Find and preview leads

VibeSell searches a B2B lead provider — people at companies — with the filters of your target. Counting and previewing are free: nobody is imported and no lead credit is spent until you approve an import.

The loop

Describe the target — save_target keeps the segment, the company profile, the champion who answers, the trigger that makes them answer now, and who to exclude.
Turn it into filters — save_target_filters keeps the lead filters every search starts from: countries, places, job titles, seniority, headcount, industries, keywords…
Count — estimate_lead_pool returns how many people match. The number is measured, not estimated.
Look at real people — preview_lead_sample shows up to 10 of them: name, job title, company, location and LinkedIn. Never their email address.
Confirm, in your words — once you’ve seen them and said they fit, confirm_target records your sentence. It opens the plan step.

Writing good filters

The filters are described on each tool’s page and their values listed in the lead filter vocabulary. What matters most:

Countries or places, not both for the same area. countries takes whole countries as ISO codes (FR, DE, US). A city or region goes in places — { "city": "Lyon", "country": "FR" } — and its country stays out of countries. excludedPlaces cuts a place out of a wider one: France except Paris is countries: ["FR"] with excludedPlaces: [{ "city": "Paris", "country": "FR" }].
Whose address? locationScope is CONTACT (where the person is — what “prospects in Lyon” means) or COMPANY (where the company is registered).
Job titles match by substring. jobTitles: ["Head of Sales"] matches every title that contains it, and several titles read as OR. Write them the way people of that market write them — Chirurgien-dentiste for French dentists.
Headcount: numbers or brackets, never both. Use headcountMin / headcountMax, or headcountBands. Given together, the brackets are refused and the numbers kept.
Sub-industries instead of industries. subIndustries (for example Medical Practice) is for a niche; sending it with includedIndustries widens the search instead of narrowing it.
Keywords describe the company, never the person: ["orthodontie"] finds orthodontic practices. Give each word its own item, in every language those companies use.
Silent zeros. A technology, a website or a look-alike the provider doesn’t know comes back as zero leads without an error. That zero is not an empty market.
Funding filters drop bootstrapped companies: a company that never raised matches no stage.

A filter left out of a count or preview is taken from the target saved in the workspace; the result lists the filters actually searched (filters), the saved ones left out (notSearched) and any value it did not recognise (ignored).

Reading a count

{
"kind": "lead_count",
"status": "counted",
"total": 4812,
"atLeast": false,
"filters": [{ "label": "Job title", "values": ["Chirurgien-dentiste"] }],
"notSearched": [],
"ignored": [],
"excludedCompanies": 0,
"measuredAt": "2026-10-23T09:12:44.000Z",
"message": null
}

status is counted, needs_filters (no filter at all), unsearchable or failed; message says why when nothing was counted.
When atLeast is true, the counter hit its ceiling: total is a floor.

Importing

import_leads takes the same filters plus a batch name (cohortName), a maximum (maxLeads) and a one-sentence summary. It does not import: it opens an approval card.

It costs one lead credit per lead asked for in maxLeads, whatever comes back, and it cannot be undone.
It needs an active plan — lead credits come with it.
The imported people come with their email addresses, in a named batch you find again in VibeSell.

See Approval cards for what happens on your click.