Shenzhen factory visits and escort interpreting: what 1,002 Shenzhen Canton Fair exhibitors do and do not tell you

Checked on 19 September 2026. We are an interpreting company in Guangzhou — not an audit or inspection body, not a law firm, and not the Canton Fair organiser. This is not legal advice and it is not compliance advice. Nothing below decides anything for anybody: the plant decides whether you get in, an audit body decides whether an audit passes, and the organiser decides who enters the Pazhou halls. Wherever a figure comes from our own data rather than from an official document, we say so in the passage that uses it.

Shenzhen is the largest single city in our Canton Fair exhibitor directory: 1,002 of 26,270 companies, ahead of Taizhou (884), Hangzhou (849), Shanghai (767) and Guangzhou (717), and well ahead of Foshan (594). It is also the city whose registered identity tells you the least.

In Foshan, two exhibitors in three are registered as manufacturers, and that one fact carries most of a visit plan. Shenzhen splits three ways almost evenly — 343 manufacturers, 342 manufacturer-and-trading companies, 300 trading companies — and the manufacturer-and-trading share, 34.1%, is the highest of the 39 cities in the file with at least 200 exhibitors. Underneath the city average the districts are not a matter of degree but of kind: in Luohu, 70 of 90 records are trading companies; in Guangming, 65 of 69 carry a production side and 2 are trading companies.

One more thing belongs on the first screen rather than the last: none of the 812 interpreter profiles in our pool is registered in Shenzhen. All 812 are registered in Guangzhou. A Shenzhen plant day is therefore an inter-city day, and the travel time inside it is a cost somebody has to carry.

1. Which row is you

2. What this data is, and what it is not

Every number in §3 to §6 comes from one file we maintain ourselves: a Canton Fair exhibitor directory of 26,270 companies, built from the 139th session exhibitor list (full version, dated 16 April 2026). 1,002 of those are registered in Shenzhen — 3.81% of the file, and 20.8% of the 4,811 Guangdong records in it.

Six limits, each of which really does change how you should read what follows.

This is the 139th session, and the 140th is a bigger event. At the State Council Information Office press conference on 18 September 2026, the Ministry of Commerce put the 140th session at 32,000 exhibiting companies. Our file has 26,270 rows and describes who exhibited in the spring. Some companies appear in both; we cannot tell you how many, because we do not have the 140th session's exhibitor list and so cannot compute it.

We hold street addresses, but they are registered addresses, not plant addresses, and we do not publish them. Being precise about this: when the directory was built we kept a street address for very nearly every record (1,001 of the 1,002 Shenzhen rows have one). It was collected together with contact names, mobile numbers and e-mail addresses, it belongs to the personal-information category, our public directory has never shown that column, and this article will not either. More to the point, publishing it would not help you: it is the registered business address, not the address of a production line. A company registered in Futian can perfectly well run its line in Dongguan, in Jiangxi, or nowhere at all. Those two facts together are why §7 refuses to give you a travel time.

The district column is not in the source data. We cut it out of the address. The source table has no district field. The build script takes that registered address, strips the province and city prefix, and takes the first 区 / 县 / 镇 segment it finds. For Shenzhen it produced a district for 949 of the 1,002 records and failed on 53. Those 53 are not companies without a district: we went back and checked, and 15 of them have a district named in the address in plain characters — Longgang (5), Bao'an (4), Nanshan (3), Guangming New Area (1), Longhua (1), Futian (1) — and the parser simply did not reach it. So the "district not parsed" row in the tables below is our parse residue, not a class of company. Please read it that way.

We merged three pairs of spellings, and that merge is ours. The addresses carry two forms of the same three place names, one with "new area" and one without. Unmerged, the counts are Guangming District 47 plus Guangming New Area 22, Longhua District 100 plus Longhua New Area 20, and Pingshan District 26 plus Pingshan New Area 7. We add each pair together in the tables below. The only basis for that is that the two spellings point at the same place name in the address string; we did not consult any official document on administrative-division history. If you want to cite these numbers, cite the unmerged pairs. One further record parsed as "Pingdi Town"; we left it out of every district row rather than assign it.

Only 375 of the 1,002 records list a product (37.4%, the lowest of the three Guangdong cities we have looked at: Guangzhou 47.4%, Foshan 40.7%). Almost all of them do carry free-text keywords — 1,000 of 1,002 — and §5 and §6 use both fields. Where §6 says a district "leans" to a phase, the denominator is the records our keyword map could place, not the district's full count. A blank field is not evidence of anything.

