# The Recruiter-Search Keyword Bank, Built From Postings to Placed Fields

*You will build a ranked keyword bank from your own target postings, place each term in its highest-weight field, and confirm you are surfacing.*

- Canonical URL: https://www.refolk.ai/candidates/guides/recruiter-search-keyword-bank
- Pillar: Positioning and materials
- Format: Playbook
- Published: 2026-09-21
- Last reviewed: 2026-09-21
- Reading time: 16 min

This is the ordered method for getting recruiters who search for your target role to actually find you. It is for anyone rewriting a LinkedIn profile who is tired of "add keywords to your headline" advice and wants a documented procedure instead: how many postings to sample, which term beats which, where each term goes, and how to confirm it worked. Follow it start to finish and you will have a ranked keyword bank routed into the right fields, plus a baseline you can watch and a trigger to rebuild.

The premise is simple and load-bearing. Over 90% of recruiters search the member database for candidates with specific keywords in their profile. If your profile does not contain the strings a recruiter types, into the fields their search reads first, you never enter the result set. There is no near-match consolation prize.

## Why keyword routing beats keyword count

The single biggest lever is not how many keywords you have, it is where they sit. Recruiter search filters by title before it reads anything else, so the same term is worth far more in your headline than buried in a decade-old bullet.

When a recruiter uses LinkedIn Recruiter, they usually filter by job title first, then skills, then location. Title-based filtering runs *before* full-text relevance, which means placement decides whether you enter the candidate set at all, and relevance only reorders the people who already made it in. A keyword-stuffed About section does nothing if your title field is empty of the term they filtered on.

Sources converge on this weight order even where they disagree on the exact math.

#### How recruiter search reads a profile

1. **Headline and current job title** - Highest weight, title-matched, filtered first
2. **Skills** - Next weight, especially the pinned top three
3. **About** - Full-text searchable, first ~300 characters slightly heavier
4. **Experience** - Depth once a recruiter is already reviewing you

*Title fields decide who enters the result set; lower layers reorder the survivors.*

Two cautions about the numbers you will see elsewhere. A widely repeated claim that "headline plus current position is about 60% of ranking weight" traces to a document that cannot be verified as official LinkedIn material. A keyword vendor asserts specific "10x/7x/5x/4x" zone weights. Both are marketing claims, not LinkedIn-published facts. Use the order - title, skills, About, experience - as directional, and do not build decisions on the exact percentages.

> **Rule:** Title fields are non-negotiable
>
> Your plain, searchable job title must appear in both the headline and the current position. Nobody recruits for "Storyteller"; if "Software Engineer" sits nowhere near the top, you do not come up.

## The acronym gap is the highest-leverage cheap edit

Spell out every acronym and also include the acronym itself, because the two strings are matched literally and a recruiter never sees the form you omitted. This one edit costs minutes and closes a gap measured in orders of magnitude.

The scale is not marginal. In Refolk's index of professional profiles, the skill "SEO" returns 216,586 people in the United States, while the spelled-out "Search Engine Optimization" returns 2,870. That is roughly a 75x difference in how many candidates a recruiter reaches depending on which form they type.

**75x - The SEO acronym gap in Refolk's US index**

"SEO" returns 216,586 people; "Search Engine Optimization" returns 2,870. Carry only one form and you are invisible to searchers of the other.

**B. Acronym vs full phrase, one market (Refolk's index)**

| Skill term | Country | People |
|---|---|---|
| "SEO" | United States | 216,586 |
| "Search Engine Optimization" | United States | 2,870 |
| Acronym-to-full multiple (derived) | - | ~75x |

The mechanism is dumb literal matching. LinkedIn maps some synonyms on its own, but the coverage is unreliable, so do not gamble your visibility on it. Store both forms for every term: "Identity and Access Management" and "IAM", "Search Engine Optimization" and "SEO". For certifications, write both the full name and the abbreviation, for example "Project Management Professional (PMP)". That single pattern turns a tidy piece of advice into the highest-return edit on the page.

> Spell both forms of every term, because the string you leave out is a door a recruiter can walk through past you.

## How many postings to sample and where to get them

Collect 8 to 10 real postings for your exact target title, from listings you would genuinely apply to. That is enough repetition to rank terms by frequency without turning collection into its own project.

Guidance splits, and the split is worth knowing so you can defend your choice. LinkedIn's own content recommends studying 3 to 5 postings; Austin Belcak sets a minimum of five, on the logic that five job descriptions balance time invested against output quality; and some guides push to 10 to 15. The tension is real: too few and one company's house-style phrasing dominates your ranking; too many and you spend an afternoon on diminishing returns. Eight to ten is the defensible middle.

