OnlineDocTranslatorGuides

Running your practice

Localisation Project: How Translators and Agencies Scope, Run, and Deliver One

A localisation project is a single multi-locale job that converts source content into reviewed, QA-passed target files on a fixed scope, SLA, and budget.

Inside a localization program, work ships one job at a time. Each job carries its own scope sheet, its own translator panel, its own SLA clock, and its own certification gate. That job is the localisation project — the atomic unit of delivery a translator or LPM actually runs.

What we offer

What is a localisation project?

A localisation project is a single bounded job inside a localisation program that converts a defined source scope into reviewed, QA-passed target-language files for one or more locales under a fixed SLA, cost base, and acceptance checklist signed off by a localization project manager. The project is the delivery unit; the program is the standing operation above it.

What are the key elements of a localisation project?

A localisation project has five scoping elements the translator inspects before quoting: the extracted resource bundle, the project translation memory, the client term base, the locale style guide, and the pseudo-localisation or staging build used for in-context review.

ArtifactPurposeFormat
Resource bundleSegments, placeholders, limitsXLIFF, JSON, .strings, strings.xml
TM leverage reportVolume by match bandCSV or TMS export
Term baseApproved and forbidden termsTBX
Locale style guideRegister, tone, format rulesPDF or wiki
Pseudo-loc or staging buildIn-context reviewAPK, URL, build link

What goes in the resource bundle and locale pair list?

The resource bundle lists segment count, placeholder density, character limits per key, and context fields across formats such as XLIFF, JSON, .strings, and strings.xml. Locale pairs are declared as BCP 47 tags combining ISO 639-1 and ISO 3166-1, so es-MX and es-ES generate separate quote lines.

  • es-MX ≠ es-ES — separate TMs, separate term bases.
  • pt-BR ≠ pt-PT — separate reviewers.
  • fr-CA ≠ fr-FR — separate style guides.

JSON and XLIFF handling for front-end frameworks is covered in React i18n: How Translators Fit Into a React Localization Pipeline.

What TM leverage report does a localisation project need?

The TM report breaks volume into ICE, 100%, 95–99%, 85–94%, and no-match bands, which drive the per-word discount schedule and set realistic throughput for the LPM's schedule.

  1. ICE matches: in-context, zero edit expected.
  2. 100% matches: segment-level identical.
  3. 95–99% fuzzy: light edit.
  4. 85–94% fuzzy: moderate edit.
  5. No match: full translation rate.

What does a usable localisation brief include?

A usable brief lists target locale, tone register, forbidden terms, brand voice reference, staging URL, character limits per key, and milestone deadlines — not just an EN-to-ES language pair. A website brief, for example, names the staging URL and in-page character limits; see Website Localization: What Translators and Agencies Actually Do on the Project for a non-app lane example.

What we offer

What are the core phases of a localisation project?

A localisation project runs through six core phases: scope and kickoff, file preparation and TM leverage, translation, bilingual review, LQA sampling, and delivery with certification sign-off where the lane requires one.

Who runs a localisation project and what do they each do?

A localisation project runs on five roles: the localization project manager coordinates the job, the translator produces target segments, the reviewer checks every segment, the LQA tester samples against a scorecard, and the localization engineer handles file prep and build integration. Each role is a distinct cost line and a distinct quality gate — removing any one of them creates a gap that the others cannot cover.

RoleProducesBilled as
LPMScope, schedule, delivery10–15% of project value
TranslatorTarget segments$0.10–$0.25 per source word
ReviewerEdited segments$0.04–$0.08 per source word
LQA testerScorecard with error typing$40–$90 per 1,000 words
Localization engineerFile prep, pseudo-loc, build integrationHourly or amortized

What does the localization project manager (LPM) own on a single job?

The LPM is the single point of contact for the client across the entire job — clients deal with one coordinator, not thirty linguists or engineers. On any given project, the LPM confirms scope, assigns linguists, dispatches through the TMS, tracks SLA, runs pre-delivery QA, and handles exceptions. A working LPM typically carries 8–15 concurrent projects at once, which means exception-handling speed and clear communication discipline are non-negotiable skills at that desk. The LPM's fee is priced at 10–15% of total project value, reflecting the coordination overhead that makes every other role more efficient.

