Talent Sourcing: Finding People Before They Apply

Talent sourcing is the part of hiring that happens before anyone applies. You go and find the people who fit, instead of waiting to see who turns up. This page covers what sourcing actually involves, where candidates are found, and how to make it repeatable rather than a scramble every time a role opens.

Sourcing is not recruiting

The two words get used interchangeably, and that causes real confusion about what to measure. Recruiting is the whole process: the job description, the interviews, the offer, the negotiation. Sourcing is one step inside it — building a list of people worth contacting.

The distinction matters because they fail differently. A recruiting problem looks like candidates dropping out after the second interview. A sourcing problem looks like an empty pipeline: nobody good to interview in the first place. Fixing the second with better interview questions does nothing.

Why the best candidates never apply

Roughly nine in ten professionals are not actively looking for work at any given moment. They have a job, it is fine, and they are not reading job boards. That does not mean they would refuse a better offer — it means nobody has made them one.

This is the whole argument for sourcing. If you only consider people who applied, you are choosing from the fraction of the market that happened to be looking during the three weeks your posting was live. The strongest specialist in your city may simply never have seen it.

Where sourcing actually happens

Most sourcing tools index LinkedIn and stop. That covers stated experience — what someone wrote about themselves — and misses demonstrated skill almost entirely.

• GitHub and Stack Overflow. Code someone wrote and technical questions they answered. For engineering roles this is closer to the truth than any CV.
• Reddit and Discord. Where specialists discuss problems in their field without an audience of recruiters. Largely untouched by sourcing tools — see community sourcing.
• LinkedIn. Professional history and context. Useful, but as one source among several rather than the only one.
• Your own ATS. People who applied to a previous role and were good but not right for that job. The cheapest source you have, and the most neglected.

A process that survives contact with a real week

Sourcing collapses when it depends on someone having a quiet afternoon. A process that holds up looks roughly like this:

1. Define what the role genuinely requires. Not the wish list — the two or three things that decide whether someone can do the job.
2. Search where that evidence exists. For a backend engineer, public code. For a salesperson, track record and references.
3. Assemble one profile per person. Fragments across sources belong to the same human; keeping them separate is how you rate the same candidate twice.
4. Decide who to contact and why. A shortlist without reasoning is just a list. Write down what matches, what does not, and what you would ask at interview.
5. Reach out with something specific. Generic outreach gets generic response rates. Referencing someone's actual work does not.

How to tell whether sourcing is working

Volume is a misleading measure. A hundred sourced profiles that produce no interviews is worse than fifteen that produce four, because someone had to read all hundred.

The numbers worth tracking: how many sourced candidates reply, how many reach a first interview, and how long a role takes to fill compared with relying on applicants. If reply rates are low, the problem is the message. If replies are fine but interviews are not, the problem is who you are picking.

Where leadl.recruit fits

Paste a job description and it searches public professional activity across GitHub, Stack Overflow, LinkedIn, Reddit and Discord in parallel, assembles one profile per person from what it finds, and returns a shortlist where every candidate carries a written verdict: what matches the role, what does not, the risks it found, and what to ask at interview.

The decision stays with you. The reading — the part that eats an afternoon per role — does not. Scoring and how the verdict is produced are covered on the AI screening page.

Pricing is public: a free plan, then €49, €149 and €399 a month, billed monthly. You pay per search run, not per contact revealed, so a role costs what it costs before you start. See pricing.

It is not an applicant tracking system and does not replace one.

See also: talent acquisition software — the four categories and which you need.

Tool: boolean search string generator.