Inquiry review · 2026
We counted 60 export inquiries, line by line: what overseas buyers actually ask
Buyers from 24 countries. Eight in ten open with sampling. Six in ten pack eight topics into one email — and half of those emails wait more than 24 hours for a human reply.
Sample: every inbound inquiry received by a Chinese bag and pet-product OEM factory, 7 Jul – 7 Sep 2026
Everyone in export sales carries a theory about what buyers want, and that theory usually comes from the last few deals and the most memorable customers. So we went looking for a complete dataset instead: every inquiry a Chinese bag and pet-product OEM factory received over two months, counted topic by topic, to see what overseas buyers actually open with.
Where they come from
60 inquiries, 55 with an identifiable country, spread across 24 countries
The US is the largest single source at 12. But add the European countries together and you get 22 — more than North America’s 16. The Netherlands, France, Austria, the UK, Germany, Spain, Italy: two or three each, none large enough to stand out on its own.
Europe always looks small in a report because it is split across a dozen countries. Only when you add them up does it turn out to be the biggest block.
What they ask
Counted once per inquiry (a customer who writes ten times still counts once) — 55 inquiries with incoming mail
The ranking matters less than the density
Read that chart the other way round: 87% of inquiries raise three or more topics at once, 76% raise five or more, and 60% raise eight or more. Buyers are not asking one question. They are working through a checklist and scoring you on it.
The most typical example listed twelve items under "please confirm or provide": fabric, mesh grade, base reinforcement, zippers, safety anchors, detachable modules, sampling cost and lead time, MOQ, six price tiers, Incoterms and a delivered price, custom labels and packaging, QC standard and certification.
These buyers are asking three to five factories at the same time. They will not chase the items you skipped — they will simply mark you down as unable to do it.
So the first reply is not a writing exercise, it is an alignment exercise: answer point by point, in their order. Where you have a basis, give the specific answer. Where you do not, say exactly what information you need before you can confirm. Not knowing is survivable. Skipping is not.
Four in ten buyers have no drawings
41% of inquiries state plainly that they have no technical drawings — only an idea, reference photos, or an AI-generated rendering. These buyers have a brand, a budget and a launch date. What they lack is the step that turns an idea into a product.
By contrast only 9% asked for a small trial order. Most of them are not trying to order less; they simply do not know where to start. Whoever spells out the path from concept to sample wins this group.
Another 23% want an NDA signed before they share drawings. That is not an obstacle, it is standard practice for brand buyers — agreeing readily and handing over your own legal entity details on the spot works far better than negotiating over it.
Where the delay actually is
This section does not flatter anyone, which is exactly why it is worth reading.
With automatic acknowledgements excluded, the median first human reply lands at 25 hours. Only half of all inquiries get a human answer within 24 hours, and just a third within 12 — while the factory’s own auto-acknowledgement promises a reply within 12 hours.
The delay is not typing. It is lookup. MOQ sits in the price sheet, the sample-fee credit rule sits in the contract, test reports sit in another folder, and freight needs a call to the forwarder. Assembling answers to an eight-topic email takes an hour or two before a single sentence is written.
Which is why an instant "we have received your enquiry" solves nothing. The buyer did not ask for a receipt. They asked eight questions.
One surprise: ChatGPT is sending buyers to the factory
Among inquiries with an identifiable source, AI assistants are the largest channel — all of them ChatGPT. Buyers ask it where to find a factory for custom pet travel bags, get pointed to the site, and arrive with the source still tagged on the landing-page URL. A further dozen said in plain text that ChatGPT had recommended the factory.
Traditional search engines account for a single-digit number over the same period. And the factory had never done anything to optimise for this.
Four things you can apply today
- 1Answer point by pointHowever many items they listed, answer that many, in their order. They want to tick boxes, not read an essay.
- 2Turn your standard answers into something searchableMOQ, sampling rules, credit conditions, certifications, Incoterms — write them once, then reuse them in every reply. Lookup is what makes you slow, not typing.
- 3When you do not have the answer, say what is missing"Tiered pricing once dimensions and weight are confirmed" beats an invented range every time. A number you made up will contradict your own quotation later.
- 4Do not watch only GoogleAI assistants are already recommending factories. Give them facts they can quote — MOQ, materials, certifications, process — written as plain text rather than a screen of animation.
About the data
- The sample is every inbound inquiry received by a Chinese bag and pet-product OEM factory between 7 July and 7 September 2026: 60 customers and 213 incoming emails. Outbound prospecting is excluded.
- Topics are counted once per inquiry. A customer who writes ten times about the same thing counts once; counting per message would multiply a single question tenfold.
- Reply times exclude automatic acknowledgements. Two acknowledgement templates were in use over the period; filtering only the newer wording treats the older ones as instant human replies and yields a median of 0.5 hours — off by a factor of fifty.
- Countries come from what buyers entered or from their email signature. Five could not be identified and are excluded from the country breakdown.
- No customer name, brand or contact detail appears anywhere in this article, and the factory is not named.