Boolean Search for Recruiters: How to Build, Test and Widen a String for One Req

By Sparsh Goyal, Founder at The Cognitive. Published 2026-08-14 · Last updated 2026-10-05

• Boolean search in recruitment is a keyword string built with AND, OR, NOT, quotes and parentheses that finds the candidates one req needs in LinkedIn, a job board, your ATS or Google.
• Build it in 7 steps: one-sentence search map, 3 intake buckets, title group first, one must-have group, channel syntax, read the first 20 results, then calibrate with 5 profiles.
• When the title group keeps growing or one req needs 4 rewrites, a plain-English brief in The Cognitive reads the role into editable filters across ~900M public profiles.

Boolean search in recruitment is a keyword string built with AND, OR, NOT, quotes and parentheses, so LinkedIn, Google, a job board or your ATS returns the candidates one req needs. The syntax takes an afternoon to learn. The real skill is translation: turning a hiring manager's brief into the words candidates actually use, then testing and widening the string until page 1 looks right.

It still pays. In Gem's 2026 benchmarks, sourced candidates are nearly 8 times more likely to be hired than inbound applicants, and 46% of sourced hires came from people already in a company's CRM or ATS, up from 26% in 2021. Both numbers reward recruiters who search well, outside their database and inside it.

This guide is the Boolean search recruiting workflow: 7 steps for one req, a worked example from brief to string to results check, where each channel accepts Boolean, and the mistakes that quietly empty a search. It links out for the rest. Syntax lives in our Boolean operators guide. Ready-made strings by role are in 51 Boolean string examples. LinkedIn plan by plan is in LinkedIn Boolean search. Google technique is in LinkedIn X-ray search and X-ray search beyond LinkedIn. If you want a string written for you, use our free Boolean search generator.

Full disclosure: I run The Cognitive, an AI recruiting platform that sources candidates from a plain-English brief and interviews them live with AI. Near the end I say when a brief beats a string and when it doesn't. Every channel fact below comes from that vendor's own help page, read on 4 October 2026.

What is Boolean search in recruitment, actually?

Boolean search in recruitment is controlled keyword logic applied to finding candidates. You decide which words must appear, which alternatives can appear, which words must not appear, and which phrases have to stay together. Recruiters use it to combine titles, skills, tools, locations and exclusions in one line.

The operators are the easy part. The hard part is choosing which words belong in the string at all. A basic recruiting string looks like this:

(backend OR back-end OR server-side) AND (Python OR Django OR FastAPI) AND (PostgreSQL OR Postgres) NOT intern

That asks for people who describe backend work in 3 different ways, show at least 1 Python signal, mention a Postgres signal, and are not interns. It makes room for variation instead of hunting for one perfect resume.

Recruiters go wrong when they treat precision as adding more required terms. Boolean search does not know what you mean; it only knows what you type. Extra terms help only when they stand for real evidence. If the hiring manager needs someone who has built distributed systems, and your string looks only for the phrase "distributed systems engineer", you are searching for a label and missing everyone who describes the work itself.

Boolean sourcing is the same skill pointed at people who have not applied: profiles on LinkedIn, public pages through Google, resume databases on job boards. Boolean recruiting inside your ATS is the other half, rediscovering people you already paid to attract. For a plain-English definition and the full operator matrix, see our explainer on what Boolean search is. If you are unsure where sourcing ends and recruiting begins, our note on sourcing vs recruiting draws the line.

Boolean searching meaning, in recruiter terms

Think of a string as a map and candidate profiles as the terrain. The map only helps if it matches the ground. A hiring manager asks for a Senior Platform Engineer. People with the right background may call themselves:

• Infrastructure Engineer
• Site Reliability Engineer
• DevOps Engineer
• Backend Engineer, Platform
• Cloud Engineer
• Systems Engineer

Search only the job description title and you miss everyone who can do the work under a different name. Putting all 6 titles in one OR group fixes that before you touch a single skill.