> **Tip:** Sample from your real target, not the aspirational one
>
> Collect postings you would actually click apply on. If your target spans two adjacent titles, sample both and keep the banks separate. A bank built from roles you will never pursue optimizes you for the wrong searchers.

Market size should shape which titles you sample. In Refolk's index, the title "Product Manager" with the skill Product Management returns 12,606 people in the United States and 3,273 in the United Kingdom, roughly a 3.85x difference.

**A. Same title, two markets (Refolk's index)**

| Title + skill | Country | People |
|---|---|---|
| Product Manager / Product Management | United States | 12,606 |
| Product Manager / Product Management | United Kingdom | 3,273 |
| US-to-UK ratio (derived) | - | ~3.85x |

The takeaway is counterintuitive. In a thinner market the exact-title pool is smaller, so title variants and adjacent titles matter *more* for surfacing, not less, because a recruiter in that market casts a wider net across near-titles. If you are in a smaller market, sample a slightly broader set of titles and note the variants in your bank.

## The procedure, end to end

Run these eight steps in order. The whole thing takes roughly two to three hours the first time, and under an hour on a refresh once you have a bank to update. Each step names what "done" looks like so you can stop when it is done rather than when you get bored.

#### Build and place the keyword bank

1. **Assemble the posting sample** - Search for your exact target title and collect 8 to 10 real postings into one document. Done when the postings sit in a single file.
2. **Extract raw terms** - Pull titles, hard skills, tools, and certifications from the combined text; a word cloud over the merged postings surfaces the frequent ones. Done when you have an unranked term list.
3. **Rank by frequency** - Count how many postings each term appears in and sort descending. If four postings say "stakeholder management" and one says "cross-functional leadership," the higher-frequency phrase wins. Done when the bank is ranked with counts.
4. **Pair acronyms with expansions** - For every term, record both the acronym and the spelled-out phrase in one entry. Done when each entry holds both forms.
5. **Route each term to its highest-weight field** - Plain job title into headline and current position first, top skills pinned, About and experience for depth. Done when every top term is placed.
6. **Write the headline within limits** - Front-load the searchable title in the first ~60 characters, use the rest up to 220 for supporting keywords. Done when the headline is under 220 with the title in the opening line.
7. **Verify via analytics** - Open Analytics, then Search appearances, and record the baseline plus searcher job titles and companies. Done when a dated baseline is written down.
8. **Refresh on trigger** - Re-run the whole method quarterly, or when your target role, industry, or location changes. Done when the bank carries a fresh date.

For extraction at step two, an open-source keyword extractor can process a batch of job descriptions and surface the most frequent terms, and AI tools can do the same in seconds over 20 to 30 postings. A manual word cloud over the merged text works fine for 8 to 10. Either way, the output you want is a frequency-ranked list, because frequency across your own target set is the closest proxy you have for what recruiters in that market actually type.

At step three, prefer specific over generic. Use "Python" and "Salesforce", not "data analysis" and "CRM". Specific terms match specific searches and mark you as someone who did the actual work rather than someone describing a category. The frequency count tells you which specific terms your target market repeats.

Refolk automates the parts of this that are pure friction. When I tailor a resume to a posting, I extract the posting's terms, rank them, and score how well your history already covers them, which is the same keyword-bank logic applied per application rather than to your standing profile. [Refolk](/candidates) does the extraction and the fit scoring so the ranking step is not a manual word-cloud exercise every time your target shifts.

## Writing the headline inside the character limits

The headline has a hard limit of 220 characters, but only a fraction shows in a search preview, so front-load the searchable title in the first 60 characters and use the rest for supporting keywords. Everything past the fold still indexes; it just will not display in the preview a recruiter scans.

The visible cut varies by source and device, which is why "first 60" is the safe rule rather than a precise figure.

**C. Headline characters visible in search (published estimates)**

| Source | Hard limit | Visible in search |
|---|---|---|
| authoredup | 220 | ~60-70 |
| justpollen | 220 | ~60-80 |
| pursuenetworking | 220 | ~120 desktop / ~60 mobile |
| posttruncate | 220 | ~100 |

These are observed UI behaviors, not LinkedIn-published constants, and mobile truncates earlier than desktop. Because the lowest credible estimate is around 60 characters, treat 60 as your budget for the part that must be visible: the plain title. Load the remaining ~160 characters with your ranked supporting keywords, which index and help relevance even though they may not show in the preview.

