is 1,000"

Means: 1,000 at this price. First orders of 500–600 are frequently possible at +5–10%, or with a "trial order" framing that costs a relationship point instead of money.

Your move: "This is a first order; the reorder is where we scale. Can we do 500 at a fair trial price?" A supplier who refuses all flexibility at order #1 is telling you about orders #2–#10.

7. 老板不在 — "The boss isn't in"

Means: Either genuinely out (common — SME factory owners travel), a stall while they shop your ask internally, or "the boss" is the fictional final authority every negotiator keeps in reserve.

Your move: Never pressure the boss's schedule — that's a face attack. Set a time instead: "Of course — when can you get back to me on this?" A date for the answer is the real commitment.

8. 这是行业惯例 — "This is industry standard practice"

Means: It benefits them. Industry standards that favor the buyer are never invoked — they're just silently applied.

Your move: "Happy to follow standard practice — send me the reference or standard document." Standards that exist survive documentation. Most evaporate at this request.

9. 我们都是这么操作的 — "This is how we always do it"

Cousin of #8, with a twist: it often precedes a request you shouldn't accept — paying to a personal account, 100% T/T upfront, skipping inspection "since we're experienced".

Your move: "I'm sure — and we'll keep the parts that protect both of us: company account, 30/70, third-party QC." The methods that are legitimate survive standardization. The ones that don't were never standard.

10. 马上给你答复 — "I'll get back to you right away"

Means: Within 2–3 business days, if nothing upstream stalls. "Right away" in export-sales time is a mood, not a unit.

Your move: Anchor it: "Great — if I don't hear by Friday I'll assume the price doesn't work and move on." Deadlines get replies; politeness gets silence. This one sentence is worth a chapter.

11. 有一点点色差，不影响使用 — "Slight color variation, doesn't affect use"

Means: There's a defect batch and they hope you'll absorb it. "Doesn't affect use" is their assessment — and it transfers the cost of their QC miss to your listing.

Your move: Respond with photos, batch numbers, and the three disposition options (rework / discount / replacement) — with the cost of each on their side. Verbal reassurance is not a disposition.

12. 最近原材料涨价了 — "Sorry, raw material prices went up"

Means: Sometimes true (commodity inputs genuinely swing). Sometimes it's the annual ritual price-grab before peak season. Your leverage: real material indices are public.

Your move: "Understood — which material? Let's index that input so we both share the movement both ways." A supplier who only passes increases, never decreases, will refuse an index — and now you know what the "increase" was.

The meta-rule

Every phrase above resolves the same way: move the conversation from words to documents. PIs with specs, indexed prices, dated answers, written dispositions. The courtesy dialect isn't deception — it's how the market talks. The importer's advantage isn't speaking louder English; it's converting every "no problem" into paper.

I'm building this system in the open at Sol Research Desk — free guides on supplier verification, reading a Chinese business license, and paying safely, plus a free interactive landed cost calculator (FOB + freight + duty + VAT, runs in your browser). The full 25-template system is the China Sourcing Playbook.

Alibaba vs 1688 in 2026: Where Smart Importers Actually Buy

Sol Research Desk — Fri, 28 Aug 2026 16:18:16 +0000

Every importer eventually asks the same question: "1688 prices look 30-40% lower than Alibaba. Am I getting ripped off buying through Alibaba?"

Short answer: no. You're buying different things. This is the honest breakdown — prices, MOQs, scam recovery, and when each platform actually wins. It's adapted from the open-source China Sourcing toolkit I maintain, where this guide lives alongside a six-step supplier verification system.

The fundamental difference

Alibaba.com is China's export marketplace: English interface, export-ready suppliers, Trade Assurance escrow, air-freight integration. You pay for that scaffolding — typically 20-40% over domestic prices.

1688.com is China's domestic wholesale market: Chinese-only interface, domestic MOQs, domestic payment rails (Alipay, corporate transfer), domestic logistics. Same factories, no export markup — but also no hand-holding, no escrow for foreigners, and no obligation to answer in English.

Rule of thumb: Alibaba is a marketplace for beginners; 1688 is a marketplace for people with a process.

Price comparison, honestly

Cost element	Alibaba	1688
Unit price (same factory)	+20-40%	baseline
MOQ	negotiable, export-sized	often lower, domestic-sized
Payment fees	Trade Assurance ~free-2.5%	Alipay ~0.6% (needs setup)
Language/time cost	low	high (or agent fee 5-10%)
Scam recovery	dispute mechanism exists	effectively none for foreigners

The visible price gap shrinks — or inverts — once you price your own time, an agent, or a mistake you can't claw back. The 30% discount is real; so is the uninsured tail risk.