The "attended the 139th" flag is derived from an exhibiting-history column, not from an attendance roll. 703 of the 1,002 Shenzhen records (70.2%) carry it. The 299 without it are not blanks either: 156 of them have a history that stops at the 138th session, 78 at the 136th, 56 at the 137th, and 9 have no history at all. The source table is itself called the 139th session exhibitor list, so that 70.2% reads much more like "what share of history columns had been updated to the latest session" than like an attendance rate. Do not use it as one.

3. The biggest city on the list, and a registration category split three ways

Company type is a field the directory carries; almost every record fills it in. Shenzhen against the cities it usually gets compared with, and against the whole file:

Rank Shenzhen among every city in the file with at least 200 records — 39 cities, the full set, not a sample, and the same cutoff we used in the Foshan piece so the two articles can be read side by side. On manufacturer share Shenzhen comes 26th of 39, below Guangzhou at 22nd and far below Foshan at 5th. On manufacturer-and-trading share it comes 1st of 39, at 34.1% against a 16.0% average across the file. Lower the cutoff to 100 and you get 67 cities; Shenzhen is then 51st on manufacturer share and 2nd on manufacturer-and-trading share, behind Yangjiang at 40.1%. Changing the cutoff does change one neighbour: at 100 records Yangjiang takes first place on manufacturer-and-trading share and Shenzhen drops to second. On both cutoffs, though, Shenzhen sits in the bottom half on manufacturer share and at the very top on manufacturer-and-trading share. There is a blunter way to put it: across those same 39 cities, Shenzhen's largest single category (manufacturers) accounts for 34.2% of its records, the lowest figure of all 39 — no city on that list has a flatter registration split.

What that gives you: in Foshan, base rates work for you — pick an exhibitor at random, ask to see a line, and roughly two times in three the company has told the organiser it is a manufacturer. In Shenzhen the same question is closer to a coin toss, and the largest single answer is neither yes nor no. It is "both".

What it does not give you: "manufacturer and trading company" is a self-declared registration category collected by the organiser. It is not an inspection and not a certificate, and it says nothing at all about where the line is. A Shenzhen company can be a manufacturer in every honest sense and have no production within a hundred kilometres of the address you were given. §7 is about that gap.

There is a second thing the three-way split does to you, and it is the practical one. In Foshan you can plan first and verify second, because the base rate is on your side. In Shenzhen the verification has to come first, because a third of the list is the ambiguous category and a third is explicitly a trading company. The sequence changes, not just the odds.

4. The spread between districts is far wider than the city average

Split the 1,002 by district. The spread is much wider than the city average suggests:

Those nine rows account for 992 of the 1,002. The remaining ten are Yantian 7, Dapeng New Area 2 and Pingdi Town 1 — too few to say anything about, which is why they are not in the table. Rows do not always add to the record count because the city also has 10 companies filed under "other", 1 research institute and 6 with the field blank.

Read the last column rather than the manufacturer column, because in Shenzhen the manufacturer-and-trading category is where a third of the city sits. Once you do, the city separates cleanly into two groups and one that is genuinely in between:

Luohu and Futian are the trading end. Only 4 of Luohu's 90 records are registered as manufacturers, and only 8 of Futian's 74. Those are not marginal figures; they are the lowest two in the table by a wide margin. We checked all nine rows, not a sample. If your supplier's address is in either district, the sensible default is that the address is an office and the line, if there is one, is somewhere else — and the thing to establish before you buy a ticket is which legal entity owns that line and where it is.

Guangming, Pingshan and Bao'an are the production end. Guangming is the sharpest: 65 of its 69 records carry a production side and exactly 2 are registered as trading companies. Pingshan is 31 of 33. Bao'an is the biggest of the three in absolute terms, 198 of 232.

Longgang, Longhua and Nanshan are in between, and Longgang matters most because it is the single largest district in the data with 258 records. Its split is 95 manufacturers, 96 manufacturer-and-trading, 63 trading — the three-way city pattern in miniature. Nanshan is the one to watch for a different reason: at 50.8% it is almost exactly balanced, so the district name predicts nothing.

None of this tells you that a particular company does or does not have a plant. It tells you what a reasonable prior is before you ask, and which question to ask first. The Foshan equivalent of this section found one reversed district inside an otherwise manufacturing city; Shenzhen has two districts that are reversed, one that is balanced, and two — Guangming at 94.2% and Pingshan at 93.9% — with a higher production-side share than any district in Foshan, whose highest is Shunde at 87.0%. Bao'an, at 85.3%, does not clear that bar. That is one city on paper and several in practice.