The core Boolean search operators: AND, OR, NOT, quotes and parentheses

The 5 core Boolean search operators each change candidate results in one direction. AND narrows, OR expands, NOT excludes, quotes force an exact phrase, and parentheses group the logic so the engine reads it the way you meant. Most bad strings use the right operator in the wrong place. Below are the recruiter rules. The full list of 12 operators, and which engine supports each one, is in our operators guide.

AND: use it when a requirement is truly required

AND means both sides must appear. React AND TypeScript returns profiles that mention both. That helps when both skills are needed and hurts when TypeScript is a preference. Every extra AND cuts the pool, so make a term mandatory only when the person cannot do the job without it, not because it sits in the job description.

OR: use it to capture how candidates actually speak

OR accepts any term in a group, and recruiters underuse it. It covers title variants, abbreviations and tool families:

(SRE OR "site reliability" OR DevOps OR infrastructure OR platform)
(Kubernetes OR K8s)
(PostgreSQL OR Postgres)
("customer success" OR "customer support" OR "technical support")

Keep each OR group to one hiring signal: a title group, a skill group, a domain group. Unrelated words in one group give you more results and worse ones.

NOT: use it carefully, because it can remove the person you need

NOT drops any profile containing the word. Profiles carry context, so a senior engineer who "mentored interns" disappears under NOT intern. Bullhorn's own help page warns that AND NOT "should be used with caution, because it eliminates candidates that have that keyword in any of the areas you are searching in." If results are full of sales reps when you want sales engineers, strengthen the positive terms first:

("sales engineer" OR "solutions engineer" OR "pre-sales engineer") AND (API OR SaaS OR "technical demo")

That defines the signal. A string like sales AND engineer NOT account NOT SDR NOT quota NOT marketing spends all its energy fighting noise.

Quotes: use them for exact phrases, not every phrase

Quotes force words together, in order. Use them for titles, certifications and product names: "machine learning engineer", "registered nurse". Avoid them for capability. Few strong candidates write "built scalable backend services" word for word. They write APIs, latency, throughput or billing infrastructure.

Parentheses: use them to keep your logic honest

Parentheses tell the engine which terms belong together. LinkedIn's help page lists its reading order as quotes, parentheses, NOT, AND, then OR, so OR binds last. backend OR platform AND Python can therefore read as backend, or platform with Python. (backend OR platform) AND Python is unambiguous. The pattern that keeps a string debuggable:

(title synonyms) AND (core skill synonyms) AND (domain or tool signals) NOT (obvious noise)

When results go wrong, you know which block to change.

A practical operator table for recruiters

How do recruiters build, test and widen a Boolean string for one req?

Recruiters build a Boolean string for one req in 7 steps: write the need as one sentence, sort the intake, build the title group, add one must-have group, adapt the syntax to the channel, read the first 20 results, then calibrate with the hiring manager and save the version that worked. Each step changes one thing, so you can see what moved the results.

1. Write the search map in one sentence. Before any operator, fill this in: we need someone who has done [work], in [environment], using [tools], at [level], with [constraints]. If you cannot write it plainly, the string will hide the confusion instead of solving it.
2. Sort the intake into 3 buckets. Must-have: they cannot do the job without it. Strong evidence: it suggests similar work. Nice-to-have: helpful, never worth excluding someone over. Only the first bucket gets hard AND logic.
3. Build the title group first. Ask what else a qualified person would call this job, and put every answer in one OR group. The fastest source is the hiring manager: ask for 3 profiles that are right but carry a different title.
4. Add one must-have group, then run it. Title group AND one skill group is enough for the first pass. If that returns too few people, your title list is the problem, and adding more AND will not fix it.
5. Adapt the syntax to the channel. LinkedIn rejects wildcards, Greenhouse accepts them, Dice adds word endings for you, and Google does not document OR. The channel table below has the rules.
6. Read the first 20 results and change one block. Mark each profile yes, maybe or no. Too few yes profiles with the right titles means widen: add OR terms or drop an AND group. Too many near misses means tighten: add an evidence group or quote a title. Change one block per rerun.
7. Calibrate with 5 profiles, then save the string. Send the hiring manager 5 profiles, not 50, and ask which is closest and why. Save the version that worked under the role and channel name, so the next req starts from it.

