The Source Pack Methodology: A Search Operating System for Hard-to-Fill Roles

The short answer

A source pack is the search operating system for a requisition. It translates a vague hiring request into a testable market model: what evidence supports the role, which title families and donor companies matter, which independent source lanes will be tested, what queries belong to each lane, what noise to expect, and what evidence would justify changing the plan.

The important shift is from searching for people to testing a market hypothesis. A sourcer should be able to explain not just who they found, but which path found them, what that path missed, and what the next experiment should be.

Why difficult searches turn into random activity

Hard requisitions usually do not fail because a recruiter forgot one Boolean synonym. They fail because the search model is unstable. A title is treated as a skill. A preferred company becomes a hidden requirement. A public clearance mention becomes stronger than the evidence supports. A manager rejects three profiles for the same unstated reason, but the search never changes.

Without a source pack, those changes live in Slack, memory, browser tabs, and individual recruiter intuition. The team keeps doing work, but it cannot tell which assumptions are producing the wrong market.

A source pack creates a controlled surface for those assumptions. It makes the search debuggable.

The 9-part source pack

Build independent search lanes, not one giant string

A lane is a distinct route into the same talent market. Good lanes use different evidence, not just slightly different wording.

Title lane

Use when titles are stable enough to carry signal. Keep an exact-title lane separate from adjacent-title expansion so you can see what the expansion actually adds.

Skill and environment lane

Search for the technical or operational context that proves the work: Kubernetes plus Terraform, Epic plus a specific module, RMF plus ATO work, PyTorch plus model-serving context, or another role-specific combination.

Artifact lane

For roles with public work, search the work itself: repositories, model cards, papers, talks, standards contributions, package ecosystems, or technical writing. Artifact evidence should support investigation, not substitute for recruiter confirmation.

Donor-company lane

Search organizations that produce the same environment, customer, stack, mission, compliance model, or scale. A donor company narrows where to look; it does not prove that every employee there fits.

Owned-history lane

ATS rediscovery, prior finalists, silver medalists, referrals, and internal networks are independent lanes because they can add people that external search never surfaces.

Use the Search Lane Expander when the strict market stops producing new evidence-fit leads.

Use the source pack to change hiring-manager conversations

Weak calibration asks, “Do you like this profile?” Strong calibration asks which parameter should change after the market produces evidence.

• Evidence: What work proves the must-have?
• Title: Which adjacent title is acceptable if the evidence is strong?
• Domain: Does the person need the industry, or the operating environment?
• Donors: Which companies are genuinely comparable, and why?
• Location: Is the constraint legal, customer-driven, team-driven, or preference?
• Compensation: Is the target market compatible with the approved range?
• Verification: Which facts can be researched publicly, and which require an authorized later process?

For a ready-made sequence, use the 25 hiring-manager intake questions.

Measure the search plan, not recruiter busyness

A source pack becomes valuable when the team can compare lanes. Track raw lead volume, evidence-fit saves, duplicate pressure, unique contribution, rejection patterns, response outcomes, and recruiter time. Do not let “profiles viewed” become the main success metric.

If a lane keeps returning people already surfaced elsewhere, its marginal discovery value may be falling. Measure that with Unique Contribution Rate. If multiple independent lanes are producing high duplicate pressure and very few new leads, use the Search Exhaustion framework before declaring the market empty.

Worked example: cleared platform engineer

Suppose the requisition asks for a platform engineer supporting a federal environment. The weak approach is a single search for “cleared DevSecOps engineer.” The source-pack approach decomposes the market.

1. Outcome: operate and automate a secure container platform in a regulated environment.
2. Strong evidence: Kubernetes, Terraform, CI/CD, infrastructure automation, and federal security or authorization context.
3. Adjacent titles: SRE, cloud engineer, platform engineer, DevOps engineer, infrastructure engineer.
4. Donor groups: primes, systems integrators, cloud vendors, mission-tech companies, and subcontractors with comparable delivery environments.
5. Independent lanes: title, secure-delivery evidence, donor companies, public technical artifacts, and owned ATS history.
6. Boundary: public clearance language is a breadcrumb for manual follow-up, not confirmation of current status.
7. Stop rule: do not call the market exhausted until strict, adjacent, donor, and at least one independent evidence lane have been tested and the new-lead rate has materially flattened.

That pack can be reviewed with the hiring manager before the team spends another week editing the same Boolean string.

Where this connects to SourcingOS

The JD Strategy Tool creates the first source-pack draft. BooleanOS turns a understood role into query variants. Search Lane Expander opens adjacent paths. Candidate Search keeps source evidence visible, and the Candidate 360 sample shows what the evidence can become after recruiter review.

FAQ