You do not have to take the ownership answer on trust. Ask for the business licence of the entity that owns the line, and check the unified social credit code on it against the code on your contract and your invoices; that code can be looked up by anyone on the National Enterprise Credit Information Publicity System at gsxt.gov.cn, and the lookup is free. One caveat: that site blocks automated access hard — three script requests to it on 19 September 2026 all came back HTTP 521 — so open it in a browser, and we cannot confirm for you whether it opens from your network. A supplier is entitled to decline. A refusal is not evidence of anything — it just means you cannot verify the ownership chain before you travel, and the value of the trip should be re-estimated on that basis.

5. The product column has no centre of gravity

The other thing people infer from "Shenzhen" is a product. The data does not support a single answer, and the way it fails is itself informative.

Across the 1,002 Shenzhen records there are 3,676 keyword mentions covering 3,048 distinct keywords. The ten most common keywords together account for just 3.0% of all mentions. The single most frequent keyword in the whole city appears 19 times out of 1,002 companies. For comparison, Foshan's 594 records produce 1,868 distinct keywords, its top ten cover 4.3% of mentions, and its most frequent keyword is a tie at 12 between furniture and the word "factory", which is not a product at all. Foshan is more concentrated than Shenzhen on a much smaller base.

There is a second surprise in the raw counts, and we would rather show it than hide it: the most frequent keyword in Shenzhen is tableware, not anything electronic. Do not over-read that in either direction. Only 37.4% of Shenzhen records list a product at all, the keyword field is free text a company wrote about itself, and a nineteen-out-of-a-thousand lead is not a lead. What can be said with a denominator is this: of the 520 Shenzhen records our keyword map could place into an official zone, 226 touch one of the five electronics-family zones — 电子消费品及信息产品, 电子电气产品, 家用电器, 照明产品, 新能源 — which is 22.6% of all Shenzhen records and 43.5% of the ones we could place. The largest block inside that union is household appliances at 93 records; the consumer-electronics zone itself is 76 (7.6%). A further 97 records (9.7% of all records) touch the three ceramics-and-kitchenware zones, 日用陶瓷, 餐厨用具 and 工艺陶瓷. Electronics is the largest single family, and it is nowhere near a majority.

That matters for the day you are planning because the inference people actually make is not "Shenzhen is electronics" on its own. It is "Shenzhen is electronics, so my supplier is in Phase 1, so I will come in the third week of October". §6 takes that chain apart.

6. Which phase your supplier is in, and which days the fair does not hold it

Start with what is official and fixed. The booth announcement for the 140th session sets the dates (the lead line and the three phase lines, quoted as one block):

第140届广交会计划于2026年10月15日开幕，分三期举办线下展，其中： 第一期：2026年10月15—19日 第二期：2026年10月23—27日 第三期：2026年10月31日—11月4日

The same announcement says the session has 54 zones, and its Annex 1 assigns every zone to a phase: 19 zones in Phase 1, 14 in Phase 2, 21 in Phase 3. That is the whole mechanism. Your supplier's phase follows from its zone, and its zone was settled when it applied for a booth — which is why the reliable way to learn it is to ask the supplier which phase it has been placed in and what its booth number is, not to deduce it from anything in this article. If the supplier will not say, the only route that does not depend on it is the organiser's own: customer service on 4000-888-999 inside China, (0086) 20-28888999 from abroad, e-mail info@cantonfair.org.cn — the channel published in the 139th session's accompanying-personnel notice. Whether it will tell you which phase a given exhibitor is in is the organiser's decision, not ours to promise.

Between the phases there are two gaps, and the trade services zone announcement of 4 September 2026 names them in a single sentence about its own opening hours:

服务时间为每天9:30—18:00，其中10月20—22日、10月28—30日换展期间闭展。

So the changeover gaps are 20–22 October and 28–30 October. Do not read "gap" as "days this supplier is free", though. A changeover is one phase tearing down while the next moves in, so the window on either side of a phase is exactly when that phase's exhibitors are in the hall shifting crates. Counting it the other way round is more useful: on the published calendar, what pins your supplier to Pazhou is its own phase plus the changeover window immediately beside it.