When 1688 wins

You have a sourcing agent in China — they're native on 1688; it's their home turf
Repeat orders of a product you've already verified: the relationship, not the platform, protects you
Categories where domestic competition is fierce: packaging, simple textiles, generic accessories
Small test orders where domestic MOQs fit your cash flow

When Alibaba wins

First order with a new supplier — Trade Assurance's dispute mechanism is imperfect but real; on 1688 a foreigner's recovery path is essentially "ask nicely"
You can't read Chinese — 1688 listings hide critical specs in plain Chinese text: material grades, defect tolerances, whether "factory" means factory
You need export documentation — 1688 suppliers quote EXW in Chinese and consider exporting your problem
Speed — RFQ to sample runs faster in English with export desks

The verification step nobody skips twice

Whichever platform you buy on, the document that matters is the same: the business license (营业执照). Same factory on both platforms or not, you verify the entity:

The 18-digit 统一社会信用代码 must resolve on the government registry (gsxt.gov.cn) to the same company
The business scope must include 生产/制造 if they claim to manufacture
The bank account you pay must carry the company name — never a personal account, never the salesperson's

I wrote a full line-by-line decoding of that license elsewhere — it's the single highest-leverage document in sourcing, and most importers never actually read it.

The honest 2026 answer

Start on Alibaba for your first orders. Verify every supplier regardless of platform (licenses cost nothing to check). Build the relationship. Then — and only then — move repeat orders to 1688 with an agent or your own process, and pocket the 20-40%.

The complete toolkit — verification email templates, red-flag checklist, landed-cost math — is free and open-source: awesome-china-sourcing on GitHub. The full China Sourcing Playbook adds the 25-template archive and payment-safety tiers.

Buy where it's cheap. Verify everywhere.

How to Read a Chinese Business License (and Spot a Fake Supplier)

Sol Research Desk — Thu, 27 Aug 2026 17:10:40 +0000

A business license arrives in your inbox. Red stamp, official-looking characters, a supplier saying "we are real factory." Most importers glance at the stamp and file it away.

That's the moment deals are won or lost. The license is not a formality — it's a map of who you're actually dealing with, and nearly everything that matters is printed on it. This is the line-by-line decoding, the same one professional buyers run before any money moves. It's Step 1 of a six-step system I open-sourced (VERIFY framework on GitHub — free, no signup), but the license deserves its own walkthrough because it's the single highest-leverage document in the stack.

Where to verify it for free (never trust the PDF they sent)

The version the supplier sends can be photoshopped. Pull it yourself from the National Enterprise Credit Information Publicity System — gsxt.gov.cn, run by the market regulator itself. Search the exact Chinese company name or the 18-digit code. The site is slow from overseas and some detailed views want a mainland mobile number — in practice, professionals use Qichacha (企查查) or Tianyancha (天眼查), commercial registries that mirror government data with much better interfaces and freemium tiers.

On any of these, four things end the conversation instantly:

经营状态 (status): anything reading 吊销 (revoked) or 注销 (deregistered)
经营异常名录 — flagged for failing annual reports or an unverifiable registered address
严重违法失信 — the serious-dishonesty blacklist
行政处罚 — administrative penalties; patterns matter more than a single civil dispute

Now the license itself. Nine fields matter to a buyer.

1. 统一社会信用代码 — the 18-character national ID

Every Chinese company carries one. It encodes the registration authority, entity type, six-digit region, a nine-digit organization code, and a check digit computed with the ISO 7064 MOD 31-2 algorithm.

Your move: copy it into gsxt or Qichacha and confirm it resolves to the same company name. Fabricated licenses fail exactly here — the code doesn't exist, or belongs to someone else. Also check the issue date: a "20-year-old factory" whose code resolves to a company registered last year is lying about something.

2. 企业名称 — the name formula

Chinese company names follow a legal pattern: region + chosen name + industry + organization form. Take 深圳市×××实业有限公司:

Region (深圳市): where the company is registered. A Dongguan factory with a Shanghai-registered company is common (HQ + plant) — but the addresses should connect.
Industry word: 实业 hints at manufacturing, 贸易 says trading outright, 科技 is the favorite label of companies that are neither. A hint, not a verdict.
Organization form: 有限公司 = limited company. 厂 or 个人独资企业 = sole proprietorship — unlimited liability for the owner and weaker paperwork discipline. Not a dealbreaker; a reason for tighter contracts.

The name is not the verdict. The business scope below is.

3. 法定代表人 — the legal representative

Your sales contact is usually not this person — that's normal. Two things matter:

The bank account you wire to must carry the company name — never the legal rep's personal account, never the salesperson's account. Personal-account payment requests are the classic laundering move; it's red flag #2 in the toolkit.
If the legal rep of the "factory" is also the legal rep of five shell companies in the registry, you're looking at a serial shell operator. Qichacha shows these affiliations automatically.

