Candidate Sourcing Software: How to Find People Who Are Not Applying

The numbers behind this

What does candidate sourcing software do that an applicant tracking system does not?

Sourcing software finds people who are not applying to you. Everything else in your stack is built to move people who already did. That single distinction decides the tooling: a tracking system is designed to advance known applicants through stages, while a sourcing tool is designed to discover names, find a working contact detail, and start a conversation from cold. Most teams need both, and the seam between them is exactly where sourcing programmes fail. A recruiter finds forty strong profiles, messages them out of a spreadsheet, and none of it ever reaches the pipeline the hiring manager reviews on Monday. Good sourcing tools write back into the same record system, so a sourced contact who replies becomes a tracked candidate without anyone retyping a name. If you are still deciding what belongs in the core system underneath, the features an applicant tracking system needs is the better place to start.

Where do sourced candidates actually come from?

There is no single database where the people you want sit waiting. Sourced hires come from a handful of channels and each one covers a different slice of the workforce. Professional networks are dense for corporate and technical roles with recognisable titles, and close to empty for shift work, trades and field sales. Public portfolios only reach practitioners who publish, which in engineering is a minority. Community groups reach specialists but punish anything that reads like a broadcast. Your own historical applicants are the one channel where people already know your name. The practical move is to pick two channels per role family, get genuinely good at them, and stop paying for the rest. Coverage beats volume: a channel that reaches sixty per cent of a population and gets read is worth more than one that lists everybody and gets ignored. The wider tooling map shows where each of these sits.

Should you learn boolean search or trust semantic matching?

Learn boolean first, then let semantic search widen it. Boolean is exact. You name the strings, the operators and the exclusions, so you always know why a profile appeared. That precision is what makes a search repeatable, auditable and easy to hand to a colleague, and it is why boolean still runs most serious sourcing desks. Semantic search reads intent instead of keywords. It surfaces the payments engineer who never wrote the word fintech anywhere on their profile, but it hides its reasoning, so a bad result is hard to diagnose. Use them in sequence. Write a tight boolean string for the non-negotiables, then run a semantic pass to catch people who describe identical work in different vocabulary. If you want a starting point for a role, the free boolean search string generator builds one from titles, skills and exclusions you can edit.

What makes sourcing outreach get replies?

Relevance and brevity, long before follow-up count. Cold messages that work name the specific work the person has done, say in one line why that maps to the role, and ask a small question somebody can answer without booking anything. Long pitches with salary bands, a benefits list and a calendar link in the first touch read as bulk mail and get treated as bulk mail. Sequences help, but the return flattens fast. A second and third touch spaced several days apart recover a real share of non-responders. A fourth mostly irritates people. Change channel rather than repeating yourself in the same inbox. Track replies by sequence and by role instead of as one blended number, because a template that hums for QA engineers will die in a finance search. And stop a sequence the second somebody answers, a rule that applies to everything you automate in hiring.

Why is your own database the cheapest source of hire?

Because you already paid to acquire those people. A team three years into hiring is sitting on thousands of applicants who were qualified but lost to somebody marginally stronger, plus everyone screened for a role that got frozen. Rediscovery means searching that history before you open a new search, and it produces the shortest time to first conversation of any channel: no licence fee, no cold introduction, no paragraph explaining who you are. The reason it goes unused is that most databases cannot be searched. Notes are free text, rejections have no reason code, and nothing records which skills were verified by an actual interview. Fix the input and the channel appears. Tag rejection reasons, keep scorecards attached to the person, and re-run last year's searches first. See how Pitch N Hire keeps a saved talent pool of past candidates if you want the mechanics.

How do you source for diversity without crossing a legal line?

Widen the search, never filter on the person. The lawful and effective version of diversity sourcing changes where you look and how you write, not who is allowed into the pipeline. In practice that means sourcing from schools, employers and communities outside your usual list, dropping requirements that quietly screen out career changers, and writing outreach that does not assume an unbroken career. Selecting or excluding on a protected characteristic is unlawful in most jurisdictions whatever the intent, and any tool offering to filter by inferred ethnicity or gender is a liability rather than a feature. Measure the top of the funnel instead. If a role family keeps producing the same narrow mix, the search string and the channel are the cause, not the market. Aggregate reporting, collected voluntarily and kept away from hiring decisions, is the honest way to see whether the widening worked.

How do you know whether sourcing is working?

Three numbers cover most of it: reply rate, sourced-to-hire ratio, and days from first message to first interview. Reply rate exposes targeting and message quality and is the fastest thing to fix. Sourced-to-hire tells you what a hire really costs in contacts, and for a senior search that number is often well into three figures, which is why sourcing capacity belongs in a plan rather than in somebody's evenings. Days to first interview shows whether replies are converting or dying in an inbox while a recruiter is stuck in interviews. Attribute every sourced contact to its channel at the moment you save it, because source of hire becomes fiction within a quarter otherwise. Then report sourcing in the same place as everything else rather than in a private spreadsheet. The recruiting metrics worth tracking covers the funnel side of that picture.

What is source of hire, and why does it stop being accurate?

Source of hire records where a hired person originally came from, and it decays faster than any other recruiting number. Three things break it. Most candidates touch several channels before applying, so a single field forces a choice between first touch and last touch, and teams rarely agree which one they are recording. Self-reported answers on the application form describe how somebody remembers finding you rather than what happened. And manual tagging stops within weeks of anyone ceasing to check it. The fix is unglamorous: tag the channel automatically at the moment a record is created, keep the referrer separate from the application source instead of overwriting one with the other, and accept that referrals need their own field because they are a relationship rather than a channel. Then report it beside interviews and hires rather than applications, which is where recruitment analytics makes the comparison worth having.

Do you need a dedicated sourcer, or should recruiters source?

Split the roles when sourcing keeps losing to interviews, which is the point at which it stops being a preference. A full-cycle recruiter owns everything from intake to offer, which suits roles that fill from applications and teams too small to specialise. A dedicated sourcer builds and works the top of the funnel, which suits scarce skills, senior searches and any market where the people you want are not applying anywhere. The split creates one new risk worth planning for: the handoff. A sourced candidate arriving as a name and a link converts far worse than one arriving with the context of the conversation that got them interested, so agree what transfers, keep it on the record rather than in chat, and let the sourcer stay visible until the first interview. Measure the two roles on different numbers, since a sourcer judged on hires is being judged on somebody else's process. The recruiter workflow has to hold both.

Sourcing channels compared: what each one is good for and what limits it