**Searchable headline skeleton (max 220 chars)**

```
[Plain Target Title] | [Specialty or Seniority] | [Top Skill], [Top Skill] | [Full Phrase] ([ACRONYM]) | [Industry/Domain]

Example:
Product Manager | B2B SaaS | Product Discovery, Roadmapping | Search Engine Optimization (SEO) | Fintech
```

*Keep the plain title in the first ~60 characters; fill the rest with ranked supporting keywords and both acronym/full forms.*

Note what the example does. "Product Manager" - the title a recruiter filters on - sits in the first line. The specialty and top skills follow. The acronym-plus-full-phrase pair rides along in the middle. Nothing is a clever tagline, because a clever tagline contains no searchable title and there is no near-match rescue.

## Verify that you are actually surfacing

Do not assume the edits worked; check the search-appearance data and, specifically, who is searching. Go to your profile, click "See all analytics" beneath your introduction section, and open the "Search appearances" tab, where profile views, search appearances, and profile appearances sit together along with the job titles and companies of the people whose searches you appeared in.

Two things gate and confound this number, and both get misread constantly.

First, access is gated. Full analytics require three or more unique profile viewers in the past seven days; below that the dashboard is limited or empty *by design*. A blank dashboard on a new or low-traffic profile means under-viewed, not badly optimized. Check your unique viewer count before you conclude your keywords failed.

Second, the raw appearance count is confounded. Appearances aggregate every surface where your name, photo, and headline showed up, whether or not anyone clicked, and they jump for reasons unrelated to keyword fit. In one analysis, 41% of profile appearances came from comments, not posts. So a keyword-perfect but silent profile can show flat numbers, and a spike can come from commenting activity rather than from recruiters finding your terms.

> **Watch out:** The searcher job-titles list is the real signal
>
> The appearance count answers "how often did I show up." The searcher job-titles list answers "did the right people find me." Read the titles and companies, not just the number. Flat numbers with the wrong searchers means your keywords are off; growing numbers from your target companies means they are working.

The premium tier reveals which keywords searchers used, starting around $29.99 a month. That is the only native way to see the exact search strings, but it is not required to run this method. The free searcher job-titles list plus a dated baseline is enough to tell direction. Guidance on the number itself is directional only: there is no credible per-role benchmark for absolute search-appearance counts, so aim for steady growth, especially from your target companies or industries, rather than a magic threshold.

#### Reading the search-appearance dashboard correctly

1. **Check viewer gate** - 3+ unique viewers in 7 days, or the dashboard is blank by design
2. **Record the count** - Note search appearances with today's date as a baseline
3. **Read the searchers** - Check the job titles and companies, not just the number
4. **Cross-check surfaces** - Compare against profile views and inbound messages to rule out a comments-driven spike

*Confirm the dashboard is even populated before you judge your keywords.*

## How this goes wrong

Most keyword work fails in predictable ways, and each failure has a cheap check. Run these against your profile before you call the job done.

- **Trusting the "60% weight" and "10x zone" numbers.** These trace to SEO vendors citing an unverifiable "LinkedIn Recruiter Search Guide." *Check:* demand a LinkedIn.com or Help-center URL; there isn't one. Use the field order, not the fake percentages.
- **Optimizing the headline for humans only.** A tagline like "Storyteller" contains no searchable title, so you never surface. *Check:* does the plain title appear in both the headline and current position?
- **Front-loading past the fold.** Keywords after ~60 characters still index but vanish from the search preview. *Check:* read your first 60 characters alone; is the title there?
- **Reading search appearances as interest.** Appearances count every impression, clicked or not, and jump for unrelated reasons. *Check:* compare against profile views and inbound messages, and read the searcher titles.
- **Empty analytics misread as invisibility.** Below three unique viewers in seven days the dashboard is blank by design. *Check:* confirm your viewer count before blaming your keywords.
- **Carrying one form of a term only.** Dataset B shows a ~75x gap between "SEO" and its spelled-out form. *Check:* is every term stored in both acronym and full-phrase form?
- **Set-and-forget bank.** Terminology drifts and a stale bank silently decays. *Check:* is there a date on the bank, and is it under a quarter old?
- **Keyword stuffing.** LinkedIn can detect and penalize unnatural keyword density, and stuffed fields read like a robot wrote them. *Check:* does each field read like a human wrote it?

#### Diagnose flat or empty search appearances

Horizontal axis runs from Few unique viewers to Many unique viewers. Vertical axis runs from Keywords well-placed to Keywords weak or misplaced.

| Quadrant | What it means |
| --- | --- |
| Under-viewed, well-optimized | Dashboard is blank by design; drive views, do not touch keywords |
| Surfacing correctly | Working as intended; watch the searcher titles for drift |
| Under-viewed and weak | Fix placement first, then drive views to populate the dashboard |
| Visible but wrong terms | Re-rank the bank and re-route; you show up for the wrong searches |

*Where a profile sits decides whether the problem is keywords or traffic.*

The false positive that costs the most is the "visible but wrong terms" quadrant: your appearance count grows, you feel optimized, but the searcher titles are all off-target. That is why the searcher job-titles list, not the raw count, is the verification signal.