• A Phase 1 supplier exhibits 15–19 October and tears down 20–22 October; from 23 October the fair's calendar no longer holds it.
• A Phase 2 supplier moves in 20–22 October, exhibits 23–27 October and tears down 28–30 October; 15–19 October, and 31 October onwards, are not the fair's.
• A Phase 3 supplier moves in 28–30 October and exhibits 31 October to 4 November; the thirteen days from 15 to 27 October are entirely free of the fair for it, and so is everything after it closes.

That is our reading of the published phase dates plus the one sentence about the halls closing during changeover; it is not the organiser telling you about any one exhibitor, and which day actually works still has to come from the supplier. And it covers only one thing: "the fair is not using them" is not "the plant is running". Whether it works weekends, and how many days it takes for the National Day holiday in early October, are questions for the plant.

Now our own numbers, with the method and its flaw both stated. We keep a keyword list that maps product and keyword strings to the official zone names, and Annex 1 maps zones to phases. The zone names and the phases are the organiser's. The mapping from a company's words to a zone is ours. The list was written zone by zone off Annex 1, so every one of the 54 zones carries at least one key. It is not evenly deep, though: 237 keys across the 19 Phase 1 zones (12.5 per zone), 140 across the 14 Phase 2 zones (10.0) and 153 across the 21 Phase 3 zones (7.3) — and the two deepest zones in the whole list, 电子消费品及信息产品 with 43 keys and 家用电器 with 40, are both in Phase 1. We are telling you that because it turns out to change the answer.

First, what the list says. The narrow version (530 unambiguous product nouns) places 520 of the 1,002 records: 272 touch a Phase 1 zone, 183 a Phase 2 zone, 157 a Phase 3 zone, and 84 of the 520 touch more than one phase, which is why those three counts add to more than 520. Adding 67 generic category words a trader might use makes a broader list of 597 keys, which places 638: 368, 222 and 182. Phase 1 both times — but the broader list contains the narrow one entirely, so the two lean the same way by construction and their agreement was settled before the count was run. That is not a test.

The real test is to cap every zone at the same number of keys, so that all 54 zones carry equal weight. Capped at 2 keys per zone, Shenzhen reads 56 / 72 / 44 — Phase 2. At 4 keys, 89 / 101 / 87, still Phase 2. At 6, 109 / 117 / 107, still Phase 2. It only turns over at 7, and then by a single record (135 / 134 / 119); it settles on Phase 1 at 10, at 12 and uncapped. Drawing 6 keys per zone at random instead, over twelve seeds, gives Phase 1 nine times and Phase 2 three times. So "Shenzhen leans to Phase 1" is less a fact about Shenzhen than a fact about how many words we wrote for the appliance and electronics zones. We withdraw it: we are not going to tell you which phase Shenzhen leans to, not even city-wide.

The control group shows the method itself is not broken. Under every one of those caps, from 2 keys per zone upward, Guangzhou reads Phase 3. Foshan moves the way Shenzhen does — Phase 3 at 2 keys, Phase 2 through the middle, Phase 1 uncapped. So stable cities exist and Shenzhen is not one of them, which is §5's 3,048 distinct keywords said a different way. (The Foshan piece used an earlier version of this same list; read its phase conclusions in the light of this paragraph.)

District by district it is worse. The table below uses the uncapped narrow list, with what happens on the broader list beside it:

Seven of the nine rows hold when the list is widened; two move. Longgang, the largest district in the city, reads Phase 2 on the narrow list and Phase 1 on the broader one; Futian is a two-way tie between Phases 1 and 2 on the narrow list (17 / 17 / 16) and Phase 1 on the broader one. Longgang's narrow-list margin is 53 against 51 — two records. And that is only the gap between two nested lists; the equal-depth test above unseated the city-wide answer as well, so the district level is not worth arguing about. We are not going to tell you which phase a Shenzhen district belongs to, or which phase the city leans to. When someone else tells you, ask which keyword list they used.

There is an official label pulling the other way, and it deserves to be stated rather than argued around. At the press conference on 18 September 2026, the Deputy Secretary-General of the Guangdong Provincial People's Government listed the regional brands the province will push at this session (one sentence; the transcript is in Chinese only and we have not found an official English version, so the translation is ours):

At the same time, building on Guangdong's strong manufacturing capacity and industrial characteristics, we will promote regional brands such as Shenzhen consumer electronics, Guangzhou commerce and trade, Foshan home appliances, Dongguan intelligent manufacturing, Yangjiang hardware and Chaozhou ceramics, and will hold dedicated trade matchmaking events aimed at the Asia-Pacific and other key regional markets, doing everything we can to drive deals and real results.

