First, One Thing: Where Does the List Come From

The previous lesson explained how to tier factory customers into S/A/B/C — scoring on scale, signals, reachability, and decision-chain complexity, then prioritizing S and A.

That lesson took one thing for granted: that you already had a list in hand.

In practice, that is exactly where things get stuck.

Most industrial-goods salespeople draw their lists from a few common sources:

  • Company Excels accumulated over the years — nobody knows who maintains them or whether they are still current
  • A few thousand rows exported from Tianyancha by industry — not yet verified as real factories
  • Lists passed on by colleagues or agents — unknown overlap
  • Business cards scanned at trade shows — no more than 20 of every 100 are worth calling

What gets stitched together is a "list" with names and phone numbers but no scale information, no signal tags, no tiering. Sales reps start at row one and work their way down, however far they get.

That is not a method. That is attrition.

An executable factory list should tell a salesperson, at minimum: which industrial cluster this factory belongs to, roughly what its scale is, whether there is a purchasing signal right now, and whether the priority is S or A. With those few columns in place, a salesperson can decide "who do I call first today."

This lesson gives you the 5-step process for building that list from scratch.

Why "Calling Through at Random" Costs So Much

Let's run the numbers first.

An industrial-goods salesperson makes roughly 40–60 effective dials per day. After accounting for answer rates, conversation time, and logging follow-ups, the number of genuine conversations with target customers is about 8–12.

Assume a poor-quality list. Out of 60 dials:

  • About 20 go to factories that don't match your product (too small, or they don't use what you sell)
  • About 15 go to traders or shell companies (Lesson 4 covers this specifically)
  • About 15 go to factories that already have a fixed supplier and won't switch any time soon
  • Only the remaining 10 or so actually reach a target with genuine potential need

Of those 10 effective conversations, only 2–3 can realistically be advanced to the next step.

An industrial salesperson's all-in monthly cost — salary, travel, phone bills, management overhead — runs roughly 15,000 to 20,000 yuan per month. Over 20 working days, that is about 750 to 1,000 yuan per day.

Making 60 calls takes roughly half a day; advancing 2–3 customers, at an effective advancement cost of roughly 125–250 yuan each.

If list quality improves so that the effective conversation rate goes from 10/60 to 20/60, the number of customers advanced per day doubles. The same 750 yuan moves 4–6 customers instead of 2–3.

Over a month, that is 40–60 additional customers advanced — on the same salary, with the only difference being list quality.

List quality is a multiplier on sales efficiency, not addition and subtraction.

The 5-Step Filtering Funnel: From ICP to Executable List

Step 1: Translate the ICP into Actionable Filter Criteria

Lesson 1 gave you an Ideal Customer Profile template that looks something like:

Factories making Product X, with annual output between Y and Z, located in Region W, showing Signal M.

That sentence is still one step away from being "filterable" — you need to turn each dimension into a concrete search parameter:

ICP Dimension Corresponding Filter Parameter
Product type ("precision machining shops") Industry category + main-product keywords
Scale range ("50 million to 200 million yuan") Annual output range or headcount range
Region ("Yangtze River Delta") Province / industrial cluster
Purchasing signal ("currently hiring CNC operators") Job posting keywords
Export attribute ("has export orders") Export / customs data tag

Every filter parameter must be quantifiable and searchable — it cannot stay at a vague descriptor like "mid-size factory." "Mid-size" cannot be filtered. "50–300 employees, annual output 30 million to 200 million yuan" can.

Translating the ICP usually requires 1–2 rounds of calibration. The first pass may produce criteria that are too broad (tens of thousands of factories nationwide qualify) or too narrow (only dozens nationwide), and you will need to loosen or tighten a parameter.

Step 2: Multi-Dimensional Combined Filtering — Shrink the Full Database to a Manageable Pool

With filter criteria in hand, the next step is to reduce the country's several million factories down to a manageable scale of a few hundred.

Single-dimension filtering is very ineffective. Filter by industry alone for "mechanical manufacturing" and you get hundreds of thousands of results. Filter by region alone for "Zhejiang" and you still get tens of thousands. Filter by scale alone for "annual output 50 million to 200 million yuan" — again, tens of thousands.

You must combine dimensions — industry + region + scale + signal all applied at once — to press the count down to a reasonable range of a few hundred.

Tianxia Gongchang covers 4.8 million Chinese real-manufacturing enterprises, each carrying structured tags: industrial cluster, scale tier, export signals, hiring activity, capacity expansion records, and more. Querying these dimensions simultaneously outputs a list of "factories that match all criteria" — each factory already comes with signal tags and scale information, so salespeople do not need to manually verify each one.