Steps 6 and 7 are where most strings improve, usually through a new word in the title group rather than a new operator. The worked example below runs all 7 on one req.

Worked example: from brief to string to results check

Here is one req taken from brief to string to results check. I wrote these strings by hand and did not run them for counts, because counts change by plan, network and day. Read them as the moves, not as results.

The brief

The hiring manager says: "We need a revenue operations manager for our B2B SaaS team in Chicago. They'll own Salesforce admin and the forecast. Hybrid." As a search map: someone who has run revenue or sales operations, in B2B software, owning Salesforce and forecasting, at manager level, near Chicago.

Sorted into buckets:

• Must-have: a revenue or sales operations background, and Salesforce.
• Strong evidence: forecasting, pipeline reporting, territory planning, CPQ.
• Nice-to-have: a specific forecasting tool, a SaaS company the manager admires.

Version 1: title group and one must-have

RevOps has heavy title drift. The same work shows up as Revenue Operations Manager, Sales Operations Manager, GTM Operations, Business Operations and CRM Operations. So the title group carries the weight, and location goes in the location filter, not the string:

("revenue operations" OR "sales operations" OR RevOps OR "GTM operations" OR "go-to-market operations") AND Salesforce

The results check

Read the first 20 and mark each one. On a string like this, the no profiles tend to fall into a few groups, and each group points to one fix:

Version 2: widened titles, tightened evidence

One change widens the titles. The next adds the evidence group. Each gets its own rerun:

("revenue operations" OR "sales operations" OR RevOps OR "GTM operations" OR "business operations" OR "CRM operations") AND Salesforce AND (forecast OR forecasting OR "pipeline reporting" OR "territory planning")

Only now, if one word keeps marking the no profiles, add a single exclusion and check it against the yes profiles first.

The same req, channel by channel

Each channel takes the same logic in slightly different syntax:

• LinkedIn Recruiter: put the title variants in the Job titles filter set to Can have, Salesforce in Skills set to Must have, and the evidence group in Keywords. LinkedIn takes no wildcards, so write forecast OR forecasting in full.
• Greenhouse, rediscovering past applicants: turn on Full Text Search and use forecast* to catch forecast, forecasts and forecasting in resumes, notes and scorecards.
• Indeed resume search: the version 2 string uses only operators Indeed documents, so paste it as written, with operators in capitals.
• Google: an X-ray version needs site: and drops AND. Our LinkedIn X-ray guide covers that format.

Calibration

Send the hiring manager 5 profiles from version 2 and ask 5 questions:

• Which profile is closest to the real need?
• Which title surprised you but still fits?
• Which skill did I overvalue?
• Which missing phrase should I add?
• Which rejection reason should become an exclusion, if any?

Their answers become version 3. Most of the time, the change is a word in the title group.

What is Boolean search in recruitment on LinkedIn vs Google vs an ATS?

Boolean search in recruitment works in 4 kinds of channel: LinkedIn, job board resume databases, your ATS, and Google. Each indexes different data and documents different operators, so one string pasted everywhere underperforms. Here is what each vendor documents, read on 4 October 2026:

ATS Boolean search

An ATS search is only as good as the data in it. If your ATS holds resumes, application answers, notes and interview feedback, Boolean search can resurface people you already paid to attract. That is where the Gem number bites: 46% of sourced hires now come from rediscovered candidates. Our step-by-step rediscovery guide turns that into a weekly routine. Greenhouse makes this explicit, since a full-text Boolean query searches resumes, notes and scorecards together. It is available on Greenhouse's Core, Plus and Pro tiers.

