Boolean search for recruiters: test your strings

Start with the role, then write the search

Boolean search helps recruiters combine requirements, group synonyms and exclude terms. Use a string to find documents worth reviewing for one role. It cannot establish that someone has the required experience or wants a new job.

For this exercise, we used Canonical's public Software Engineer - Python and K8s posting, observed October 20, 2026. Python and Kubernetes are sensible discovery terms. Software developer and site reliability engineer are title hypotheses to test alongside software engineer.

The posting also asks for Linux administration, public-cloud and container experience, reliable code, testing and documentation. It also specifies a degree or equivalent and travel once or twice a year for up to two weeks. The role is listed as Home based - Worldwide; US search parameters do not create a US residence requirement. Keep the full requirements in your review brief. This public example does not make Canonical a Metix customer or authorize recruiting on its behalf.

Download the blank and filled role brief (Markdown). For your own search, name the hiring owner and decide which requirements belong in the query, which have reliable platform filters and which need evidence later.

A Boolean cheat sheet with a defined scope

This example groups alternative titles and alternative Kubernetes spellings. AND connects the groups that each document must satisfy.

("software engineer" OR "software developer")
AND Python
AND (Kubernetes OR K8s)

Swipe horizontally to read the operator table.

Use straight quotes and uppercase logical operators. Keep a copy of the query before changing it. Do not assume wildcard, proximity, plus or minus syntax carries between tools.

LinkedIn Recruiter: choose the field as carefully as the words

LinkedIn's current Recruiter and Recruiter Lite help supports AND, OR, NOT, parentheses and quoted strings. It explicitly allows quotes and Boolean in Job titles, Companies and Keywords. Advice that Boolean only works in Keywords is too broad.

For a first Keywords template, widen the title group while keeping the two technology concepts:

("software engineer" OR "software developer" OR "site reliability engineer")
AND Python
AND (Kubernetes OR K8s)

In a paid seat, test title variants in the Job titles filter and technologies in Keywords rather than forcing everything into one field. Check Current, Past or both for the title scope. LinkedIn also provides Can have, Must have and Doesn't have choices in supported structured filters. Those UI choices are different from typing their labels into a query.

The Project filter is an exception: LinkedIn says it does not support Boolean search. Use its exclusion UI rather than assuming query syntax works there. LinkedIn says plus and minus are not officially supported; use AND and NOT in supported query fields. Its help also notes ignored stop words, including within quoted searches. The template above is based on documentation. We did not execute it in a paid Recruiter seat, and we cannot report its native result count.

Google X-ray search: three queries we actually ran

A Google X-ray search looks for public pages indexed by Google. It does not query LinkedIn's native database or establish complete coverage. Google documents quotes, site: and the minus exclusion operator. Put no space between site: and the domain.

G1: site:linkedin.com/in "Python" "Kubernetes"
G2: site:linkedin.com/in "Python" "Kubernetes" "software engineer"
G3: site:linkedin.com/in "Python" "Kubernetes" "software engineer" -intern -recruiter

We executed these on October 20 with English/US search parameters. The signed-in browser reported personalized results. We sampled the first five displayed links per query; this is a dated observation, not an estimate of national search demand or available candidates.

G1 included a test-framework result alongside software-engineering and DevOps results. Adding the title phrase in G2 changed the displayed sample. Adding exclusions in G3 changed it again. Inspect those changes before deciding that an exclusion helped. Ranking, indexing, personalization and text outside the snippet can affect what appears. We did not establish why any individual link moved or disappeared.

Do not paste Recruiter's NOT syntax into this example or assume a Google result snippet describes someone's current job. Open an appropriate public source and verify the relevant work before adding a person to an employer review.

Test an exclusion before trusting it

We built eight invented documents to make the cost of a broad exclusion visible. The local test checks literal words and phrases across each whole document. It is not a simulation of LinkedIn ranking, stemming or candidate quality.

A describes a software engineer who mentored an intern. E describes a software engineer who previously worked as a recruiter. L2 removes both, even though each contains the required discovery terms. B returns when we widen the title group. These are document matches; we have not judged whether any invented case meets the actual role.

G survives every version, but its text says Kubernetes was not used on the project. Matching a word does not establish a positive claim about experience. Read the sentence and the work behind it. C and D omit one of the discovery technologies, so their experience remains unknown. H is a recruiter hiring engineers with those skills; that is not evidence of doing the engineering work.

Download all eight test cases (CSV) and the query workbook (CSV). The workbook separates executed local tests, actual Google observations and the unexecuted Recruiter template. Start with the eight case rows, then add a known relevant document from a search you are allowed to run. In the query workbook, fill in your platform field and filters, record the version, and note whether the change kept that document. Add positive, missing and ambiguous cases before deciding whether an exclusion helps.

Debug the search one change at a time

1. Save the scope. Record the platform, field, current/past filters, date and exact query. A copied string alone is not enough to reproduce the search.
2. Read a small sample. Write down why each result is relevant, irrelevant or unknown. Keep a promising example that your next query ought to retain.
3. Change one condition. Add a synonym, move a title into a filter or test one exclusion. Check the saved example again.
4. Investigate a missing result. A profile may use different terminology, omit a skill or fall outside the index. Absence of the word is not proof of inability.
5. Review before outreach. Check actual work, role constraints and duplicate contacts. Search results alone do not establish availability or interest.

If you need more results, try an adjacent title before weakening an essential requirement. If you get too much noise, identify the source before appending global exclusions. For example, a current-title filter can address people currently recruiting without excluding every engineer who ever mentioned a recruiter. Check that field's current/past setting and retain an exclusion-free baseline.

Carry the evidence into an approved role

Use manual Boolean search when you need direct control of a query or database. Keep it when that control is the task. A better string does not handle candidate replies, resolve questions or book a conversation.

Metix AI is relevant when an employer has a real opening and wants a workflow that continues through interview booking. Bring the role brief, distinguish required evidence from unknowns, review candidate evidence and approve outreach. Mira handles the workflow through booking. Your team conducts interviews and makes hiring decisions. See AI sourcing tool vs AI recruiter for the scope difference.

The self-serve plans include AI Agents and monthly Credits. No Metix staff member reviews the shortlist on those plans. Staff-run Delivery is a separate purchase, quoted on a demo. Metix is not a public Boolean people-search API or MCP endpoint.