Refolk
ReferenceRecruiting and sourcing

The Local-Title Search Reference: Native Forms, Variants, and Level Traps

Look up one market row and write a first-run search with the right local title forms, gendered and script variants, and a seniority band set in years.

20 min readLast reviewed November 15, 2026Read as Markdown

Key takeaways

  • In Refolk's index, 37,043 German software profiles use the English title and 14,561 use the German one, so about 72% of the pool is invisible to a German-only search.
  • In a 25-profile German-title sample, the hyphenated "Software-Entwickler" outnumbered "Softwareentwickler" 11 to 7, so a search that uses only one spelling loses a large share of the pool.
  • About a quarter of every 25-profile title sample in Refolk's index (6 or 7 of 25) had a current title such as founder or CxO, so read 25 current titles before you trust any count.
  • Brazil's "pleno" has no national definition: published ranges run from 2-5 years to 6-9 years, so search on the word but filter on tenure.
  • Prokurist and ingenieur diplome record a legal power or a degree, not a level, so flag them and band the person on functional title and tenure instead.
  • Practitioner sources report that platform search ignores accents while a Google X-ray matches them literally, so every X-ray needs both the accented and the unaccented form.

This reference is for the in-house recruiter, sourcer or founder who has a req in a country they have never searched before: Munich, Tokyo, São Paulo, Bangalore, Paris. For each market, its row gives you the local title forms to OR together, the gendered, hyphenated and script variants, the seniority band in years and span of control, the English-versus-local split where it has been measured, and the trap that makes the title overstate or understate level. Find the row you need, build the string, and run the procedure before you trust the count.

What does a local-title search have to cover?

A local-title search has to cover five layers of the same job: the location string, the language, the spelling, the gender form and the script. If you miss one layer, the search quietly drops part of the pool and gives you no sign that anything is missing.

Three terms are used throughout:

  • Seniority band: the range of years of experience and span of control you want, written down before you search.
  • Statutory title: a title that records a legal power or a protected degree, not a rank in a hierarchy.
  • X-ray: a Google search limited to a site's public profile pages with the site: operator, used to catch what the platform's own search misses.

The five layers of one local title

  1. Location
    City aliases such as "Berlin, Berlin, Germany" and "Berlin Metropolitan Area"
  2. Language
    The English title on a local profile versus the native-language title
  3. Spelling
    Compound versus hyphenated forms, such as Softwareentwickler and Software-Entwickler
  4. Gender
    Masculine, feminine and inclusive forms, such as -in, *in and :in endings
  5. Script
    Native script, romanized and accent-stripped forms, such as 課長, kachou and kacho
Every layer multiplies the forms a single job can take, and a search that skips any layer silently loses that slice of the pool.

A note on the US ladder. None of the sources behind this reference maps these local ladders to a US ladder directly, so I don't invent one. Every row converts through two common units: years of experience and span of control. Write your US req in those units first, then read across.

How much of a market titles itself in English?

In Refolk's index, about 72% of German software profiles use the English title and 28% use the German one, roughly 2.5 to 1. Germany is the only market where both sides have been measured. For every other market, treat the split as unknown and search both languages.

Title formProfilesShare of combined
English (Software Engineer/Developer)37,04371.8%
German (Softwareentwickler/-in, Software-Entwickler/-in)14,56128.2%
English-to-German ratio2.54 : 1derived
Brazil English (Software Engineer/Developer)24,222Portuguese not measured
72%
of German software profiles in Refolk's index use the English title
37,043 English-titled profiles against 14,561 German-titled ones, a combined pool of 51,604.

The split also changes who you find. In the English sample, international employers such as Zalando, Helsing and Vercel appear. The German sample leans toward Mittelstand firms such as DATEV and Giesecke+Devrient. So the language you search in also picks the kind of employer you surface, and not just the number of profiles.

Know where this evidence stops. The split was measured for one role family (software) in one country. Commercial roles were not measured. Brazil's English count is known but its Portuguese count is not, and the other markets in this reference were not measured at all. As far as I know, nobody has published these figures for any market. You can measure the split for your own market in a few minutes: count each title form separately with the same location and compare. I took the German figures above from Refolk the same way, one form per query, and the method works in any tool that reports a result count.

Market lookup: local forms, band and trap

Each market row below gives the forms to OR, what the title tells you about level, and the trap that makes it lie. Use the summary table to orient, then read the subsection for the market you are working in.

