Sign inGet your API key
ProductDocsPricingCompareBlogSign in

Glossary

What Is Predictive Lead Generation?

A way to rank prospects by how closely they match the customers you already won, and which of those inputs you can filter for in a database.

Get your free API key →Free to start. No credit card. 1,000 records to spend whenever you like.

Predictive lead generation is the use of a statistical model, trained on your past customers, to rank prospects by how likely they are to buy.

The model looks at who bought and who did not, then finds the attributes the buyers share. Those attributes are firmographic (industry, headcount, funding stage), technographic (tools on the company website) and role-based (job function, seniority). New prospects are scored against that pattern. This sits at the top of the lead generation funnel, before any message is sent.

Why it matters for outbound

Outbound has a fixed budget of sends, mailboxes and rep hours. A ranked list tells you where to spend them first.

Without a ranking, a list of 10,000 prospects is worked in the order of the export. With one, the top 500 get your best copy and your fastest follow-up. The method only works if you have enough closed deals to learn from. With a few dozen wins, a written ideal customer profile beats a model. Re-score every quarter, because the buyers you win change as your product does.

Example

You have 60 closed customers. Most are software companies with 50 to 200 staff that raised a round in the last two years and run a CRM your product connects to.

ProspectIndustryHeadcountLast roundCRM detectedFit
AcmeSoftware1202025YesHigh
GlobexRetail900NoneNoLow

Acme matches three of the four traits shared by your buyers, so it ranks first. Globex matches none. Names and values are placeholders. Check the result by comparing reply and meeting rates for the top and bottom of the ranking. If they match, the model adds nothing.

In LeadOcean data

LeadOcean has no predictive model and no score for how likely a company is to buy. It supplies the attributes a model or your own rules score on, and it does not hold intent or behavior data. The dataset is refreshed monthly.

  • minLastFundingYear and maxLastFundingYear limit companies by the year of their last round.
  • technologies and technologyCategories match tools detected on a company website.
  • employeeRange, jobLevel and jobFunction narrow by size and by buyer role.

Run these as filters on POST /v1/companies/search or POST /v1/people/search, or through the MCP tools search_companies and search_leads. Pull the scoring set with POST /v1/exports, score it in your own model, then work the top slice. The free plan gives 1,000 records, one-off, with no card. Pro is $499 a month. See /pricing.

For more on choosing between building a list and prospecting it, read lead generation or prospecting in 2026. More terms are in the glossary.

Pull the attributes your scoring model needs, free

Free to start. No credit card. 1,000 records to spend whenever you like.

Get your free API key →