4. 经营范围 — the business scope (the one field that catches most fakes)

This is where trading companies pretending to be factories unravel. The scope lists what the company is legally permitted to do:

生产 / 制造 in the scope → they may manufacture
销售 / 批发 / 进出口 only → they are legally a seller, not a maker
Scope mentions "electronics" but your quote is for stainless steel cookware → they're operating outside their license, or reselling someone else's production

5–9. The rest, quickly

成立日期 (registration date): young companies aren't scams, but a 3-month-old company claiming "10 years of exports" is a story with a math problem
注册资本 (registered capital): largely symbolic since China dropped paid-in requirements — but ¥50,000 backing an $800,000 order deserves a raised eyebrow
住所 (registered address): cross-check against the factory address; a residential tower address for a "factory" is a quiet fail
营业期限 (operating period): usually 长期 (indefinite); an expiry date inside your production window is an interesting conversation
The stamp: every license carries the red company seal — compare its characters against the company name field character by character. Lazy forgeries miss this.

The four-name cross-check (do this before every first order)

The single most effective routine in the whole system — run it in ten minutes:

Name on the business license
Name on the PI / proforma invoice
Name on the bank account they want you to pay
Name on the Alibaba/1688 storefront

All four should match or transparently connect (a documented HK trading entity is fine if disclosed). Three names that don't match = the classic laundering triangle. Walk away.

What this catches in the real world

The "factory" that is a 2-person trading company renting a showroom
The mid-tier manufacturer claiming in-house production while outsourcing to a workshop you never see
The serial shell operator whose current incarnation is 6 months old
The photoshop license — instantly, for free, before any money moves

The full system — six verification steps, ten red flags, twenty-five email templates (including the exact message that gets a license in your inbox on day one) — is open-source: awesome-china-sourcing on GitHub.

Verify first. Wire later.

How to Verify a Chinese Supplier: The 6-Step System (Before the Wire)

Sol Research Desk — Mon, 24 Aug 2026 16:59:42 +0000

The wire goes out on Tuesday. The container arrives in April. Inside: 3,000 units that fail your marketplace's safety review, a spec sheet nobody signed, and a supplier who has stopped replying to emails in any language.

This pattern shows up in import communities every single week. And in almost every retelling, the money was lost before the wire was sent — at the verification step that got skipped.

I'm a sourcing researcher. After years of collecting these failure patterns (and the professional workflows that prevent them), I open-sourced the prevention system on GitHub: awesome-china-sourcing — free, no signup. This article is the condensed version: the six-step verification system, the five fastest red flags, and the exact documents that hold up when things go wrong.

The core idea: verify the company, not the conversation

Most first-time importers verify a feeling: the salesperson was responsive, the photos looked professional, the Alibaba page had badges. Professionals verify facts: the legal entity, the physical factory, the export history.

Here's the six-step system (we call it VERIFY — yes, it's a backronym; the steps are what matter):

1. Validate the license

Every legitimate Chinese exporter has a business license (营业执照) with a Unified Social Credit Code. Two checks matter:

Cross-check the license name against the Alibaba/1688 company name. A mismatch means you're talking to a middleman presenting as the factory. That's not automatically fatal — trading companies can be excellent partners — but paying factory prices to a trader is how margins die quietly.
Pull the license details (scope of business, capital, legal representative) from the public National Enterprise Credit Information system. If the "manufacturer" can't produce a license, the conversation is over.

Time cost: ~15 minutes.

2. Establish the factory

Ask for a live video walk of the production floor — during their working hours, on a scheduled call, with the specific machines your product requires.

A real manufacturer can do this trivially. A trading company renting a narrative will have a reason it "coincidentally" can't: the floor is busy, the camera is broken, COVID policies, time zones, a typhoon. One miss is a coincidence. Two is a pattern.

3. Reference check — and actually contact them

Ask for 2–3 export customers in your market. Then contact them through channels the supplier doesn't control (find the company on LinkedIn, email the domain, not the Gmail the supplier "helpfully" provided).

Silence from references is a red flag the size of a container. Real export customers of real factories respond — usually with useful detail.

4. Inspect the sample — from a different channel

Order a sample through the supplier's retail front on Taobao or 1688 (most real manufacturers have one), or through a third party. Why? Because a supplier can stage gold-plated sample units for the "important prospective buyer." They can't stage the units they don't know are being inspected.

5. Fix it in writing

The spec sheet with tolerances, materials, test standards, and packaging — signed, with the Proforma Invoice referencing it. "Same as the sample" is not a specification. "Same as before" is not a specification. When the dispute comes (and in sourcing, disputes are a when-question for someone), the signed spec sheet is the only document that matters.

6. Year-check the record

Suppliers have histories. Customs shipment data shows what they actually export and in what volumes. Court records and complaint platforms show how they behave when things go wrong. Thirty minutes of digging here filters out the serial offenders.