MarketForms to ORWhat the title signalsMain trap
GermanySoftware Engineer/Developer, Softwareentwickler/-in, Software-Entwickler/-inRole, not levelProkurist is a power of attorney
Japan部長, 次長, 課長, 係長, 主任 plus romanizedSpan of control部長 is not necessarily a board member
Korea부장, 과장, 팀장 plus romanizedTime served, looseningRank can report to a lower-ranked team lead
BrazilAnalista or desenvolvedor/desenvolvedora júnior, pleno, sêniorEmployer-definedPleno spans 2-5 to 6-9 years
IndiaAGM, DGMDepends on employerAGM floor is 8+ or 2-5 years
FranceIngénieur, ingénieur diplôméA degreeDegree read as level

Germany

Forms to OR: Software Engineer, Software Developer, Softwareentwickler, Softwareentwicklerin, Software-Entwickler, Software-Entwicklerin.

What the sample shows: In a 25-profile German-title sample from Refolk's index, the hyphenated "Software-Entwickler" appeared 11 times and the compound "Softwareentwickler" 7 times. No feminine form showed up among the top titles. Include the -in forms anyway. They cost nothing, and a small sample proves nothing about their absence.

Gender markers: "m/w/d" is a convention for job postings. Section 11 of the AGG (Germany's equal-treatment law) says a position may not be advertised in a discriminatory way, but the AGG does not require that particular abbreviation. Inclusive forms like "Manager*in" or "Manager:in" are also AGG-compliant. The "d" stands for "divers", which people in Germany have been able to register as since 1 January 2019. How often any of these markers appear on individual profiles has not been established. Add *in and :in forms as OR terms and check whether they return anything in your sample.

Leiter titles: Abteilungsleiter, Abteilungsleiterin and the other Leiter/Leiterin titles were not mapped to a band in the research behind this reference. Treat them as unmapped. Search all gendered forms and set the band from tenure and team size in your 25-profile sample.

The trap, Prokura: Prokura is a special commercial power of attorney governed by §§ 48-53 HGB (the German Commercial Code). It is entered in the Handelsregister (commercial register). The weaker Handlungsvollmacht is not registered, and the business owner can set its scope. A Prokurist signs with "ppa." (per procura autoritate) under § 51 HGB. None of this tells you rank. A Prokurist can be a mid-level finance lead. Look for a separate functional title.

Japan

Ladder, top to bottom:

  1. 部長 (Buchou): department chief
  2. 次長 (Jicho): deputy department head, the step between kacho and bucho
  3. 課長 (Kachou): section chief
  4. 係長 (Kakarichou): subsection chief
  5. 主任 (Shunin): below the subsection chief

What the title signals: These ranks follow span of control. No published government source ties them to years of service, so a year range here has to come from your own sample, not from the rank.

The traps: A 部長 is not necessarily a board member. Torishimariyaku is a statutory title to flag, not to use as a level. The English-versus-Japanese split has not been measured, and how platforms match kanji against romaji has not been published. Search both forms, and include both romanizations you will see (kachou and kacho).

Korea

Rungs, bottom to top: sawon, daeri, gwajang (과장), chajang, bujang (부장). One guide describes 부장 (bujang) as a general manager or senior director, with roughly ten or more years at large companies.

RungKorea Herald (years in grade)Seoulstart (total years)koreapov (typical years)
Daeri42-44-8
Gwajang57+8-12
Chajang5+4-5 after gwajang12-15
Bujang4-510+15+

What the title signals: Korean rank has followed time served more than scope. Samsung described years-in-grade for each rung, and at Samsung it usually took almost 20 years for a new employee to become an executive. Samsung later dropped minimum years-in-grade requirements of roughly eight to ten years for some promotions. Rank is therefore a weaker years proxy than it used to be, and how much weaker varies by employer.

The traps: Rank and role are separate. A chajang can serve as timjang (team lead) over a team that includes a bujang. Some firms use "pro" or the "-nim" suffix instead of ranks, so the rank never appears on the profile. Search 팀장 (team lead) separately when you need span of control.

Brazil

Forms to OR: analista júnior, analista pleno, analista sênior; desenvolvedor and desenvolvedora with the same three levels. Also include the English Software Engineer/Developer, which accounts for 24,222 profiles in Refolk's index. The Portuguese side was not counted.

SourceJúniorPlenoSênior
Zendesk BRup to 56-910+
iFood Benefíciosup to 55-1010+
Mackenzieup to 22-55+
Tera (data analyst)up to 22-55+
salario.com.br2-44-66-10+

What the title signals: Each employer defines "pleno" for itself. There is no national classification, and the published ranges barely overlap. Glassdoor Brasil's Analista Pleno figures rest on 53,738 submitted salaries, with self-reported experience from 1-3 years up to 7-9 years. Use the word to find the pool and tenure to set the band.

