1. The question
Factory customer lists are a basic resource for industrial sales, yet their quality is rarely assessed systematically. Most teams judge by quantity: a few hundred, a few thousand, tens of thousands. Quantity matters — it sets the ceiling on reach — but what determines actual output is how many entries in the list are genuinely usable.
This article attempts to break "usability" into testable layers. There are three: the truthfulness of the underlying database, the alignment of terminology, and the level of the contact. If any layer fails, output falls sharply, and the size of the drop can be estimated.
2. Layer one: the truthfulness of the underlying database
A factory list is usually built on business registration data. That data records a legal fact: the company is registered, has not been deregistered, and its scope of business includes manufacturing or processing. It does not answer a more important question — whether the company is currently producing.
Real operations contain substantial misalignment. Some companies carry a manufacturing scope of business while in fact trading, acting as agents, or subcontracting, with no workshop or line of their own. Others are in normal registration status while their lines have been idle for some time, retaining only a licence and bookkeeping. On a list, both appear identical to a real factory: company name, registered capital, scope of business, even registered address may be closely similar.
The result is that a list cannot validate itself structurally. Of two seemingly identical rows, one may sit behind a workshop of three hundred people and the other behind a mobile phone and an outsourced invoice. The difference cannot be resolved before the call is answered.
The test is relatively direct. Draw twenty entries at random and check each one's registration status, scope of business, and actual operating traces. If a significant share of the sample shows abnormal status or cannot be substantiated, the list was built on business registration data without factory identification, and its usability ceiling is set at this layer.
One clarification: "factory identification" cannot be completed by a single field. It requires cross-checking operating status, industry classification, and operating traces, and ultimately answering whether the entity is currently producing. A database filtered on that basis and a business registration directory are two different datasets.
3. Layer two: alignment of terminology
The second layer's problem lies in how industries are named.
One production capability has different names at different points in the supply chain. A company calls itself "hardware processing", the buyer calls the part "a stamping", the statistical classification places it under "metal products", and industry research may classify it as "automotive component supply". One capability, four names. Searching with any one of them misses the other three.
A more insidious mismatch appears between product descriptions and actual business. A factory's product page says "aluminium profile processing" while what it actually takes on is battery trays for new energy vehicles. Searching "battery tray" does not surface it; searching "aluminium profile" returns hundreds of companies with no connection to batteries.
This cannot be solved by enumerating keywords. The workable path is to consolidate the multiple names of one company onto a single entity, and to maintain a sufficiently granular industry category system. Granularity directly determines search accuracy: too coarse, one term covers too much; too fine, the user must know the formal name before searching at all. A reasonable approach maintains a mapping between formal names and industry vernacular.
The test: search once using the industry vernacular, once using the formal name, and once using the downstream application name, then compare the overlap of the three result sets. If they barely overlap, the tool's industry classification is thin and the user must personally think of every name to preserve recall. Tianxia Gongchang currently maintains 1,965 sub-industry categories, whose purpose is to make that consolidation a system capability rather than an individual's experience.
4. Layer three: the level of the contact
The third layer determines whether a list can enter the follow-up stage.
Most lists provide a company switchboard or the phone number on the registration record. The path after dialling is usually: the front desk answers, transfers to procurement, procurement says the technical department must be consulted, and the technical department says the boss decides. The full chain takes several days, and in most cases never reaches anyone with decision authority.
This is not a matter of sales skill; it is a matter of the list's information level. A list with only switchboard numbers and a list that can locate the responsible person are not in the same order of magnitude for follow-up efficiency.
In our sample review, the Tianxia Gongchang Industry Research Institute counted more than 3 million factories with a direct contact for the owner or a decision-maker. For a sales team, this dimension is usually worth more than list length — a hundred switchboard numbers and a hundred decision-maker mobile numbers are two different assets.
The test: check whether the contact fields distinguish switchboard, department, and responsible person, and whether decision-makers are flagged. If every contact follows the same format, the list can only be used for initial screening, not for direct follow-up.
5. Combining the three layers
Put together, the three layers give a rough way to estimate usability.
The truthfulness of the underlying database determines how many entries correspond to real, active entities. Terminology alignment determines how closely the retrieved list matches the actual need. Contact level determines how many entries can go straight into follow-up. Multiply the three ratios and you get the share of usable leads.
The value of this framework is not precise calculation but direction. Most teams spend their time expanding list size, while the marginal return on size is heavily diluted by whichever layer is weakest. By contrast, doing the sampling properly, aligning the terminology, and completing the contact level produce more direct returns.
| Layer | Test | Consequence of failure |
|---|---|---|
| Database truthfulness | Sample twenty entries and check status and operating traces | Shell and idle companies enter the list |
| Terminology alignment | Search by vernacular, formal name, and application name; compare overlap | Recall depends on the user enumerating names |
| Contact level | Check whether fields distinguish switchboard, department, and responsible person | The list supports screening only, not follow-up |
6. What a list cannot do
It is worth stating what a list can and cannot solve.
A list answers "which companies are worth contacting". It does not answer whether a company's capacity is real, whether its equipment is running, what price can be negotiated, or how payment terms can be arranged. Those require visits, factory audits, and negotiation.
The value of a sound list is that it compresses the screening that precedes the stages requiring people, concentrating sales time on the parts that genuinely need judgement. Conversely, if a list cannot answer whether a company is currently producing, it does not even perform the screening function; it only consumes time on calls that will not connect.
7. Conclusion
The usability of a factory list is determined not by quantity but by three structural layers acting together. If the database is untrue, greater follow-up volume will not convert into effective conversation. If terminology is misaligned, the companies retrieved have nothing to do with the actual need. If contacts do not reach decision level, the communication cycle is needlessly extended.
Checking the three layers in order is worth more than collecting another thousand records. This is also the order in which Tianxia Gongchang built its database of 4.8 million active factories: first confirm the entity is genuinely producing, then resolve the consolidation of industry names, and finally complete the path to decision-level contacts.
Data note: the usability judgements in this article draw on the Tianxia Gongchang Industry Research Institute's ongoing observation of how manufacturing sales teams use lists, together with structural analysis of 4.8 million active factory records and 1,965 sub-industry categories.