The goal of combined filtering is to bring the pool to 200–500 factories. Too many (over 1,000) means the criteria are still not precise enough. Too few (fewer than 50) means the ICP is too narrow and a dimension needs to be relaxed.

A few commonly used combination approaches:

  • Industrial cluster + scale + hiring signal: The most common. Locks in a geographic cluster + scale band + a signal of whether there is purchasing demand right now.
  • Industry + export tag + scale: Good for salespeople selling export packaging, export testing, or cross-border settlement services.
  • Industry + capacity expansion / new workshop records + region: Good for equipment salespeople — factories expanding capacity have the strongest equipment demand.
  • Industry + digital-role job postings + scale: Good for industrial SaaS salespeople. A factory posting for an MES implementation engineer is usually seriously considering a system upgrade.

One counterintuitive point when stacking conditions: do not apply more than 4 hard filter conditions at once. Too many conditions will filter out genuine potential customers — some factories may not be actively hiring right now, but that does not mean they have no purchasing demand, only that the timing window differs.

The recommended approach: use 3 core conditions to build the main pool, then use a 4th signal condition as a priority ranking rather than a hard filter.

Step 3: Deduplicate — Remove Obvious Noise

The pool produced by combined filtering needs one round of deduplication. Common noise sources:

1. Cross-source duplicates

If you are merging from multiple data sources (e.g., a Tianxia Gongchang export + trade show cards + the company's historical Excel), the same factory may appear 2–3 times with slightly different names ("XX Technology Co., Ltd." / "XX Technology"). Use the full registered business name to deduplicate.

2. Customers already in active follow-up

Factories currently being worked in your CRM should not re-enter a new list for duplicate outreach. Compare against CRM data before building each list.

3. Customers who have explicitly declined

Factories you have called and who clearly said "not interested" or "we have a fixed supplier and will not switch for three years" — flag them and remove them from the list, moving them into a cold pool for reactivation in 6 months.

After deduplication, the pool typically shrinks by 10–20% and becomes cleaner.

Step 4: Score and Sort Using the Lesson 2 Scoring Framework

Once deduplicated, score and rank each factory using the S/A/B/C scoring framework from the previous lesson.

Data sources for scoring dimensions:

  • Scale fit: Output range, headcount (carried over directly from the combined filtering results — no need to look them up again)
  • Signal strength: Hiring activity, capacity expansion records, bidding win records (also carried over from the filtering results)
  • Reachability: Is there a direct contact? Is the number a direct line rather than a switchboard?
  • Decision-chain complexity: Inferred from scale and partnership announcements — the larger the factory, the longer the decision chain tends to be

S tier (high score): Call first — must reach at least 2 per day
A tier: Rotate through 3–4 per day
B tier: Work through once per week
C tier: Contact once a month; if no response, consider removing from the main list

After sorting, the list transforms from "a bunch of factories in export order" into "a prioritized to-do queue."

Step 5: Set Daily Quotas — Put the List in Motion

Once the list is built, most salespeople make one mistake: they call through a batch today, call another batch tomorrow, and never track how many times each factory has been contacted, when the last touch was, or whether there has been any response.

The list needs a quota system.

Basic daily quotas:

  • S tier: 2 touches per day (one first contact + one follow-up)
  • A tier: 3–4 new contacts per day
  • B tier: At least 1 touch per factory per week
  • Total cap: No more than 60 effective dials per day — preserve time for logging and follow-up

Maximum touches per factory:

Three consecutive contacts with no response (including unanswered, or a non-productive connection) — mark as "paused" and reactivate in 1 month. Do not burn more than 3 futile attempts on the same factory.

List lifecycle:

A list from the time it is built to the completion of the first round of outreach typically takes 2–4 weeks. After that first round, responsive factories enter the CRM follow-up workflow; non-responsive ones are decided by tier whether to enter a second round.

Quotas are not a restriction — they are protection. They protect salespeople from concentrating all their energy on a handful of accounts, while also ensuring enough new customers keep entering the funnel.

Mini Case Study: How a CNC Cutting-Tool Company Built a 320-Factory Executable List

A company making CNC cutting tools, positioned as a precision CNC tooling supplier, targets mid-size factories with CNC machining workshops.