India

Forms to OR: AGM, DGM, and the spelled-out Assistant General Manager and Deputy General Manager.

What the title signals: Depends on the employer type. IBBI deputation notices for public-sector officers require at least 11 years in Officer Grade for DGM and at least 8 years for AGM. NeGD lists AGM at 2-5 years of experience. What the private-sector Executive and AM grades mean has not been published. Check whether the employer is public or private before you trust the grade.

France

Forms to OR: ingénieur, ingénieure, ingénieur diplômé, and the unaccented forms for X-ray.

What the title signals: A degree, not a level. Only the title "ingénieur diplômé" is recognised and protected by law, dating from the law of 10 July 1934. It confers the grade de master, and misusing it carries a fine of up to €15,000 and up to one year in prison. The job of engineer, however, is not a regulated profession in France, so "ingénieur" on its own is just the name of a job. For questions about a particular school's title, start at the Commission des titres d'ingénieur (CTI). To set level, use years since graduation.

UK Director and bank VP

The UK Companies Act meaning of "Director" and the banking use of "VP" were on the original list for this reference, but neither was verified in the research behind it, and how often they appear on individual-contributor profiles has not been published. I give no band mapping. If you are sourcing in those markets, treat both as unmapped: read the functional title, the size of the team, and tenure in your 25-profile sample before you decide what level either word means.

Which titles state level and which state status?

Titles in these markets record one of three things: span of control, time served, or a legal power or degree. Only span of control is a level signal you can search on directly. The other two need a second filter before they can set a band.

What a local title actually tells you

Strong signal of spanWeak signal of span
Span ranks, such as Japan's 部長 and 課長
Search the rank, then set years with a tenure filter
Anchor titles that prove both
None in this reference qualify; prove one with a 25-profile sample before relying on it
Status and employer-defined titles, such as Prokurist, ingénieur, pleno
Flag them and band on functional title and tenure
Tenure ranks, such as Korea's gwajang and India's public-sector AGM
Use as a years proxy, then find 팀장 or the team role for span
Weak signal of yearsStrong signal of years
Most local titles are strong on one axis at best, so the second axis always comes from a filter or a sample, never from the word itself.

The bottom-left quadrant causes the most expensive mistakes because the titles sound senior. Prokura is registered and its scope cannot be limited toward outsiders. Ingénieur diplômé is a master's-level degree. Both are real and both are verifiable, and neither tells you whether the person manages anyone.

A title that proves a legal power or a degree says nothing about how many people report to its holder.

Once the band is set in years and the variants are listed, the search itself should be one sentence. This is the German software row from above, written as a plain-English request:

How do platforms and Google treat accents, script and location aliases?

Practitioner sources report that platform profile search ignores accents while a Google X-ray matches them literally, so the same string returns different pools in each tool. Location strings for the same city also come in several forms, and each alias you leave out drops part of the count.

I found no official platform documentation on this. What practitioners report is consistent: LinkedIn ignores accented characters and Google can search them. The practical result:

  • Platform search is forgiving. "Sao Paulo" and "São Paulo" should return the same people, which suits a first pass.
  • Google X-ray is strict. An X-ray with only the accented form, or only the unaccented one, returns fewer results than it should. Include both.
  • Kanji against romaji is unknown. How platforms match 課長 against kachou or kacho has not been published. Run each form on its own, compare counts, and keep every form that adds people.

Location aliases split counts the same way. In Refolk's index samples, Brazil profiles appeared as "São Paulo, São Paulo, Brazil" (3) and "São Paulo, Brazil" (2). Germany showed "Berlin, Berlin, Germany" (5) and "Berlin Metropolitan Area" (4). In both samples, a search on one alias would have missed close to half of the matching profiles.

For deeper help with non-Latin and diacritic searching in general, the University of Michigan Library's guide to non-English searching is a useful reference.

Step-by-step: building a first-run search in a new market

The procedure turns a US req into a local search in seven steps, about two and a half hours on a first market and less on any market you have already done. No source documents a full workflow, so the timings in each step are my estimates and have not been measured.

