The short answer
A Boolean string is not good or bad in isolation. It selects a population. Title-heavy strings select title vocabulary. Skill-heavy strings select self-described capability. Evidence lanes select public work. Adjacent-title strings select transferable backgrounds. Donor-company strings select employer context. Run multiple archetypes and measure what each adds instead of endlessly polishing one string.
Definition: query archetype
A query archetype is a structural family of search string defined by the evidence type it selects rather than the exact keywords it contains. Two queries belong to the same archetype when they rely on the same underlying signal—for example titles, tools, public artifacts, adjacent roles, or employer context. The concept is useful because a more precise version of the same archetype can still be structurally blind to the populations another archetype reaches.
Why “better Boolean” is the wrong benchmark
Precision matters, but precision is not coverage. A highly refined title query cannot return a person whose employer used a different title. A narrow skill query cannot return a strong practitioner who barely maintains a public profile. The benchmark therefore measures result composition and unique contribution, not which string looks the cleverest.
The five archetypes
The examples below use a senior middleware or integration engineer working with IBM MQ for consistency. Adapt the terms; preserve the structural shape.
1. Title-heavy
Selects: Employer title vocabulary
("Middleware Engineer" OR "Integration Engineer" OR "MQ Engineer" OR "Messaging Engineer" OR "Middleware Administrator") AND ("IBM MQ" OR "WebSphere MQ" OR "MQSeries")What it can find: People whose profile uses titles close to the requisition.
Structural blind spot: People doing equivalent work under broader or company-specific titles.
2. Skill-heavy
Selects: Stated tooling and capability
("IBM MQ" OR "WebSphere MQ" OR "MQSeries") AND ("MQSC" OR "queue manager" OR "channel" OR "AMQP" OR "JMS") AND ("clustering" OR "high availability" OR "failover" OR "TLS")What it can find: Practitioners who document the technical details of how they work.
Structural blind spot: Strong practitioners with sparse or outdated public profiles.
3. Evidence-based
Selects: Public work artifacts
site:github.com ("MQSC" OR "queue manager" OR "ibm-mq") ("Terraform" OR "Ansible" OR "Helm")What it can find: People whose public work reveals relevant evidence even when conventional profile text is thin.
Structural blind spot: People whose work is closed or non-public, which is especially important in cleared environments.
4. Adjacent-title
Selects: Neighboring roles and transferable capability
("Systems Engineer" OR "Application Support Engineer" OR "Platform Engineer" OR "ESB Developer" OR "Integration Architect") AND ("Kafka" OR "RabbitMQ" OR "ActiveMQ" OR "TIBCO" OR "MQ") AND ("enterprise messaging" OR "message queue" OR "middleware")What it can find: People one step away from the exact title who may have transferable capability.
Structural blind spot: It is intentionally broader, so false-positive review cost rises and must be measured.
5. Donor-company
Selects: Employer context supported by market evidence
(validated donor-company set) AND ("middleware" OR "integration" OR "messaging" OR "MQ")What it can find: People whose work environment is relevant even when their vocabulary differs from the req.
Structural blind spot: People at employers your donor map did not include, which is why provenance and refresh dates matter.
Pre-registered benchmark protocol
- Choose three requisitions across three role families.
- Build all five archetypes for each requisition.
- Run them in the same licensed search environment, with the same filters and a 30-minute cap per archetype.
- Record total results where available, reviewed results, saved profile-like leads, and saved leads not found by earlier archetypes.
- Rotate archetype order between requisitions so first-run advantage does not automatically inflate the same lane.
- Dedupe only on a stable identity anchor and require human confirmation before merging.
- Report results by role family with sample size and collection window; do not blend unlike markets into a single headline number.
Results status: not yet collected. This page will be updated when the controlled benchmark is complete.
Three failures that ruin Boolean coverage
ANDing synonyms
If two terms are genuine alternatives for the same signal, putting them in separate AND blocks requires both and can collapse the result set. Group alternatives deliberately.
Overusing exclusions
Every NOT or minus exclusion trades recall for cleanliness. Exclusions should target known noise patterns, not become a substitute for reviewing imperfect results.
Calling one string a search strategy
One string is one lane. If it is title-heavy, polishing it does not magically turn it into a skill, evidence, adjacency, or donor-company lane.
Where BooleanOS fits
Use the Boolean Generator to draft narrow and expanded logic, then pair it with the Search Lane Expander so the search is organized by coverage rather than by one giant expression. The Search-Path Scarcity framework explains why that matters, and the Federal Contract Data lane shows how to build employer-context inputs from public evidence.
Primary-source context
Google Search Help documents operators including quoted exact matches, site:, and minus exclusions. GitHub Docs documents the syntax for GitHub code search. For occupation adjacency and alternate-title research, O*NET OnLine publishes current occupation tasks, related occupations, and reported job titles.
FAQ
Which Boolean archetype is best?
There is no universal winner. The five archetypes select different signal types. The useful question is which archetype still adds net-new qualified leads for the requisition you are working.
What is a query archetype?
A query archetype is a structural family of search string defined by the type of evidence it selects—such as titles, skills, work artifacts, adjacency, or employer context—rather than by its exact keywords.
Does Boolean still matter when recruiters use AI?
Yes. AI can help draft and expand strings, but recruiters still need to understand which signal type the query selects and which market segment it is structurally unable to reach.
Can recruiters use site: searches on Google?
Google documents site:, quoted exact matches, and minus exclusions as supported search operators. Search publicly indexed pages manually and follow the terms and policies of the underlying sites.
How long should each benchmark lane run?
The initial SourcingOS protocol caps each archetype at 30 minutes so effort is comparable. The final benchmark will report the cap and reviewed-result counts with the results.