ATS data is messy. Resumes are formatted differently, recruiter notes use shorthand, and older resumes are out of date. Boolean in an ATS is best for rediscovery:

• Past silver medalists
• People rejected for timing or compensation, not capability
• Candidates with adjacent skills
• Applicants from previous similar roles
• People who passed a technical round but lost to a stronger finalist

The Cognitive does not search your ATS history, so keep your ATS's own Boolean search for this job. It connects to 60+ ATSs, including Greenhouse, Lever, Workday and Bullhorn, and sits alongside the ATS rather than replacing it or the recruiter. For candidates you import from the ATS, it writes a note with the interview score, summary and report link back to the ATS.

Job board Boolean search

Job board resume databases reach people who are actively looking, and both big US boards document Boolean. The useful twist is Dice: you don't need an asterisk, because "the database already auto populates the various endings". So (admin) on Dice already covers administrator.

Gem's benchmarks are a reminder of the trade-off: job boards and company marketing channels generate roughly 90% of all applications but only about half of hires. A resume database search is a better use of a job board than waiting for applicants, but it still finds the people who are looking. Our pages on searching Indeed resumes and the best resume databases for recruiters cover access and cost.

LinkedIn Boolean search

LinkedIn is strong on titles, companies, skills and location, and weaker when profiles are sparse or the real work sits in a project description. It also nudges you toward title searches, which makes your title OR group matter more than almost anything else:

("machine learning engineer" OR "ml engineer" OR "applied scientist" OR "data scientist" OR "research engineer")

Then add evidence such as (PyTorch OR TensorFlow OR "model deployment" OR MLOps). The plan changes the limits, as the table shows, and our LinkedIn Boolean search guide goes product by product.

Google X-Ray Boolean search

Google X-ray search points Boolean at one site with site:, so you can search public LinkedIn profiles, GitHub, portfolios and conference pages. It is broad, which cuts both ways: it finds public pages a recruiting database misses, along with stale profiles, duplicates and old job posts. Google warns not to put a space between an operator and the term, so test page 1 before you trust the syntax. Strings live in our LinkedIn X-ray guide and our guide to X-ray search beyond LinkedIn.

AI sourcing and Boolean search are not enemies

Boolean wins when you know the exact logic you want. AI sourcing wins when the role is messy, niche or full of title variation, because you describe it in plain English instead of operators. The strongest workflow uses both, with a recruiter checking the assumptions. The section on using a brief below shows where the line falls.

Common mistakes that kill Boolean searches for recruiters

The most common mistakes in Boolean searches for recruiters are copying job description language, making every signal mandatory, using NOT to patch a weak search, ignoring title drift, skipping the first 20 results, and treating a keyword match as proof of skill. Each one makes a string look sophisticated while quietly removing viable candidates. Syntax slips, like lowercase operators or smart quotes, are covered in the operators guide.

Mistake 1: treating the job description as the candidate dictionary

Job descriptions use the company's own naming, level language and wishlist. Candidates rarely mirror it. A JD says "experience building scalable distributed systems in a cloud-native environment". A candidate writes:

• Built high-throughput billing services on AWS
• Reduced API latency by 40 percent
• Owned Kafka-based event pipelines
• Migrated monolith services into Kubernetes

Same capability, different vocabulary. Search only the JD phrase and you miss the builders.

Mistake 2: making every good signal mandatory

A hiring manager lists Python, AWS, Kubernetes, Kafka, Terraform, Postgres and CI/CD. A recruiter ANDs all 7. The pool shrinks to almost nobody, or to people who stuffed every keyword into their profile. The 3 intake buckets from step 2 fix this.

Mistake 3: using NOT to fix a bad positive search

NOT removes noise fast and good candidates invisibly. Sourcing technical recruiters with NOT agency NOT staffing drops people who learned high-volume sourcing at an agency, then moved in-house and got very good. Search for the behaviours that matter instead: sourcing, calibration, engineering hiring, LinkedIn Recruiter. Our technical recruiter hiring page shows how we describe and interview for that role.

Mistake 4: ignoring title drift

Title drift is the same work carrying different titles at different companies. Platform, infrastructure, DevOps, SRE, cloud and backend overlap depending on company maturity. When a hiring manager says a rejected candidate "had the right background but a different title", stop the search. Your string is filtering on vocabulary. Ask for 3 right profiles with other titles and rebuild the OR group around them.