First-run local-title search

  1. Translate the brief into local market terms
    Rewrite the intake brief using the titles people in the market actually use, starting from the market row above (about 30 minutes). Done when you have a list of local titles at the right level.
  2. Map the US level to the local ladder
    Convert the req to a year range and span of control, then read across the Japanese or Korean rungs, Brazil's júnior, pleno and sênior, or India's grades (about 20 minutes). Done when the band is written as a year range, with a note where sources disagree.
  3. Build every title variant
    OR the English, native-language, compound, hyphenated, gendered, native-script, romanized and accented forms (about 30 minutes). Done when every layer of the title has at least one form in the string.
  4. Strip or flag statutory titles
    Mark Prokurist, ingénieur diplômé, torishimariyaku and similar titles so they cannot set the band (about 10 minutes). Done when each is flagged and paired with a functional-title check.
  5. Normalise location strings
    Collect every alias of the target city from a broad first run and OR them (about 10 minutes). Done when every alias you saw is in the search.
  6. Run the search and sample 25 current titles
    Read 25 current titles and count the off-target ones (about 30 minutes). Done when the off-target count is under the threshold you set beforehand.
  7. X-ray Google for accented and native-script forms
    Run a Google X-ray with the accented, unaccented, native-script and romanized forms and compare it with the platform results (about 20 minutes). Done when any forms that were missing have been added to the main string.

Set the off-target threshold before step 6, not after. If you set it after you see the results, it tends to drift to whatever the results happen to be.

First-run search strings (Germany, Brazil, Japan)
PLATFORM TITLE CLAUSE - Germany, software
("Software Engineer" OR "Software Developer" OR Softwareentwickler OR Softwareentwicklerin OR "Software-Entwickler" OR "Software-Entwicklerin")

PLATFORM LOCATION CLAUSE - Germany
("Berlin, Berlin, Germany" OR "Berlin Metropolitan Area")

CURRENT-TITLE EXCLUSION - apply to current title only
NOT (Founder OR CEO OR CTO)

GOOGLE X-RAY - Brazil, accented and unaccented
site:linkedin.com/in ("São Paulo" OR "Sao Paulo") (desenvolvedor OR desenvolvedora) pleno

GOOGLE X-RAY - Japan, native script and both romanizations
site:linkedin.com/in (課長 OR kachou OR kacho) Tokyo

Swap in your own city aliases and keep every variant; delete a form only after a sample shows it returns nothing.

How local-title searches go wrong

Local-title searches fail in two directions. Some return people whose current title doesn't match (false positives), and some miss whole spellings, scripts or aliases (false negatives). Each failure below has been seen in Refolk's index or documented in the sources, and each comes with a check you can run in minutes.

6-7 of 25
current titles off target in each 25-profile sample
Founders, CxOs and unrelated roles turned up in the Germany English, Germany German and Brazil samples from Refolk's index.
  1. The title match returns people whose current role doesn't fit. About a quarter of each 25-profile sample had founder, CxO or other current titles that did not match the search: 6 of 25 for Germany English, 7 of 25 for Germany German and 6 of 25 for Brazil. A matching title somewhere in the history is not a matching current role. Check: read 25 current titles before you trust the count. This applies in any tool, Refolk included.
  2. Missing the hyphenated form. "Software-Entwickler" outnumbered "Softwareentwickler" 11 to 7 in the German sample. A search that uses only one spelling loses a large share of the pool. Check: OR both spellings together, plus their -in forms.
  3. Treating Prokurist as a level. It is a registered power of attorney, not a rank, and a Prokurist can be a mid-level finance lead. Check: look for a separate functional title.
  4. Reading "Ingénieur" as seniority. The job of engineer is not a regulated profession in France, and the protected title shows a degree, not a level. Check: years since graduation.
  5. Confusing Korean rank with role. A chajang can be timjang over a team that includes a bujang. Firms that use "pro" or "-nim" never show a rank at all. Check: search for 팀장 (team lead) separately.
  6. Collapsing the Brazil pleno band. Published ranges run from 2-5 years to 6-9 years. A search on "pleno" alone mixes people roughly four years apart in experience. Check: filter on tenure, not on the word.
  7. Dropping accents or aliases. The platform reportedly ignores accents but Google does not, so an X-ray without the accented form returns too little. Location aliases split counts as well, as with "São Paulo, Brazil" versus "São Paulo, São Paulo, Brazil". Check: use both accent forms in every X-ray and every alias in every location clause.
  8. Assuming one meaning for Indian AGM and DGM. The year floor for AGM is 8+ at IBBI but 2-5 at NeGD. Check: find out whether the employer is public or private before you band the person.

Failures 1, 2 and 7 hide the most, because the search still returns a plausible number of results. A count that looks healthy is not evidence that the search is right. Only the 25-profile sample and the X-ray comparison show you what is missing or wrong. Running the search as a single plain-English request in Refolk removes the work of assembling the string, but the sample check still applies.

Checking the result and keeping the market rows current

A local-title search is finished when the band is written in years, every variant and alias is in the string, statutory titles are flagged, and a 25-profile sample comes in under your threshold. Market rows go stale as employers change their ladders, so record what you measured and when, and re-run the count when you reopen a req in that market.