The Chinese original reads: 「同时我们也立足广东强大的制造能力和产业特征，打响深圳消费电子、广州商贸、佛山家电、东莞智造、阳江五金、潮州陶瓷等区域品牌，面向亚太等重点区域市场举办专场贸易对接，全力促成交、增实效。」

Both things are true at once, but our 22.6% needs stating precisely. It is the union of five zones — 电子消费品及信息产品, 电子电气产品, 家用电器, 照明产品 and 新能源 (consumer electronics and information products, electronic and electrical products, household appliances, lighting, and new energy; those English renderings are ours, since Annex 1 is in Chinese) — and the largest block inside it is household appliances at 93 records, ahead of consumer electronics and information products at 76. Count only the two zones with "electronic" in the name and it is 110 records (11.0%); count only the consumer-electronics zone and it is 76 (7.6%). So 22.6% is not the figure to hold up against "Shenzhen consumer electronics": one is a regional brand a province is promoting, the other is our registration data across five zones, and the number that actually lines up with the phrase is 7.6%. Neither of them is a booking rule.

One group deserves a line of its own, and the answer runs against the intuition. 157 of the placed Shenzhen records point at a Phase 3 zone — 105 of them at Phase 3 and nothing else. It is true that nothing follows Phase 3: the fair closes on 4 November. But a Phase 3 exhibitor is not at the fair between 15 and 27 October — the whole of Phase 1, the whole of Phase 2 and both changeover windows fall inside that. So this is in fact the group with the most room, provided you come before it moves in. The genuinely narrow group is Phase 2: 15–19 October, or from 31 October.

And if you are not coming for the fair at all, none of the above binds you. The only date question left is the National Day holiday in early October: the arrangements are published separately each year, each plant decides its own shutdown, and the reliable move is to ask the supplier whether it is running on the days you have in mind. We are not going to print dates we have not seen in an official notice.

7. How we price the day, and the questions §1 tells you to ask us

A Shenzhen plant day booked out of Guangzhou is an inter-city day. We are not going to tell you how long the road takes, and we would be suspicious of anyone who does it from a desk: we do not have your plant's address, traffic is not a constant, and the registered address in our file is not the plant anyway. What we can give you is the order of operations.

1. Get the street address in writing, for the plant, not the head office — and the name of the legal entity that operates it.
2. Ask the supplier to state its own door-to-door estimate from your hotel, then check that estimate yourself on a map. It is their address; the estimate is theirs to give and yours to sanity-check.
3. Ask the supplier how long the visit itself will take and which processes you will actually walk through. That number is theirs too. We are not going to estimate it for you from the size of the company.
4. Add the return journey at the real clock time you will be travelling, not at an average.
5. Book the interpreter in whole blocks against the total, not against the meeting.

That is also why our quoting works the way it does, which brings us to the two questions §1 tells you to put to any agency — including us. Asking them and then not answering them ourselves would be cheap.

Who pays for the travel time to Shenzhen. We do not absorb it. A Guangzhou-to-Shenzhen day is quoted as one whole block with the travel at both ends inside it, and transport, meals and any overnight are listed separately rather than folded into an hourly rate. Our published floor is consecutive and escort interpreting from ¥200 per hour or ¥800 per full day, depending on language pair and subject, with half-day rates available — that is the listed floor on our FAQ page, not a quote, and an inter-city day does not land on the floor.

How much the interpreter receives. We do not publish that split. That is a real answer rather than a dodge: it is a commercial term between us and the interpreter and we are not going to invent a number for an article. If the split matters to you — and for some buyers and some audit standards it genuinely does — ask before you book, and ask every agency you are comparing, not only us.

The third question, and what happens if it is cancelled. §11 adds one more: is the person arriving on the day the person whose profile you were shown? We are not going to give a generic answer to that in an article — it belongs in the confirmation for a specific booking, including how you are told if we have to substitute and whether you can refuse. Cancellation and rescheduling are the same: whether the day is chargeable when the supplier pulls out the night before, or when your own trip moves, is a term in that confirmation, not a promise in an article. Only one thing generalises: get both in writing before you book, from us and from every agency you are comparing.

What we do not do. We are the agency. The price covers the interpreter's fee plus our cost of screening them and standing behind the assignment. Whether the plant lets you in, whether you may photograph a line, whether a badge is issued and whether an audit passes are none of them ours to decide. We do not give legal advice and we do not give compliance advice; if you actually need either, instruct a licensed lawyer or the relevant certification body.