Mistake 5: failing to read the first 20 results

The first 20 results are feedback from the market. Too senior means the title or level terms are off. Too junior means you need responsibility words, not the word senior. Adjacent but wrong means the positive terms are too broad. Empty means too many AND groups. Ashby's report shows why reading beats counting: the average recruiter now processes 291 applications per hire, compared to roughly 100 in early 2021. A loose string that returns thousands of matches builds the same pile in your sourcing tab.

Mistake 6: confusing resume signal with interview signal

Boolean search finds likely candidates. It cannot show they can do the job, and AI-polished profiles make that gap wider. Someone can list Kubernetes, Kafka and distributed systems without having made a hard production call. A live conversation tests that.

The Cognitive's AI Interviewer runs a live, two-way AI video interview of 10 or 20 minutes, in 9 languages, at a slot the candidate books without an account. The invitation tells candidates the interview is run by AI. The rubric is fixed per role and the questions adapt live, following up on thin answers. The report gives a 1 to 5 score per criterion, a weighted score out of 100, a suggested verdict, the transcript and the recording, and it checks up to 5 resume claims, each marked verified, refuted or unclear. Nothing is rejected automatically.

The recording above, made on 1 September 2026, is 5 minutes of a live AI interview for a go-to-market role. The AI asks the candidate about an outbound email campaign, then follows up on reply rates and how leads were qualified. A profile can carry every keyword in your string and still fail those 2 follow-ups, which is why the Boolean list feeds the interview stage instead of replacing it.

What is a Boolean search cheat sheet for recruiters?

A Boolean search cheat sheet for recruiters is a reusable aid for building, testing and repairing strings: operator patterns, title synonym prompts, skill buckets, an exclusion check and channel reminders. This one is about the thinking. For a one-screen syntax sheet of every operator by engine, use the operators guide.

Title synonym groups

Ask "what else would a qualified candidate call this job?" before every niche search. A Data Engineer search, for example, starts from data engineer, analytics engineer, ETL developer and BI engineer. A Technical Recruiter search starts from technical recruiter, engineering recruiter, talent partner and sourcing recruiter.

Must-have skills vs evidence skills

Reusable Boolean patterns

The 4 patterns recruiters come back to most:

Title plus skill:
("backend engineer" OR "software engineer") AND (Python OR Go OR Java)

Title plus domain:
("product manager" OR "product owner") AND (fintech OR payments OR banking)

Skill plus seniority evidence:
(Kubernetes OR Terraform OR AWS) AND (owned OR led OR architected OR migrated)

Technical recruiter:
("technical recruiter" OR "engineering recruiter" OR "sourcing recruiter") AND (Boolean OR sourcing OR GitHub) AND (engineers OR developers OR software)

Each block has one job. A block without a job does not belong in the string. More by role are in our Boolean string examples.

The exclusion check

Before you add NOT, answer 3 questions:

1. Is this term always wrong, or only sometimes? If only sometimes, leave it in.
2. Could a strong candidate use this word in a good context? If yes, be careful.
3. Can I improve the positive search instead? Better signal beats heavier exclusion.

The practical Boolean search cheat sheet

When should you stop writing strings and use a plain-English brief?

Stop writing strings when the string has become the work. Four signals tell you that point has come:

• The title group passes 8 or 10 variants and keeps growing. The market has no settled name for the job, and you are guessing at vocabulary.
• The string is too long for the tool. LinkedIn's own Recruiter help says that when a Boolean query gets too long, you should move the items into the individual filters. Sales Navigator stops at 15 operators.
• You are rewriting the same req for 4 channels. LinkedIn without wildcards, Greenhouse with them, Google without AND.
• The pool is thin and you can't tell which AND group is starving it.

One case calls for neither a string nor a brief. When the role is poorly defined, or the hiring manager can't say what strong performance looks like, a long string just gives false confidence. Calibrate first, tighten the role, then source.

