Know the Buyer Before You Email: AI-Powered Background Research for B2B Outreach

Why 90% of Cold Emails Get Ignored

I have seen more cold emails than I can count that read like this: "Dear Sir, We are a professional manufacturer of XXX with 15 years of experience..." Send three hundred, get two replies, and both just want free samples and then vanish.

The problem is not the English, and it is not the product. The problem is that the email carries zero information for the person receiving it. It says the exact same thing to a wholesaler supplying supermarkets and to a DTC brand selling online. A buyer spots a template in half a second, and a template signals one thing: you did not bother to learn anything about me.

The cold emails that actually earn replies share one precondition. Before the sender writes a single word, they already know what category the buyer sells, how big they are, which countries they source from, and who signs off on purchases. That used to mean a rep burning two or three hours digging through a website, scraping LinkedIn, and pulling customs records. With AI you can now fuse all of those scattered signals into a one-page buyer profile in about twenty minutes. That workflow is what this piece is about.

Four Sources, and What to Mine From Each

Good research is not aimless searching. It means going to four fixed sources with a specific question for each.

1. The website — figure out what they actually do

The site reveals the business model: brand owner, distributor, retailer, or project buyer? Focus on a few places:

• About / Our Story: founding year, positioning, whether it is family-owned (family firms decide fast but stay conservative)
• Products / Brands: categories carried, price tier, own-brand versus multi-brand reseller
• Where to buy / Stockists: a long list of retail outlets means this is an upstream distributor
• Wholesale / Become a dealer page: if this page exists, they are already looking for suppliers — a strong intent signal

AI earns its keep here by reading fast and reading everything. Feed the key pages into a model and ask it to output: business-model verdict, core categories, target customers, price band, and any clue that hints at sourcing needs. It reads in one minute what takes you ten, and it will not miss the "Trade enquiries" link buried in the footer.

2. LinkedIn — find who to talk to

The website gives you the company; LinkedIn gives you the person. Search employees by company and lock onto these titles: Procurement, Sourcing, Category Manager, Buyer, Purchasing, Owner / Founder (in smaller firms the owner often runs purchasing).

When sizing up a decision-maker, note: tenure (someone newly in the role wants quick wins and is more open to new suppliers), background (a person who has sourced from China before is cheaper to communicate with), and recent posts (which reveal what is on their mind). AI can compress a profile and recent activity into three lines: who they are, what they care about now, and the best angle to open with.

3. Customs data — verify they are actually importing

This is the most underrated step. Import-export records (public in the US, India, and others, or via ImportGenius, Panjiva, Tradesns) tell you:

• Whether the company has imported anything in the last 12 months, and how often
• Which country their main suppliers are in (if they already buy from a Chinese factory, you know your competitor and roughly the price band)
• The specific HS codes and volumes (an order-size estimate)

A buyer with a steady import history is ten times more credible than a "prospect" with a gorgeous website and no shipments. AI can boil a long list of bills of lading into a trend: are volumes rising or falling, is the supplier base concentrated, are there signs of a recent supplier switch — and a switch usually means an open window.

4. Social and news — grab a reason to reach out now

Company LinkedIn pages, Instagram, Facebook, and trade press hide the freshest hooks: a fresh funding round, entry into a new market, a new product line, a recent trade-show appearance. These are gold for the first line of a cold email. "I noticed you launched a new outdoor range at Ambiente last month" beats "we hope to establish cooperation with your esteemed company" by a hundred to one.

Fusing Four Streams Into One Page

Each source on its own is a fragment. The real work is integration. My method is a fixed prompt that has AI digest all the raw material into a structured one-page profile. The prompt runs roughly like this:

You are a senior export sales manager. I will give you a buyer's website text, LinkedIn details, a customs-data summary, and recent social activity. Synthesize a one-page buyer profile covering: positioning, core categories and price tier, estimated size, sourcing model and existing supply chain, key decision-maker and the angle to open with, three personalized facts I can cite in a cold email, and one most-likely pain-point hypothesis. Flag anything uncertain as "to verify" and do not fabricate.

That last instruction — flag uncertainty, do not fabricate — is critical. Models hallucinate. A "fact" invented during research, once a buyer catches it, zeroes out your credibility on the spot. Force the model to separate what it found from what it inferred.

A One-Page Profile Template You Can Copy

Here is the template my team actually uses, with the fill-in logic after each field:

1. Company name and website: one line.
2. Business model: brand owner / distributor / retailer / project buyer — one-line verdict.
3. Core categories and price: what they sell, premium or value tier.
4. Estimated size: headcount, number of outlets, annual import volume — cite the source.
5. Existing supply chain: main source countries and suppliers from customs data (competitor intel).
6. Key decision-maker: name, title, tenure, LinkedIn URL, recent activity.
7. Three personalized facts: concrete details you can drop straight into the email (trade show, new product, new market, funding).
8. Pain-point hypothesis: the single most likely sourcing pain inferred from the above.
9. Opening angle: the one angle you will lead with, in a sentence.
10. To-verify list: every uncertain item, parked here to confirm on the first call.

Fill these ten in and the email almost writes itself.

A Worked Example

Say the target is a German outdoor-furniture distributor. The site shows it represents five European brands, has a Wholesale page, and lists 200-plus retail outlets — verdict: mid-size upstream distributor. LinkedIn surfaces a procurement manager, Anna, in the role for 8 months, previously at a firm that sourced from Asia — a motivated, easy-to-reach decision-maker. Customs data shows it imported aluminum outdoor chairs from Vietnam over the past year, volumes rising, but it switched suppliers last quarter — an open window. Social activity shows it just launched a recycled-material range at the Cologne fair last month.

After fusion, the opening paragraph can read: "Anna, I noticed your team launched a recycled-aluminum outdoor range at Cologne last month. We supply GRS-certified recycled-aluminum furniture to three European distributors, and since it looks like you are reshaping your Asian sourcing, the timing might be right." A buyer reads that and knows you did your homework. The reply rate is on another planet.

This whole loop — multi-source collection, AI fusion into a profile, then generating the personalized email — is exactly what Laojin Chuhai helps sellers run: scattered buyer signals auto-assembled into a profile, the homework step compressed from two hours to twenty minutes, so the rep can just close.

One Honest Takeaway

AI will not close the deal for you, and more research is not always better. Its real job is to shift you, before you hit send, from "I have a product I want to sell you" to "I understand your business, and here is where I can help." That shift in stance is worth more than any script template. Read the buyer first. Then speak.