We are also not the only route. At the same 18 September press conference, Guangdong described a programme that takes buyers to plants directly (two sentences; Chinese only, translation ours):

We are using the "Global Buyers in Guangdong" programme as the lever to open a channel linking the inside and the outside of the fair. We are precisely inviting buyers from key regions worldwide to go deep into Guangdong's specialised markets, industrial parks and manufacturing plants, promoting face-to-face negotiation and point-to-point matching between buyers and industry alliances, efficiently converting expressions of interest into landed orders and turning the traffic of exhibiting into a net gain in trade.

The Chinese original reads: 「我们以“全球采购商广东行”为抓手，打通展会内外的联动通道。精准邀约全球重点区域采购商，深入广东专业市场、产业园区、制造工厂等，推动采购商与行业联盟面对面洽谈、点对点对接，高效将意向询盘转化为落地订单，让参展的流量变为贸易的增量。」

The transcript does not say how a buyer signs up, and we are not going to guess. If it fits what you are doing, ask the organiser — the contact channel published in the accompanying-personnel notice is in §10.

8. Our interpreter pool: 812 profiles, none of them in Shenzhen

The pool behind our Guangzhou interpreter directory holds 812 profiles. Every single one records Guangzhou as its base; the set of distinct city values in the file has exactly one member. So for a Shenzhen day we are sending someone across the boundary, and §7 is how that gets counted and paid for.

By language, counting profiles rather than people-days — a profile can hold more than one language:

English 631, Japanese 248, French 100, Spanish 65, Russian 56, Korean 51, German 32, Italian 17, Vietnamese 8, Arabic 7, Thai 6, Indonesian 4, Turkish 3, Portuguese 1.

Read the tail of that list literally. There is one Portuguese profile in the pool, three Turkish, four Indonesian. (The next tier up is not deep either: Thai 6, Arabic 7, Vietnamese 8, Italian 17.) We do not have a Portuguese team; we have one person, and one person has no stand-in — which is why these are the languages to ask about earliest. That follows from the counts, not from any guess of ours about demand. If you ask us for one of those and we cannot staff it, there are three routes and each of them has an owner: we can look outside the pool, and whether we find anyone is something we will tell you after we have looked, not before; you can bring a relay language, which only works if someone on your side speaks it; or you can move the date, which is only an option if you are not coming for the fair. If none of the three applies, the honest answer is no, and you should hear it early enough to do something about it.

On industry, 154 profiles carry the Electronics & Electrical tag — the largest tag after the general foreign-trade one, and the most relevant tag for Shenzhen. It is worth being exact about what that is: a tag entered on the profile, not a certificate, not an audit, and not a city. An interpreter tagged for electronics is telling you what vocabulary they are comfortable in.

On credentials, 810 of the 812 profiles have something in the qualifications field: 555 hold a bachelor's degree, 189 studied at Guangdong University of Foreign Studies, 149 hold TEM-8, 92 hold a master's. Like the industry tag above it, this is a field the interpreter filled in; we cite it as it stands and we say where it comes from. What it is not is a licence to interpret at a plant visit — see the second bullet of §9.

9. What an interpreter does not do

• Not an auditor or an inspector. An interpreter will not sign an audit report, will not give you a verdict on a line, and should not be asked to. A social-compliance or quality audit is a separate firm with a separate scope.
• Not required to hold anything for a private plant visit — but that sentence has a range. The three documents the Fair asks for do not reach beyond the hall door. A CATTI certificate, a NAETI certificate or a foreign-language degree from a mainland institution are its conditions for issuing the "interpreter" class of accompanying-personnel badge, not its conditions for hall badges in general; and a Shenzhen plant is not a hall at all, so that rule does not arrive there. Do not read that as "nobody will ask", though: your own audit standard, your company's supplier-access policy, or the plant's own site-access and confidentiality rules may each require something of the people you bring, and none of those are ours to interpret or to waive.
• Not a witness or an expert. Do not plan to rely on the interpreter's recollection as evidence of what was promised. Write it down in the room and have both sides agree the written version. That is practice, not a statement about what counts as evidence.
• Not a decision-maker at the gate. Whether you may enter, photograph a workshop or see a specific line is the plant's decision on the day. We do not decide it and we do not promise it.
• Not a lawyer. Nothing an interpreter says about a contract, a licence or a liability is legal advice.
• And the reverse trap: the supplier's own salesperson interpreting for you is not a neutral interpreter, however good their English. That is not an accusation of bad faith — it is that they are one of the two parties in the room.