A plain-English brief hands the translation step to software. In The Cognitive AI Sourcing Tool you type the role as one sentence, or start from a job description, and it is read into editable filters you can check before anything runs. I tested it with "Senior backend engineer with 5+ years of Go who has built payments systems, Remote EU". "Remote EU" came back as a location filter for Germany plus 7 other countries, the OR group of countries you would otherwise type by hand.

The search then shows 2 phases, "Reading your brief and finding relevant candidates…" and "Ranking candidates based on your requirements…". Every person found is judged against the must-haves in your brief. Anyone who misses one drops below the people who meet them, so nobody vanishes the way NOT makes them vanish. Each result card opens with a "Why this match" line giving 1 or 2 reasons traced to the profile. When the pool is thin, the search relaxes its softest filters and tells you what it broadened. Later searches for the same role skip people you have already seen.

Boolean still wins in 3 places. Your own ATS history is one, since The Cognitive does not search it. An exact licence or certification term you want matched word for word is another. InMail-led outreach is the third, because The Cognitive sends email and SMS, not LinkedIn InMail. For how brief-to-filter search works across tools, read our guide to natural language candidate search.

Type the req you would have written as a string One sentence, read into editable filters and judged against your must-haves across ~900M public profiles. Start free

Who is The Cognitive built for?

The Cognitive is built for recruiters and hiring teams who spend more time writing strings than talking to candidates. You describe the role once, check the filters, shortlist from ranked results, reveal contacts for the people you pick, and invite them to a live AI interview in one go. It suits in-house teams and agencies hiring roles with messy titles, and small teams without a dedicated sourcer.

Contacts come back as verified emails and phone numbers, and email and SMS outreach sequences are on the Sourcing Pro plan. A search costs 1 credit per candidate it returns. Revealing an email costs 5 credits and a phone number 10, charged only when a value comes back. AI Sourcing starts at $49/month and AI Interview at $99/month, on monthly plans you can cancel anytime.

Keep, or pair it with:

• Your ATS's Boolean search for rediscovering past applicants in Greenhouse, Bullhorn or similar.
• LinkedIn Recruiter if InMail is the channel your candidates answer.
• Our free Boolean search generator when you still want a string for LinkedIn or Google.

If you are comparing sourcing tools, look past whether they accept clever strings. Ask whether they turn a messy req into people you can reach and evaluate. We cover that gap in our AI sourcing tools comparison and the Juicebox alternatives guide, and our AI recruitment platforms roundup shows how sourcing, interviewing and shortlisting fit together without replacing your ATS.

To test the idea on one real role, start with your hardest open req. Build the Boolean string and read its first 20 results. Then type the same req as a brief, compare the two lists, and invite the strongest few to a live AI interview. The Cognitive has a free trial with 100 sourcing credits.

Run your hardest req as a brief, not a string Describe it in one sentence, see who meets every must-have, then reveal contacts and interview the people you pick. Start free

Sources

• Gem, "Key takeaways from the 2026 Recruiting Benchmarks Report", gem.com, read 4 October 2026.
• Ashby, "Recruiter Productivity Trends Report", ashbyhq.com, read 4 October 2026.
• LinkedIn Help, "Use Boolean search on LinkedIn", linkedin.com, read 4 October 2026.
• LinkedIn Help, "Boolean query limitations", linkedin.com, read 4 October 2026.
• LinkedIn Recruiter Help, "Use Boolean to filter search results in Recruiter and Recruiter Lite", linkedin.com, read 4 October 2026.
• Indeed, "What is Boolean search?", indeed.com, updated 28 August 2025, read 4 October 2026.
• Dice, "Build Better Boolean Search Strings", dice.com, 26 March 2024, read 4 October 2026.
• Greenhouse Support, "Search candidates using Boolean queries", support.greenhouse.io, updated 2 March 2026, read 4 October 2026.
• Bullhorn Knowledge Base, "Understanding Boolean Logic", kb.bullhorn.com, read 4 October 2026.
• Google Search Help, "Refine Google searches", support.google.com, read 4 October 2026.

All posts · thecognitive.io