The 5 Key Metrics for Technical Sourcing Efficiency
If you run a technical recruiting team and you're only tracking time-to-fill and cost-per-hire, you're flying blind. Those are lagging indicators. By the time they look bad, the damage is done.
The metrics that actually let you steer the ship are more granular, more actionable, and surprisingly rarely tracked. I'm going to walk through five of them with real benchmark numbers so you can see where your team stands.
1. Source-to-Screen Rate
What it measures: Of all candidates sourced (not applied - sourced), what percentage make it to a recruiter screen?
Why it matters: This is the single best indicator of sourcing quality. A low source-to-screen rate means your team is spending time finding people who don't respond, aren't interested, or aren't qualified. A high rate means your targeting is sharp.
Benchmarks:
- Below 15%: Your targeting criteria need work. You're casting too wide a net.
- 15-25%: Average. Room to improve but not broken.
- 25-40%: Strong. Your sourcers know who to go after.
- Above 40%: Exceptional. Usually seen with referral-heavy pipelines or very precise tooling.
How to improve it: Better data before outreach. If you know a candidate's current tech stack, salary range, job tenure, and recent activity before you reach out, you can filter out the mismatches before wasting anyone's time.
2. Response Rate by Channel
What it measures: For each outreach channel (LinkedIn InMail, email, Twitter/X DMs, conference introductions, referrals), what percentage of messages get a response?
Why it matters: Not all channels are equal, and the differences are dramatic. Most recruiting teams over-invest in their lowest-performing channel because it's the most familiar.
Benchmarks:
- LinkedIn InMail: 10-18% response rate (down from 25% five years ago)
- Personalized email: 20-35% response rate
- Referral introductions: 45-65% response rate
- Community/conference connections: 30-50% response rate
- Twitter/X DMs: 8-15% response rate
The trend is clear: channels with built-in trust (referrals, community) dramatically outperform cold channels. If your team is spending 80% of its effort on InMails, the math says you should rebalance.
3. Time-to-Fill by Role Type
What it measures: Average calendar days from opening a req to accepted offer, broken down by role category.
Why it matters: "Our average time-to-fill is 45 days" is meaningless without segmentation. Frontend roles fill differently than infrastructure roles. Niche language roles (Rust, Go) take longer than generalist roles. If you blend them all together, you can't identify where the bottleneck is.
Benchmarks:
- Frontend (React/Vue/Angular): 30-42 days
- Backend (Python/Java/Node): 35-48 days
- Infrastructure/DevOps: 40-55 days
- Data/ML Engineering: 45-60 days
- Niche languages (Rust/Go/Zig): 55-80 days
- Engineering Management: 50-70 days
If your infra roles are taking 80 days while your frontend roles close in 35, that's not a general "hiring is slow" problem. That's a specific sourcing gap you can address.
4. Cost-per-Qualified-Candidate
What it measures: Total sourcing cost divided by the number of candidates who pass the initial technical screen. Not cost-per-hire. Not cost-per-applicant. Cost per candidate who is actually qualified.
Why it matters: Cost-per-hire hides a multitude of sins. If you hire one person after screening 200 unqualified candidates, your cost-per-hire looks fine but your efficiency is terrible. Cost-per-qualified-candidate tells you how efficiently your sourcing dollars convert to real pipeline.
How to calculate it: (Total sourcing tool costs + sourcer salary allocation + agency fees) / number of candidates who pass technical screen in the same period.
Benchmarks:
- High efficiency: $200-$500 per qualified candidate
- Average: $500-$1,200 per qualified candidate
- Needs attention: $1,200-$2,500 per qualified candidate
- Burning money: $2,500+ per qualified candidate
If you're paying $2,000 per qualified candidate through agencies and $400 through direct sourcing tools, the ROI argument for investing in better tooling writes itself.
5. Pipeline Diversity
What it measures: The demographic and background diversity of your sourced candidate pool at the top of the funnel.
Why it matters: You can't hire diversely if you don't source diversely. Most diversity problems in engineering hiring are pipeline problems, not evaluation problems. If 90% of your sourced candidates come from the same 10 companies and the same 3 universities, your hires will reflect that.
What to track:
- Gender distribution of sourced candidates vs. industry baseline
- University diversity (not just top-10 CS programs)
- Company background variety (big tech, startups, non-tech industries, self-taught)
- Geographic distribution
Benchmark: There's no single "good" number here because it depends on your goals. But if your sourced pipeline is less diverse than the overall industry talent pool, your sourcing strategy has a systematic gap.
Putting It Together
These five metrics give you a complete picture of sourcing health. Source-to-screen tells you if you're targeting right. Response rate by channel tells you where to invest effort. Time-to-fill by role type tells you where the bottlenecks are. Cost-per-qualified-candidate tells you if the economics work. Pipeline diversity tells you if you're reaching broadly enough.
Candyfloss AI gives you the data layer to improve every one of these metrics. Better targeting data improves source-to-screen. Personalization data improves response rates. Niche filters reduce time-to-fill for hard roles. Efficient tooling reduces cost-per-qualified-candidate. And broad, multi-source indexing helps you reach candidates you'd never find on a single platform.