10. Meeting at the stand first, and what a badge costs you

Plenty of buyers want to meet the supplier at its booth in Pazhou before committing a day to Shenzhen. That is a good idea, and it is the only part of this article where Canton Fair badges matter at all: a plant visit in Shenzhen needs no Canton Fair badge of any kind.

If you do want your interpreter inside the halls, here is the state of play. As of 19 September 2026 the 140th session's accompanying-personnel badge notice has not been published. We checked by pulling 1,200 items from the official site's content stream, which runs unbroken through the stream from 2 November 2025 to 18 September 2026 and carries 179 distinct publication dates (the site published nothing on the other days), and by confirming that the same pull does return the 139th session's notices of 30 March and 2 April 2026 — a negative result is only worth something if the same method finds the thing you know exists. What the site does say is in the purchasing-representative pre-registration notice of 3 September 2026, whose closing paragraph states that since the 139th session the Chinese accompanying personnel of overseas buyers may no longer obtain a purchasing-representative badge and may apply for the accompanying-personnel badge instead, and — in the sentence immediately after that one — that pre-registration will open at the end of September. So the route through a purchasing-representative badge is closed. When the badge notice itself will appear, the site does not say, and we are not going to guess.

Until it appears, the 139th session's rules are the only text there is, and they may change. Under those rules: only the overseas buyer can file the application and the accompanying person cannot start one; the application must go in at least two calendar days ahead, and both language versions land on the same date — the Chinese says 2 个自然日 and the official English says only "at least 2 days", but the official English's next sentence works the example with must, giving 13 April for attendance on 15 April, which is exactly those same two calendar days; the accompanying person must hold, in the official English, "a People's Republic of China resident identity card"; they are registered as an "assistant" or an "interpreter", and an "interpreter" must additionally submit a CATTI certificate, a NAETI certificate or a foreign-language degree from a mainland institution; the quota is two per buyer per phase, one of each, and buyers holding a Chinese passport or coming from Hong Kong, Macao or Taiwan may apply only for an assistant; the fee is ¥300 per day plus a ¥50 administrative fee, and a badge already used is not refunded — the notice defines "used" as "scanned by entry gates".

The clause that most often decides whether the specific person you picked can be badged at all is Section VII, "Restrictions on Badge Application". Two sub-clauses, quoted verbatim, and their reach is not the same:

（一）本届同一展期内，成功办理参展商证、参展代表证的人员，无法办理当期境外采购商陪同人员证。 （二）本届成功办理筹撤展证件的人员，或申请办理参展商证、参展代表证但被驳回的人员，无法办理本届境外采购商陪同人员证。

Note the difference between them: the first is limited to the same phase of the current session, so a person blocked in Phase 1 is not blocked in Phase 2; the second reaches the whole current session. The official English:

i. For the same phase of the current session, individuals who have successfully obtained an exhibitor badge or exhibitor representative badge are not eligible to apply for an accompanying personnel badge for that phase. ii. Individuals who have successfully obtained a move-in/move-out badge for the current session, or whose exhibitor/exhibitor representative badge application was rejected, are not eligible to apply for an accompanying personnel badge for the current session.

In plain terms: if the person you had in mind is also working a stand at this fair, or has a move-in/move-out badge, or was refused an exhibitor badge, you will find out at the application, not at the gate. Ask before you book them. Section VI adds that an irregular badge record from earlier sessions triggers an alert and that anyone flagged must submit extra documents for review; the notice does not define "irregular" and does not say how long the review takes, and we are not going to fill that in. Section VIII is the one with teeth: buyer and accompanying person must have a genuine accompanying relationship, failing which the organiser may confiscate both parties' badges, require them to leave, and place them on the Canton Fair badge blacklist, with no refund. The full procedure and the consequences are in our piece on the accompanying-personnel badge rules; what a badged day actually costs, all in, is in the cost breakdown.

Two practical notes. The ¥300 per day plus the ¥50 administrative fee is paid to the organiser; it is not inside our interpreting price and we do not collect it. And badges are not our call: the organiser publishes its own channel in that notice — customer service on 4000-888-999 inside China, (0086) 20-28888999 from abroad, e-mail info@cantonfair.org.cn. Whether a badge is issued is the organiser's decision and we will not speak for it.

