Candidate Sourcing

Candidate Sourcing

Build a qualified candidate pipeline through multi-channel sourcing strategies.

Workflow

1. Parse the job description to extract: required skills, experience level, location preferences, deal-breakers, and "nice-to-have" differentiators.
2. Build a target candidate profile: ideal company backgrounds, tech stack familiarity, career trajectory patterns, and signals of high performers in this role.
3. Execute parallel searches across channels (see channel matrix below).
4. Score each candidate against the profile using a fit-score model (0–100).
5. Deduplicate across channels — merge records for the same person.
6. Segment candidates into tiers: Hot (ready to engage), Warm (nurture), Cold (long-term pipeline).
7. Draft personalized outreach messages for Hot and Warm tiers.
8. Track response rates per channel and message variant; optimize iteratively.

Channel Strategy Matrix

| Channel | Best For | Search Approach | Response Rate | |---------|----------|-----------------|---------------| | LinkedIn Recruiter | Passive professionals | Boolean + filters (title, company, tenure, skills) | 15–25% | | GitHub | Engineers, DevOps | Language, repo activity, org affiliation, commit frequency | 10–20% | | Stack Overflow | Senior engineers, specialists | Reputation score, tag activity, answer quality | 8–15% | | Indeed / Glassdoor | Active job seekers | Keyword search, company page applicants | 30–40% | | Employee Referrals | Cultural fit, speed | Internal campaign, incentive tracking | 40–60% | | Industry Communities | Niche talent | Slack/Discord groups, conferences, meetups | 20–35% | | Internal Talent Pool | Redeployment | Skills inventory, performance data, career aspirations | 25–40% | | University Pipelines | Juniors, interns | Career fairs, CS club partnerships, hackathons | 15–25% |

Boolean Search Builder

Construct Boolean strings for LinkedIn/Indeed/GitHub:

// Senior Backend Engineer example
("Senior" OR "Staff" OR "Lead")
AND ("Software Engineer" OR "Backend Engineer" OR "Server-Side")
AND ("Python" OR "Go" OR "Rust" OR "Java")
AND ("AWS" OR "GCP" OR "Azure")
AND ("distributed systems" OR "microservices" OR "scalability")
NOT ("intern" OR "junior" OR "entry-level" OR "bootcamp")

Rules:

• Use quotes for exact phrases, parentheses for grouping
• Limit to 3–5 AND clauses (too many = zero results)
• Add NOT clauses sparingly — they can exclude edge-case great candidates
• Test each Boolean string and iterate based on result count

Candidate Scoring Model

Score each candidate 0–100 across five dimensions:

| Dimension | Weight | Scoring (0–10 each) | |-----------|--------|---------------------| | Skills Match | 30% | How many required skills are demonstrably present? | | Experience Level | 25% | Does tenure and scope match the role level? | | Company Pedigree | 15% | Prior companies relevant to our industry/domain? | | Activity Signals | 15% | Recent job changes? Active on platforms? Open to opportunities? | | Referral Boost | 15% | Referred by current employee? (+10 auto) |

Composite Score = Σ(dimension_score × weight)

Tier assignment:

• ≥ 75: Hot — outreach within 48 hours
• 50–74: Warm — add to nurture sequence
• < 50: Cold — hold in talent pool, review quarterly

Outreach Message Templates

Cold Outreach (LinkedIn InMail)

Hi [Name],

I came across your profile while sourcing for a [Role] position at [Company].
Your experience with [specific skill/project from their profile] stood out.

We're building [1-sentence about what the team is doing].
The role would involve [key impact statement].

I know you may not be actively looking, but I'd love to share more
if you're open to a casual 15-minute chat. No pressure either way.

Best,
[Recruiter Name]

Warm Outreach (Referral Introduction)

Hi [Name],

[Employee] suggested I reach out — they mentioned your work on
[specific project] and thought you'd be a great fit for a role
we're building at [Company].

[2-sentence pitch about the role and impact]

Would you be open to a quick call this week?

Best,
[Recruiter Name]

Follow-Up (No Response After 5 Days)

Hi [Name], just circling back on my note below.
I completely understand if the timing isn't right —
I'll hold your info in our talent pool and reach out
when something more aligned comes up.

Best,
[Recruiter Name]

Outreach Rules

• Personalize the opener: Reference something specific from their profile (project, post, contribution)
• Lead with value, not ask: What's in it for them? (impact, growth, compensation range)
• Include a soft CTA: "Open to a chat?" not "Apply now"
• Timing: Tuesday–Thursday, 10am–12pm local time shows highest response rates
• Frequency: Initial + 1 follow-up at 5 days + 1 follow-up at 14 days, then drop
• Transparency: Include compensation range in the second message if they ask

Deduplication Rules

When the same person appears across multiple channels:

1. Match by: email address (primary key), then name + company combo, then LinkedIn URL
2. Merge all data into a single candidate record
3. Retain the highest engagement channel as primary contact method
4. Log all channel appearances for source-attribution reporting

Pipeline Metrics to Track

| Metric | Target | How to Calculate | |--------|--------|-----------------| | Source-to-Applied Rate | > 30% | Applicants ÷ Outreach sent | | Applied-to-Screened Rate | > 20% | Screens ÷ Applicants | | Time-to-First-Contact | < 48h | Hours from profile ID to outreach | | Channel ROI | Varies | Cost per hire per channel | | Pipeline Depth | ≥ 5× headcount | Qualified candidates ÷ open roles |

Edge Cases

• Hard-to-fill roles (ML engineers, security specialists): Expand search to adjacent skill sets, consider bootcamp grads with strong portfolios, tap alumni networks
• Mass hiring (50+ roles): Shift from individual outreach to employer branding campaigns, event-based sourcing, and agency partnerships
• Geographic constraints: Use remote-work filters, consider EOR (Employer of Record) options for international talent
• Diversity goals: Use blind-sourcing tools (seek out candidates from underrepresented groups without bias toward pedigree)