Step 1 — Translate the ICP into filter parameters:

  • Industry: Mechanical machining (precision parts, molds, machine-tool components, etc.)
  • Region: Yangtze River Delta (Zhejiang + Jiangsu + Shanghai)
  • Scale: Annual output 50 million to 200 million yuan
  • Signal: Hired for "CNC operator," "CNC machining," or "machining technician" roles in the past 3 months

Step 2 — Multi-dimensional combined filtering:

Tianxia Gongchang applied all four conditions simultaneously and returned about 320 factories, each carrying their industrial cluster, scale tier, and a summary of their most recent hiring activity.

Without combined filtering, searching "Yangtze River Delta mechanical machining factories" alone would return over 30,000 results — no salesperson can work a list of 30,000. With four conditions stacked, the pool shrinks to 320, every one of which simultaneously matches on scale, region, and signal as a genuine potential customer.

Step 3 — Deduplication:

Compared against the company CRM, removed 23 already in active follow-up and 8 with a history of explicit refusal. Pool shrinks to 289.

Step 4 — Score and rank:

Scored 289 factories using the scoring framework and found:

  • S tier: ~35 factories (very active hiring + scale in the 100–200 million yuan range, high tooling purchase frequency)
  • A tier: ~90 factories (hiring records present, scale 50–100 million yuan)
  • B tier: ~110 factories (scale match but no strong recent signal)
  • C tier: ~54 factories (smaller scale, weak signals)

Step 5 — Set quotas:

Prioritize 2 S-tier + 3–4 A-tier factories per day; rotate through B tier once per week; contact C tier once a month.

At this pace, the first round of outreach for S+A's 125 factories takes about 3 weeks; B tier's first round takes about 5 weeks.

Feedback from this company's salespeople: previously, building a list meant manually exporting from Tianyancha and then filtering — a process that took 2–3 days per list, with no signal tags on the output and many traders in the results. Now Tianxia Gongchang outputs a structured list directly; building the list itself dropped from 2–3 days to half a day. More importantly, each factory comes with a signal summary, so the first call has something to say — it is not a cold pitch in the dark.

This is a representative case we have observed; specific numbers will vary with actual operations. The core conclusion is: combined filtering transforms the work from "manually searching through a sea of data" to "calling through an already-filtered list in priority order."

A Field-Ready List of Columns

A properly built factory list should include the following columns for every row:

Basic information

  • Full registered company name
  • Contact name + title (if available)
  • Direct line / mobile number
  • Province, city, district + industrial cluster name

Scale information

  • Annual output range (use a range, not a precise number — precise numbers are often unreliable)
  • Headcount range
  • Factory floor area (if available)

Signal tags

  • Latest hiring activity summary (which roles, how recently)
  • Export tag (yes / no / suspected)
  • Capacity expansion / new workshop record (yes / no, with date)
  • Bidding win record (yes / no)

Tiering and tracking

  • S/A/B/C tier
  • Date of first contact
  • Date of most recent contact
  • Number of touches
  • Current status (not yet contacted / initial conversation done / in follow-up / paused / abandoned)
  • Date of next follow-up

Notes

  • Key points from initial conversation (what was discussed on the first call, what need the contact mentioned)
  • Competitive situation (which competitor was mentioned)

Not all 20 columns need to be filled in — signal tags and scale information come over directly from Tianxia Gongchang during the combined filtering stage; salespeople do not need to look them up factory by factory. Tiering and tracking columns fill in progressively as work advances.

The core principle: a list is not an address book. It is a dynamic workbench. It should tell a salesperson "who to call today, what happened last time, and when to follow up next" — readable in one glance, no memory required.

Two Common Pitfalls

Pitfall 1: Building the list and leaving it static

Many salespeople treat the list as a one-time task — call through it once and never update it. In reality, factory signals change. A factory with no expansion signal last month may open a new hiring push next month. Lists need their signal tags refreshed every 4–6 weeks, with high-signal factories that have newly appeared moved up in priority.

Pitfall 2: More conditions is better

Some salespeople stack seven or eight filter conditions when building a list and end up with a pool of a few dozen factories they cannot even work through in one month. Three to four core conditions is enough. Extra conditions belong in the scoring weights, not in the hard filters.

One Additional Note

This lesson assumes that the 289 or 320 factories produced by filtering are all genuine production-type factories — with real workshops, active products, and purchasing needs.

In practice, lists do contain traders, shell companies, and shuttered operations. These "fake factories" are the most expensive form of list waste. Once they get in, salespeople do not know who they are actually calling, and all call time burns on unproductive conversations.

The next lesson is devoted entirely to this problem: how much it costs when "fake factories" infiltrate a list, the five disguise types they use, and how to screen them out at the combined-filtering stage — that lesson is called "The Costliest Waste in a Sales List: Calling a 'Fake Factory'."