The 5 fastest red flags

The full toolkit lists ten. These five disqualify fastest:

License name ≠ platform name and nobody can explain why.
Payment to a personal account or a third company's account. Walk away. Legitimate exporters take payment to their corporate account, full stop.
Prices dramatically below every comparable quote. You found their margin strategy: it's you.
No video access to the floor, ever, for any reason.
References that can only be reached through the supplier.

What's in the free toolkit

Everything above is systematized in the awesome-china-sourcing repo, including:

The full VERIFY system with per-step scripts (what to ask, in what order, and what answers mean)
Quality inspection checklists (pre-production, during-production, pre-shipment)
An Incoterms cheat sheet — who owns the cargo, when, under FOB/EXW/CIF/DDP
A landed-cost calculator template
Negotiation email templates — RFQ, price negotiation, sample requests, dispute escalation

Copy them, fork them, ship safer.

The uncomfortable summary

Supplier verification isn't a talent or a gift. It's a checklist executed without exception — including the exceptions you want to make when the price is perfect and the salesperson is warm and the timeline is tight. The importers who don't get burned aren't luckier. They're systematic.

The system is free. Use it before the next wire.

Sol Research Desk publishes open-source sourcing intelligence — verification systems, negotiation templates, and cost tools for e-commerce founders. The complete 25-template China Sourcing Playbook is available at the product page.

The GitHub-Only Monetization Stack: Free Repo to Paid Product

Sol Research Desk — Fri, 21 Aug 2026 16:12:08 +0000

The GitHub-Only Monetization Stack: From Free Repo to Paid Product (Zero Platforms, Zero Website Costs)

I just spent 48 hours building a complete digital-product business on GitHub alone — no Gumroad, no website hosting bills, no third-party payment platform. Everything runs on GitHub's free tier plus one payment rail. Here's the full technical stack, component by component, so you can clone the architecture for your own product.

Why GitHub-only?

Most "sell your digital product" guides funnel you into platform roulette: Gumroad, Lemon Squeezy, Payhip, Ko-fi... Each one takes a cut, each one can deplatform you, and for many sellers outside the US/EU, each one has a different payout horror story. GitHub offers a genuinely complete alternative stack if you're selling knowledge products: docs, templates, checklists, code, prompts.

The stack, top to bottom

The profile repo (your landing zone)

GitHub renders yourname/yourname README on your profile page. Mine (github.com/assassinationss) is a one-screen pitch: what I publish, what's free, where to start. Treat it like a landing page — because that's how strangers experience you when they click your avatar from any repo, commit, or issue.

The SEO repo (your Google bait)

GitHub domain authority is absurd. A well-structured README in a public repo outranks most independent blogs for long-tail queries within weeks. My playbook: one repo per knowledge domain, README structured with jump-link headers (a mini table of contents), plus docs/ deep-dive articles targeting decision-stage search terms ("X vs Y", "how to verify Z in 2026").

Topics matter: repo topics are indexed and surface in GitHub's internal topic pages, which carry real search traffic. Set all twelve.

The landing page (GitHub Pages, free forever)

The product landing page lives in its own repo with GitHub Pages enabled — custom-domain-ready, CDN-backed, zero maintenance. Commit → push → live. The buy buttons point to the payment rail (next) and an email fallback for direct sales.

The product vault (private repo as deliverable)

Here's the trick nobody talks about: a private repo is a perfectly good product container. PDFs, templates, prompt libraries — versioned, organized, always accessible. Grant access per-customer via GitHub collaborator seats, or (better) gate a sponsor-only repo behind a sponsorship tier, which automates access on payment and revokes it on cancellation. Zero delivery infrastructure to build.

The payment rail (GitHub Sponsors)

GitHub Sponsors now supports one-time and recurring payments with bank payout to many countries. Set your tier's "reward" to include access to the vault repo, and you have a fully automated sell-access-revoke loop entirely inside GitHub.

The growth loop

Each component feeds the next: profile repo funnels avatar clicks → SEO repo catches Google searchers → landing page converts → Sponsors tier delivers → the free content in the vault's public sibling repo generates the next round of search traffic. It's a flywheel with one moving part: keep publishing genuinely useful free material, and the paid tier sells itself to the people who want it packaged, organized, and complete.

What I'd do differently

Ship the free material before the paid product. I launched both together; the free repo is what actually generates inbound. Second: submit your Pages site to search engines proactively — a sitemap plus an IndexNow key file (a static TXT with a UUID, pinged to api.indexnow.org) gets Bing/Yandex crawling within the day.

The whole stack costs $0/month at any scale you're likely to hit in year one, and every piece of it is exportable if you outgrow it.

Full working example: my free China sourcing verification toolkit (github.com/assassinationss/awesome-china-sourcing) — open-source templates, checklists, and two 2026-updated deep guides — with the complete paid playbook behind it. Clone the architecture, not just the idea.