Cloning your best client: lookalike prospecting

Lookalike prospecting means profiling your best existing clients — their sub-vertical, size and situation — and building a list of companies that share those traits. It is the fastest route to a high-fit outbound list, because it replaces guesswork about who might buy with evidence about who already has. The pattern is in your closed-won record; the work is extracting it and making it searchable.

Why start from existing clients?

Because your client base is the only targeting dataset that contains the answer. Every other source — industry reports, gut feel, whoever replied to the last campaign — tells you who might fit. Your closed-won deals tell you who signed, stayed and paid. When firms skip this step, they typically default to targeting companies that look like the clients they want rather than the clients they win, and the list underperforms from day one. Good copy cannot save bad targeting, and this is where targeting starts. The full route from profile to verified list is laid out in the B2B database building guide; lookalike work is the first mile of it.

What does "best" actually mean?

Not biggest. The clients worth cloning score well across four dimensions:

• Margin. Revenue after the true cost of serving them, not the headline fee.
• Retention. They renewed, expanded or came back.
• Sales velocity. They closed quickly, with few stakeholders and little discounting.
• Delivery friction. They were straightforward to serve — few escalations, sensible expectations.

A large logo that closed slowly, haggled hard and consumes your delivery team is not a template. A mid-sized firm that signed in three weeks and renewed without a conversation is. If you have never scored clients this way, five minutes with a spreadsheet is usually revealing.

How do you extract the pattern?

The mechanism runs in four steps. When you pull your client list, then rank it on the four dimensions above and take the top five to ten. When you have the shortlist, then record the observable traits of each: sub-vertical, headcount band, geography, ownership structure, who signed, and what was happening when they bought — new hire, growth spurt, a failed supplier. When a trait appears in most of the shortlist, then it becomes a list-building filter; when it appears once, then it is an anecdote and gets discarded. When the filters are set, then write them down as a one-page profile — this is your ICP, stated as searchable criteria rather than adjectives. Defining the ICP precisely is the difference between a filter a data source can execute and a wish it cannot.

The discipline is "observable". "Ambitious founders who value quality" is not a filter. "UK recruitment agencies, 10–30 staff, two or more consultants hired in the last six months" is.

How does the pattern become a list?

Each observable trait maps to a source. Sub-vertical and headcount map to company databases and the public register; roles map to LinkedIn; situational traits map to job boards and filings. From there the standard pipeline applies — source, identify the decision-maker, enrich, verify, load. Two companion pieces cover the adjacent decisions: where list data stops and CRM data begins, so your cloned list never pollutes your system of record, and intent signals, which tell you when to contact the companies the lookalike work tells you to contact at all. Who, then when — in that order.

Where does lookalike prospecting go wrong?

Three failure modes recur. Over-fitting: with only five to ten reference clients, it is easy to treat coincidence as pattern — three clients in Manchester does not make geography a filter unless you can say why it would be. Cloning the past: if your strategy has moved upmarket or into a new service line, yesterday's best clients describe yesterday's business; adjust the profile deliberately rather than inheriting it. Unsearchable traits: qualities you cannot filter for belong in the messaging, not the list criteria.

Handled honestly, though, this is the highest-leverage hour in outbound. It is also, quietly, most of what a good BDR does in their first month — at £35k+ a year — before they start sending anything. Making the pattern explicit means the system can do the finding, and the humans can keep the judgement.

---

Next step: the Growth System Audit — £450, seven days, credited against any build — maps where your growth system leaks and what to build first.