The short answer
To automate candidate sourcing, write down what a qualified candidate looks like, let software search your own ATS before anywhere else, add outside sources, score every match against the same criteria, and send personalized outreach sequences that follow up on their own. A recruiter approves the criteria and a sample of matches before anything goes out, then reviews interested candidates instead of raw search results. Done well, sourcing becomes a process that runs every night rather than a task a recruiter squeezes in between calls. The steps below work whatever tool you use.
What automated sourcing can and can't do
Automation is very good at the mechanical parts of sourcing: running searches, de-duplicating profiles, checking who has already been contacted, writing a first draft of each message, and following up on schedule. It is also good at rediscovery, finding people already in your ATS who fit a new role.
It is weaker at things that need inside knowledge: knowing that a target company just had layoffs, that a candidate's current manager is a friend of your CTO, or that "five years of Kubernetes" is a requirement the hiring manager will drop once asked. Plan for a recruiter to stay involved at the start of each search and at the hand-off.
Step 1: Write a sourcing brief the software can use
Most failed sourcing automation starts with a vague brief. Before you switch anything on, write these down for the role and get the hiring manager to sign off:
- Three to five must-haves that can be checked from a profile or a short conversation (for example "has owned a quota for a mid-market SaaS product").
- Nice-to-haves, clearly labeled so the software doesn't treat them as filters.
- Exclusions: current employees, people rejected in the last six months, companies you have agreements not to recruit from.
- Location and work model: where the person must be, or which time zones work.
- The pitch: two or three true reasons someone would take this job.
The same brief becomes the scoring rubric later, which keeps sourced, referred and inbound candidates on equal terms. Our interview scorecard template is a good format.
Step 2: Search your own ATS first
Your ATS is full of people who already applied, interviewed, or reached the final round and lost out narrowly. They know your company and often reply at higher rates than cold prospects. This is called talent rediscovery, and it is the cheapest sourcing you have.
Automate it by letting your sourcing tool read past candidates and their outcomes, then match them to each new role. Exclude anyone with a recent rejection for the same role, and check the notes from earlier interviews before outreach. A former finalist should hear something like "we remember you," not a generic cold message.
Step 3: Choose your outside sources
For new names, sourcing tools work in one of two ways. Some search a large profile database they maintain or license. hireEZ, for example, says it searches more than a billion profiles across 45+ platforms, Pin says 850 million+, and Juicebox says 800 million+ across 30+ data sources. Others search the open web, job boards and professional networks you have access to. Database size matters less than coverage of your specific market, so test with a role you know well and see how many of the people you'd expect actually show up.
Also plan for inbound. Posting to job boards such as Indeed and ZipRecruiter brings applicants to you, and the same scoring step should handle them.
Step 4: Build the search, then calibrate it
Older sourcing relies on Boolean search strings like ("customer success" OR "account management") AND SaaS NOT intern. Newer tools let you describe the role in plain English and generate the search themselves. Either way, calibrate before scaling:
- Run the search and review the first 20 to 30 results by hand.
- Mark each one as a yes, maybe or no, and write down why.
- Adjust the criteria. Common fixes: a title filter that is too narrow, an industry filter that excludes adjacent industries, or a years-of-experience floor that shuts out strong people.
- Share a sample with the hiring manager. If they reject half of your yeses, go back to Step 1.
Calibration usually takes an hour or two per role family, and it saves weeks of outreach to the wrong people.
Step 5: Score and rank every match
Once the search is calibrated, have the software score each match against the brief and explain the score in a sentence a recruiter can check. "Strong match: six years in mid-market SaaS customer success, owned renewals" is checkable. "92% match" with no reason is not.
Scoring is also where fairness risk enters. A ranking used to decide who gets contacted or advanced is a selection step. Under the US Uniform Guidelines, the four-fifths rule compares each group's selection rate with the highest group's rate, and a ratio below about 80% generally indicates adverse impact. Track who your sourcing selects, not only who gets hired. In New York City, a tool that substantially assists hiring decisions may need an annual bias audit and candidate notice. This is not legal advice. Our AI resume screening guide goes further.
Step 6: Automate outreach sequences
An outreach sequence is a planned series of messages that stops when someone replies. A solid default for a professional role:
Rules that keep sequences effective and respectful:
- Personalize the first line from something real in the profile. Generic flattery hurts reply rates.
- Only text or message people who have given permission, and honor opt-outs right away.
- Stop the sequence the moment someone replies, on any channel.
- Answer common questions (salary range, remote policy, visa sponsorship) from an approved list, and pass anything unusual to a person.
- Send from a real recruiter's name and inbox, and be honest that automation helps you reach people.
Step 7: Hand off cleanly to screening and scheduling
Sourcing automation that stops at "interested" leaves recruiters chasing calendars. Connect the next steps: an interested candidate should get a structured first interview or a booking link straight away, and their profile, messages and score should land in your ATS so the hiring team sees one record. Our guide on adding AI to your ATS covers how the write-back usually works, and interview scheduling explains the scheduling side.
Step 8: Measure what matters
Review these weekly for the first month, then monthly. If reply rates fall, change the message before you change the search. If calibration falls, go back to the brief.
Picking a tool for automated sourcing
Two broad options exist. Dedicated sourcing tools are built around a large searchable database and deep filters, and they are strong when finding names is the hardest part. End-to-end agents treat sourcing as one step in a process that also covers outreach, screening, interviews and scheduling inside your ATS, and they are strong when the bottleneck is everything that happens after the search.
We build NTRVSTA, which is in the second group. Our agent Carolina sources for an open role, runs outreach by email, SMS and WhatsApp, scores and ranks every candidate with its reasoning, and hands interested people straight to a live AI interview and scheduling, with each action set to suggest, ask first or auto. Be aware of the limits: NTRVSTA doesn't publish a proprietary profile database, so if deep search over a huge database is your main need, SeekOut, hireEZ, Juicebox or Pin may serve you better. Compare them in our best AI sourcing tools list, or see the best AI headhunting tools for senior searches.
Questions
Can candidate sourcing be fully automated?
Most of the mechanics can: searching, de-duplicating, scoring, messaging and follow-ups. A person should still approve the brief and search criteria, review a sample of matches before outreach starts, and decide who advances. Fully hands-off sourcing tends to drift from what the hiring manager wants.
How many candidates should an automated sequence contact per role?
Start smaller than you think, around 30 to 50 well-matched people, and look at reply rates and hiring manager feedback before scaling. Contacting hundreds of loosely matched people damages your brand and wastes the hiring manager's time.
Is it legal to text candidates I sourced?
Rules on texting and messaging differ by country and state, and many require consent. Get permission before using SMS or WhatsApp, honor opt-outs right away, and check with counsel for the places you recruit. This is not legal advice.
Should I source passive candidates or focus on applicants?
Both, with the same scoring. Applicants are already interested, so screen them quickly. Passive candidates take more persuasion but are often the strongest people for hard-to-fill roles. Score both groups against the same brief so neither gets an unfair advantage.
Curious what sourcing looks like when it runs straight into interviews and scheduling? Book a demo and bring a live role, or try a sample interview first.
