Two disclosures first, because they determine how you should read the rest. This site takes a commission when someone opens an account through any of its eight agent links, CnFans and KakoBuy included. And roughly 84% of the outbound agent clicks it sends go to KakoBuy, which makes KakoBuy by a wide margin the agent that pays us most.
A comparison from us that concluded "KakoBuy" would be worth precisely nothing. That is a large part of why this one concludes nothing at all.
Start with what the evidence contains: neither of them
This site is built on 237 Reddit comments quoted verbatim across 124 sellers. Searched for agent names, the result is blunt.
CnFans appears zero times. KakoBuy appears zero times. Not once, in either direction, positive or negative.
The agent names that do appear are incidental — a buyer mentioning which forwarder handled a haul. Superbuy is named in seven comments, three of them an automated sticky posting beginner-guide links rather than anybody's experience. MuleBuy appears three times, all from that same sticky. AllChinaBuy twice, EastMallBuy once, HubBuyCN once, PandaBuy once.
That is the entire agent record here, and what it means is that this corpus is about sellers. People write haul posts about a jacket, and the agent who forwarded the box is a footnote. Anyone — us included — building an agent ranking out of Reddit sentiment is building it out of something that is not there.
Two different ideas of what a buy link is
The one genuine, verifiable mechanical difference between these two is how each accepts a product.
- KakoBuy takes the entire source URL, encoded, as a parameter on an item-details path. It does not care what shape the link is; it resolves it on its side.
- CnFans prefers the item pulled apart first: a platform token, drawn from a three-word vocabulary covering Weidian, Taobao and the 1688 wholesale side, plus the numeric item id. It accepts a whole encoded URL as a fallback when an id cannot be parsed.
Practically, the whole-URL style survives being pasted into more places, because there is no id to extract wrongly. The id-and-platform style is more brittle at the moment of construction and more legible afterwards — you can read a CnFans link and see which marketplace and which item it points at.
Both are recoverable either way, which is why the decoder can rebuild working links for all eight agents from a CnFans or KakoBuy URL that stopped opening years ago. Save the original Weidian or Taobao link rather than the agent one, though: the decoder rebuilds the second from the first forever, and not always the other way round.
The referral plumbing differs, and it matters once
KakoBuy's signup link travels through a separate short host before landing on the site. CnFans' is a registration page on its own domain carrying a referral number.
This is plumbing, not service, and we raise it for exactly one reason: an extra hop is an extra thing that can fail, and a failed hop reads to a buyer as "the agent is down". It usually is not.
The one incident on file, and why it was not an outage
We monitor two URLs per agent — the referral redirect and a real buy link built from a live Weidian item — rather than homepages, and that design came from a specific false alarm.
KakoBuy's bare host once answered 451, "unavailable for legal reasons", to an automated request from a European edge, while both URLs this site actually emits returned 200 at the same moment. A checker pinging homepages would have declared an outage that did not exist. So the classifier answers in three states: alive on a 2xx, dead on a 5xx or a broken redirect chain, and inconclusive for a 403 or a 451, which mean a server refused to talk to a robot and say nothing about whether a person can buy through it.
The practical version for you: if either agent's homepage will not open, test an actual product link before concluding anything. CnFans has produced no equivalent incident in our checks, which is a statement about our sample and not a clean bill of health.
The cost stack is the same shape at both
Neither charges one number, and comparing headline service fees alone will mislead you. Both build an order from item price plus domestic Chinese shipping, a service or purchase fee, storage that is free for an opening window and then charged, international freight quoted only once the parcel is packed, and add-ons for extra photos, box removal, repacking and insurance.
We publish no percentages for either. We have not audited them, they are the agents' to change, and freight — normally the largest line — moves for reasons unrelated to which agent you picked, as why the same line quotes differently sets out. A lower service fee attached to a worse freight rate is a worse deal on almost any real haul. The only comparison worth anything is a matched basket: same items, same weight, same destination, same week, all-in. We have not run it.
What would be needed to actually rank them
- A matched basket priced at both in the same week, freight included.
- Order-to-warehouse timings across enough orders to survive one bad month.
- QC photo sets of the same item from both, judged side by side.
- Dated dispute outcomes — granted, refused, ignored — with sources.
- Payment success rates by country, where agents quietly differ most.
None of that exists here, so neither is ranked, exactly as a seller with no votes stays unranked. The method page explains why not knowing is not a grade.
Deciding in ten minutes, without us
- Open both and check payment. Which methods each accepts from your country is the difference most likely to actually bind you, and the easiest thing to verify.
- Test your real item, not a hypothetical. Paste the product you intend to buy into the decoder and open both results. If one resolves and the other does not, the decision has been made for you.
- Read both fee pages the same afternoon. That is the only fee comparison that is current, and it takes ten minutes.
- Consider your existing links. If your saved finds are already in one agent's format, converting them is friction you can simply not have.
- Start small, on something from the shelf or a seller with quoted evidence on the seller pages, and ship a real parcel before forming an opinion. Then read when to consolidate and when to ship before the second one.
Longer walkthroughs live at the CnFans review and the KakoBuy review. The other pairings people ask about are KakoBuy vs Superbuy and CnFans vs MuleBuy, and the reasoning behind refusing to name a best agent is its own guide.