Before calling the search done

  • Every title form in the market row is ORed together, including the hyphenated and feminine forms.
  • The target band is written as a year range and a span of control, not as a local rank word.
  • English and native-language forms are both in the search, unless the split for this role and market has been measured.
  • Prokurist, ingénieur diplômé, torishimariyaku and similar titles are flagged and paired with a functional-title check.
  • Every location alias seen in a broad first run is in the location clause.
  • 25 current titles have been read and the off-target count is under the threshold set before the run.
  • A Google X-ray with accented, unaccented, native-script and romanized forms has been run and compared with the platform results.
  • The English-versus-local count and the sample result are recorded with the date measured.

Some rows will change, and you can recheck them as follows:

  • English-versus-local split: re-count each form separately with the same location. The German split is a measured figure, so it can drift as employers change. Record a new split each time you measure.
  • Korean years-to-rank: Samsung has already removed years-in-grade floors for some promotions. If a new sample shows rank and tenure diverging, trust the tenure.
  • Brazil pleno: sources disagree today, and new salary surveys can add yet another range. Keep the table's spread in mind, and set your band from the employers you are actually targeting.
  • Unmapped rows: for Abteilungsleiter, the UK Director, bank VP and Indian private-sector grades, your own dated sample is the best evidence available. Write it down.
Market row record
Market:
Role family:
US req band (years / span of control):
English forms searched:
Native-language forms searched (compound and hyphenated):
Gendered and inclusive forms searched:
Native-script, romanized and accent variants:
Location aliases:
English count / local count / date measured:
Statutory or degree titles flagged:
Local band sources and their year ranges:
25-profile sample: on-target / off-target / threshold:
Trap seen in this market:

Fill one per market and role family, date it, and re-run the counts when you reopen a req there.

Questions practitioners ask

What does analista pleno mean in Brazil, in years of experience?

Analista pleno means mid-level, but no national standard fixes the years, so the band depends on the employer. Published ranges run from 2-5 years (Mackenzie, Tera) through 4-6 (salario.com.br) to 6-9 (Zendesk BR). Glassdoor Brasil's 53,738 self-reported salaries span 1-3 to 7-9 years of experience. Search on the word to find the pool, then filter on the tenure your req actually needs.

What are bucho and kacho in English?

部長 (bucho) is a department chief and 課長 (kacho) is a section chief. 次長 (jicho) is the deputy department head between them, 係長 (kakaricho) is a subsection chief below, and 主任 (shunin) sits below that. The ranks describe span of control, and no government source ties them to years of service. A 部長 is not necessarily a board member. Search the kanji and romanized forms together.

Is a Prokurist a senior person in a German company?

Not necessarily. Prokura is a commercial power of attorney under §§ 48-53 HGB. It is entered in the Handelsregister, and a Prokurist signs with "ppa." It proves legal authority to act for the company, not rank or span of control, and a Prokurist can be a mid-level finance lead. Look for a separate functional title and tenure before you place the person in a band.

Should I search German profiles in English or German?

Search both, every time. In Refolk's index about 72% of German software profiles carry the English title and 28% carry the German one. A German-only search misses most of the pool. An English-only search drops about 14,561 profiles that skew toward traditional Mittelstand employers. OR the English forms with the compound and hyphenated German forms and their -in variants.

Does LinkedIn search match accented characters?

Practitioner sources say LinkedIn ignores accents and Google matches them literally. I found no official platform documentation for either behavior. In practice the platform search is forgiving and a Google X-ray is strict, so an X-ray for "São Paulo" alone returns too little. Put the accented and unaccented forms in every X-ray. Test kanji against romanized forms yourself, because how platforms match them has not been published.

How senior is an AGM or DGM in India?

It depends on the employer type. In IBBI's public-sector deputation notices, a DGM needs at least 11 years in Officer Grade and an AGM at least 8. NeGD lists AGM at 2-5 years of experience. What these grades mean in the private sector has not been published. Check whether the employer is public or private, then band on tenure instead of the grade letters.

Try it on the search you came here for

Stop building boolean strings. Just describe the person.

Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.

  1. 01Describe them

    One plain sentence. Role, city, stack, stage, whatever matters to you.

  2. 02I read the web live

    GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.

  3. 03You read the shortlist

    Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.

  • No boolean, no filters, no seat to buy. One box.
  • Read at search time, so a profile updated yesterday counts today.
  • Every step visible as it runs, every name with its reason.

500 free credits on sign-up. No card, no demo call. See real searches.

Read next