11. If it has already gone wrong

• You arrived and it is an office. Say so, politely, and ask for the plant address and whose licence that plant sits under. If they offer to take you to a "partner factory" instead, go if you want to — but record it in your own notes as a partner factory, not as this supplier's plant. §4 tells you which districts make this outcome most likely.
• The supplier cancels the day before. Use the calendar in §6 to see whether the week holds a second day at all; if it does not, decide whether the trip is still worth making rather than filling the day with a plant you have not checked. Whether the interpreter day is chargeable is a term in your confirmation with the agency, us included — §7 says why that has to be settled before you book rather than in an article.
• You booked on a day your supplier is at the fair. That afternoon probably cannot be rescued — the plant is still running; what is missing is the people. Move it using the calendar in §6: a Phase 1 supplier from 23 October; a Phase 2 one on 15–19 October or from 31 October; a Phase 3 one anywhere in 15–27 October. Confirm with the supplier that the plant is working that day.
• Your interpreter has no accompanying-personnel badge and you needed one. Three things, in this order. First, you will probably not get one for tomorrow: under the 139th session's timing the application goes in at least two calendar days ahead and only you, the buyer, can file it. Second, the subject of this article — going to a plant in Shenzhen — needs no Canton Fair badge at all, and if your supplier is not at Pazhou that day anyway (the calendar in §6 will tell you), moving the meeting outside the halls saves it. Third, whether there is any other route is the organiser's to answer, not ours: customer service on 4000-888-999 inside China, (0086) 20-28888999 from abroad, info@cantonfair.org.cn. One more thing belongs here because it is easiest to forget at this moment: a badge already scanned at the gates is not refunded, and borrowing someone else's accompanying relationship can, under Section VIII, cost both parties their badges and put them on the organiser's blacklist.
• You already booked through another agency. Put the two questions from §7 to them in the same words — how the travel time is counted, and what the interpreter receives — and add a third: is the person arriving on the day the person whose profile you showed me? You are entitled to ask us all three, and our answers to the first two are in §7.
• You land tomorrow and have nobody. WhatsApp or WeChat us on +86 139 2416 6640 with the district, the date and the language. We will tell you whether we can take it; if the honest answer is that we have to check a calendar — which is the likely answer for the thin languages in §8 — that is what you will get, rather than a "should be fine". If you would rather not use a messaging app, the contact page reaches the same place. We are not the only route, either: if the trip is being run by an audit firm, ask whether they bring their own interpreter; if the supplier offers to provide one, §9 explains why that is not neutral, not that you must refuse.

12. Sources and dates

• Announcement on Booth Applications for the Export Exhibition of the 140th China Import and Export Fair, Business Office of the China Import and Export Fair, 5 May 2026 — the phase dates, the "54 zones" figure, and Annex 1, which assigns each zone to a phase: Canton Fair website. The zone-to-phase table is the organiser's; our product-keyword-to-zone map is ours.
• Announcement Inviting Exhibitors to the Trade Services Zone of the 140th Canton Fair, 4 September 2026 — opening hours and the two changeover dates: Canton Fair website.
• State Council Information Office press conference on the 140th Canton Fair, 18 September 2026 — the 32,000-exhibitor figure, the regional-brand list including Shenzhen consumer electronics, and the "Global Buyers in Guangdong" programme: Canton Fair website. Chinese only; we did not find an official English version, and the two quotations from it above are our working translation.
• Pre-registration for Purchasing Representatives of the 140th Canton Fair Is Open, 3 September 2026 — accompanying personnel may not hold a purchasing-representative badge, and their pre-registration opens at the end of September: Canton Fair website. Chinese only; that paragraph is our working translation.
• Notice on Badge Application for Personnel Accompanying Overseas Buyers for the 139th Canton Fair, 2 April 2026 — applicant, deadline, types, quota, fee, alert, restrictions, responsibility, and the organiser's contact channel: Chinese and official English. The 139th session's rules do not automatically apply to the 140th.
• All of the above were consulted on cantonfair.org.cn on 19 September 2026, which is also the date of the content-stream check described in §10.
• Exhibitor counts, company types, districts, keywords and phase mapping: our own Canton Fair exhibitor directory, built from the 139th session exhibitor list (full version, 16 April 2026), 26,270 companies. These are our numbers, not the organiser's. Interpreter counts: our own interpreter pool, 812 profiles.
• Our published interpreting floor prices are the ones listed on our own FAQ page.

Related: the Guangzhou plant-visit piece, which has the Guangzhou district breakdown, and the Foshan one, whose city leans the other way on company type.