## The pre-publish checklist

Before you call the profile done, confirm each of these. This is the standard the finished work has to pass.

#### Confirm before you call it done

- [ ] The plain, searchable target title appears in both the headline and the current position.
- [ ] The title sits inside the first 60 characters of the headline.
- [ ] The headline is under 220 characters total, with supporting keywords filling the rest.
- [ ] Every term in the bank is stored and placed in both acronym and full-phrase form.
- [ ] The top three skills are pinned and match the highest-frequency terms from the sample.
- [ ] Specific terms (tools, languages, named skills) are used instead of generic categories.
- [ ] Each field reads like a human wrote it, with no stuffed keyword density.
- [ ] A search-appearance baseline is recorded with today's date.
- [ ] The searcher job-titles list has been read to confirm the right people are finding you.
- [ ] The keyword bank carries a date and a note on which titles and market it was built for.

## Keep the bank alive

A keyword bank is a dated artifact, not a one-time task, so re-run the whole method quarterly or the moment your target role, industry, or location changes. Terminology drifts, new tools enter the postings, and titles rename themselves; a stale bank quietly stops matching what recruiters type.

Both a schedule and an event trigger appear in the guidance. One recruiter source advises refreshing at least every six months or whenever your target changes; another recommends saving AI-generated keyword lists by target role and updating quarterly as you collect new postings, so you can watch how the terminology evolves. Quarterly is the tighter, safer cadence. Keep the ranked banks by target role in one document, each stamped with its date and the market it was built for, so a refresh is an update rather than a rebuild from zero.

> **Note:** The refresh trigger is the date plus the target
>
> Two conditions fire a rebuild: the bank is more than a quarter old, or your target role, industry, or location moved. Either one means re-collect the sample and re-rank. The date on the bank is what makes "set and forget" impossible to do by accident.

When the target shifts, the expensive part is re-extracting and re-ranking terms from a fresh set of postings, and then re-scoring how well your existing history covers them. That is exactly the work Refolk runs per posting when it tailors materials to an application, so if your target is moving often, let the extraction and fit-scoring run automatically and spend your own time on placement and headline judgment, which no tool should make for you.

## Frequently asked questions

### How many job descriptions do I need to build a good keyword list?

Aim for 8 to 10 real target postings. Practitioner guidance splits: LinkedIn's own content suggests 3 to 5, Austin Belcak sets a minimum of five, and some guides push to 10 to 15. Eight to ten is a defensible middle that balances the time invested against the quality of the frequency ranking, giving you enough repetition to separate signal terms from one-off phrasing without turning collection into a project.

### Which profile fields do recruiters actually search?

Recruiters using LinkedIn Recruiter typically filter by job title first, then skills, then location, before full-text relevance runs. Headline and current job title carry the most weight, skills next (especially your pinned top three), and About plus experience matter for depth once someone is already reviewing you. Vendor claims of exact zone weights like 10x or 7x are marketing, not LinkedIn-published, so treat placement order as directional, not precise.

### Should I use the acronym or the full phrase on my profile?

Use both, every time. In Refolk's US index the skill SEO returns 216,586 people versus 2,870 for Search Engine Optimization spelled out, roughly a 75x gap. The two strings are matched literally, so a recruiter who types one form never sees a profile carrying only the other. Store each entry as acronym plus full phrase, and for certifications write both, for example Project Management Professional (PMP).

### How do I check my LinkedIn search appearances?

Go to your profile, click See all analytics beneath your introduction section, and open the Search appearances tab. It shows profile views, search appearances, and profile appearances together, along with the job titles and companies of searchers. Full analytics are gated: they require three or more unique profile viewers in the past seven days, so a low-traffic profile may show a blank or limited dashboard by design.

### My search appearances are flat. Are my keywords wrong?

Not necessarily. Appearances aggregate every surface where your name showed up, and in one analysis 41% of profile appearances came from comments rather than posts, so a keyword-perfect but silent profile can look flat. Before blaming keywords, check the searcher job-titles list to confirm the right recruiters are finding you, and confirm you have three unique viewers in seven days so the dashboard is even populated.

### How often should I rebuild the keyword bank?

Refresh quarterly, or immediately whenever your target role, industry, or location changes. Recruiter guidance ranges from reviewing at least every six months to updating the bank quarterly as you collect new postings. Terminology drifts, so a stale bank silently decays. Stamp the bank with a date and treat the date as the trigger: if it is more than a quarter old or your target moved, re-run the collection and ranking.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/recruiter-search-keyword-bank*