Skip to main content
  • Sales & Growth
  • Running
  • Client work, anonymised (a consulting firm and its clients)

A lead engine that finds real decision-makers for a few cents each

Describe the ideal customer in keywords, roles and geography; the engine turns that into precise public-web searches, keeps only the people whose title fits, finds an email before spending anything, and remembers who it already gave you.

running for one client, whose own interface stays withheld. The screens below are our Open Research Engine: the same method, our product, on invented data.

Talk to us about a build like thisAll 7 projects

A look inside

These are not the client's screens and never will be. They are our own Open Research Engine, the product that runs this method for us, on nine invented freight companies and the invented people at them: a lead engine's real output is real people's names and addresses. Everything the engine does here is the shipped engine doing it. Only what it was fed is made up.

The Open Research Engine's workspace: an ICP panel on the left with a preset loaded, a results grid of twenty-three people with name, title, company, LinkedIn and email, tier pills reading enriched or candidate, and a footer counting people, LinkedIn coverage, fit and enrichment targets.

One screen, no page hops: the profile on the left, the people on the right, and a footer that keeps score. The tier pill on every row is the engine's verdict on how reachable that person actually is.

1 of 4

The work

  1. Stage: points

    The problem

    Buying lists meant paying for names that did not fit and paying again for names already held. The firm needed a source-agnostic way to build lists for a narrow, unusual market where the buyers do not sit in any vendor's database.

  2. Stage: connections

    What we built

    A discovery engine and a small web app on top of it. The engine writes search-operator queries over the public web and directories, runs them, and passes every candidate through a hard fit gate decided on the person's title alone, with a deny list, a leadership override and generic tokens that can never rescue an off-target title. Kept contacts are tiered by reachability: send-ready, enrichable, name only. A free-first email waterfall runs before any paid lookup; a dedupe master keeps every run returning new people. The app shows each run live, stage by stage, with the cost per lead.

  3. Stage: resolved

    The result

    Runs on the researcher's own machine on demand. Over 255 runs it produced 350 contacts, 124 of them send-ready, at an average of about one and a half cents each.

Evidence at a glance

  • hard ICP-fit gate on the title, every tier
  • free-first email waterfall before any spend
  • no contact returned twice across runs
  • about a cent and a half per lead, blended

Want one of these pointed at your business?

Talk to us about a project like this