Build a LinkedIn prospect list around evidence, not job titles

A job title can tell you who someone is. It rarely tells you why they should buy from you this month.

That is the problem with a prospect list built around “founder,” “managing director” or “head of marketing.” Those people might be able to approve your service. They might also run the wrong kind of company, have the work covered internally or have no reason to change anything.

Before you write a clever LinkedIn opener, establish why this person belongs in the conversation. A smaller list with a clear reason beside each name is easier to use than a large export you have to research again before every message.

Harry the Marketer's lead-research tutorial points to four places to look: competitors' audiences, conversations about the problem, business events and relevant communities. Those are useful starting points. None is a list of confirmed buyers.

Describe the job your service would do

Start with the work, then decide which people could reasonably need it.

“We sell content” leaves nearly everyone in the market. “We turn a founder's existing technical explanations into useful videos for the people evaluating their software” gives you something to qualify against. You need access to expertise, an audience that benefits from explanation and a company able to use the resulting material.

Write a first version of your inclusion rule in ordinary language. This is an illustrative example for that service:

We want to speak with founders or marketing owners at B2B software companies whose buyers need product education. The company has someone willing to contribute expertise and a plausible use for regular video. Exclude consumer apps, production agencies and companies already telling us they have this covered.

That rule has several separate tests. A founder title does not pass all of them.

Look at the decision role as well. The person who notices the problem may be different from the person who approves spending. A marketing manager can be a useful contact even when the founder signs the agreement. A founder can be a poor contact for an operational problem they have delegated entirely.

Company constraints matter too: size, location, language, delivery capacity, existing tools, likely procurement requirements and the economics of the service. Do not collect a field simply because a database offers it. Collect it when the answer would change whether you proceed or what you ask.

Harry recommends using AI to work through this exercise in his first-agency-client video. Make the model interview you about your service, useful customers, unsuitable work and delivery limits. Then challenge its proposed criteria against actual conversations. AI can make a tidy customer profile from assumptions that have never met a buyer.

Research competitors' conversations, not an imaginary private list

Choose competitors serving a genuinely similar buyer. The largest creator in your industry may attract jobseekers, peers and international readers who cannot use your offer. A smaller specialist with the right conversations can be more useful.

Review publicly available posts and discussions. Who is asking a question your service could help answer? Who describes the company situation? Who has the right role and business model?

Harry's tutorial acknowledges that much of a competitor's audience may be unsuitable. That qualification matters more than the promise of a ready-made client list. Employees, students, consultants, friends and existing suppliers can all follow or engage with a company.

A competitor's private company-Page follower list is not something you should assume you can export. Use information legitimately available to you and respect the service's access rules. Our competitor-audience research guide explains the difference between visible conversation context and a promised follower export.

Avoid sending an opener that announces, “I saw you follow our competitor.” It adds little business relevance and may feel intrusive. The useful finding is the person's actual question or circumstance, where that finding is public and appropriate to mention.

Read the discussion around a specific problem

Search for the problem in the language customers use. “Lead generation” is broad. A discussion about enquiries disappearing between a webinar and the first sales call has a more precise operational job attached to it.

Look across several related posts rather than taking one popular post as a market map. Compare what people ask, what they disagree with and which examples they recognize. Specialist authors can help you find those conversations, but their follower count is not your qualification rule.

There is a useful order of evidence here. Someone explaining their current difficulty gives you more context than someone pressing Like. A detailed question about a process tells you more than a general compliment. Even then, check whether the commenter is discussing their own business, a client or an academic example.

Here are fictional research decisions for the video service above:

The last person has given you a clearer reason to respond. It still does not mean you should ignore the request's details and send your standard pitch.

Connect business events to a question you can defend

Hiring, funding, a new product, a leadership change or expansion into a market can make a conversation timely. They are not automatic purchase signals.

Take hiring. A company recruiting a marketing manager may be investing in marketing. It may also be bringing work in-house specifically to reduce agency spending. The observable fact supports both possibilities.

Record the event, date and source. Then write the consequence as a hypothesis:

The company announced a new technical product. Buyers may need clearer educational material. We do not know whether the launch team already has that covered.

That gives you a reasonable question about the launch material. It does not justify telling the founder their content strategy is failing.

Freshness matters. A funding announcement from eighteen months ago is weak support for a message about “your recent funding.” A past job title can point you toward the wrong owner. Check current sources before allowing a saved trigger to appear in an opener.

Do not turn uncertain research into confident personalization. A wrong company event makes the entire message look careless, however polished the wording.

Use groups and events as places to participate

A specialist event can reveal which questions matter to a market. A group can show you how people describe a problem before they have chosen a solution.

Read the rules and the conversation. Answer a relevant question where you can make a useful contribution. If someone asks for help, respond to what they asked and suggest continuing privately only when that makes sense.

Membership alone is weak evidence. Some attendees are learning, selling, recruiting or supporting a colleague. The useful research is the actual context, not the mere fact that their name appears near the topic.

This is also a reason to keep relationship opportunities separate from a buyer campaign. A knowledgeable peer or potential referral partner may be worth knowing. Mislabeling them as a hot prospect creates the wrong next step for both of you. Harry's discussion of participating in other creators' conversations is useful here: a contribution should earn interest through what it adds, rather than serve as a pretext to pitch everyone present.

Keep the evidence beside the name

Your research record should survive the handoff to whoever writes or answers the message. A useful first version contains:

• Current person, role and company, with the source checked.
• The inclusion criteria they meet and any exclusion you tested.
• The public observation, its URL and date.
• A possible business consequence, clearly marked as an inference.
• What you still need to ask.
• The owner, next action and any existing relationship or refusal.

Do not write “hot lead” where the evidence says “right role, possible need.” Those labels lead to different behavior. One invites a relevant exploratory question; the other can encourage an unjustified sales push.

Harry's Claude outreach workflow uses qualification before research and asks for three recent observations before drafting. You can apply that sequence manually. If AI helps, require the source for each observation and allow “could not verify.” Three weak facts are less useful than one meaningful one.

Stop researching when you can make a proportionate next move

You do not need a dossier to ask a short, relevant question. Excess research can become an excuse to avoid testing the offer.

You do need enough evidence to explain why this person, why this topic and why your proposed next step fits. Review a manageable initial batch yourself. If many people fail the same test, change the list rule or source before adding more names.

A filtered database or ordinary LinkedIn search may be the efficient starting point when the market is well defined. Public conversations can then supply context. Sales Navigator, a licensed data source and manual research can work together; this is not a requirement to abandon tools or buy another one.

Track what happens after the first reply. Did the person recognize the problem? Were they the right owner? Could your service help? Did the conversation reach a sensible next step? Negative answers should change your research criteria, not merely the wording of your opener.

Once those answers become more consistent, expand carefully. If they remain unclear, a larger list will give you more of the same uncertainty to manage.