What do translators, reviewers, and LQA testers each produce?

Translators deliver target-language segments against the active TM and term base; every segment they produce is constrained by pre-approved terminology so that forbidden terms hard-fail at the editor level before review even begins. Reviewers then edit every segment for accuracy and locale fit — not a sample, every segment — producing a corrected file that feeds directly into delivery. LQA testers work last, sampling a defined slice of translated content against an MQM-style scorecard that buckets errors into accuracy, fluency, terminology, locale convention, and style categories. Each tester's output is a scored artifact, not an opinion, and that scorecard travels with the delivery package as evidence of quality. Being connected across all three of these functions is precisely what gives a strong LPM the visibility to catch problems before they reach the client.

For role depth beyond the project unit, see Localization Management: How Translators and Agencies Run Multi-Locale Programs.

Pricing

How much does a localisation project cost and how is it budgeted?

A localisation project is budgeted on four stacked line items in US dollars: translation at $0.10–$0.25 per source word, review at $0.04–$0.08 per word, LQA at $40–$90 per 1,000 words, and project management at 10–15% of project value, plus tooling amortized per project.

Line itemRateBasis
Translation$0.10–$0.25per source word
Review$0.04–$0.08per source word
LQA$40–$90per 1,000 words
Project management10–15%of project value
Certificate of accuracy$25–$150per 250-word page (certified lanes)
Notarization$10–$50per document, optional

What does a mid-size multi-locale localisation project budget look like?

A 5-language, 50,000-word localisation project budgets at $25,000–$75,000 end to end, depending on TM leverage, turnaround speed, and whether the lane requires a certificate of accuracy. The planning layer above this budget is covered in Localisation Strategy: How Agencies Plan, Price, and Deliver Multi-Locale Programs.

How does AI pre-translation shift the project cost mix?

AI pre-translate drops the translation line and shifts spend into review at $0.04–$0.08 per word and LQA at $40–$90 per 1,000 words, which carry more of the quality burden once raw output is machine-generated. The scorecard gate does not move; the sampling density often tightens.

Get a quote

How do you evaluate whether a localisation project is good or janky?

A localisation project is evaluated against an MQM-style LQA scorecard priced at $40–$90 per 1,000 words, covering accuracy, fluency, terminology, locale convention, and style. A 'janky' output is one that fails locale convention and terminology first — those two categories are the ones end users notice immediately, even when they are comfortable reading the source language alongside the translation. A polished output passes all five categories with zero critical errors and carries a scorecard artifact as proof.

Error categoryJanky signalPolished signal
AccuracyDropped clauses, wrong numbersSegment-level fidelity
FluencySource-language word orderNative reading rhythm
TerminologyTerm base missesZero forbidden-term hits
Locale conventionWrong date, currency, addressLocale-correct formats
StyleRegister driftStyle guide alignment

Which error categories most often make a localisation project feel janky?

Locale convention errors — wrong date format, incorrect currency symbol, misformatted address — and terminology slips against the client term base are the two categories end users notice first, even when they also read source-language content comfortably. Accuracy errors matter enormously for certified or legal content, but for general product copy it is the locale convention failures that break trust fastest, because they signal that the content was never adapted for the target market — only translated. Fluency and style errors read as awkward rather than wrong, so they erode confidence more slowly; terminology and locale convention errors read as factually incorrect, so they erode it immediately.

A project that has been run well leaves a clear audit trail: the LQA scorecard is aligned to ISO 17100 quality standards, every critical error has a resolution note, and the final file matches the locale style guide on date, number, and address formatting. A project that has been run badly lacks that artifact chain — the scorecard is missing, the term base was never loaded, or the locale style guide was never issued at kickoff. Evaluating a project's quality is therefore as much about inspecting the process artifacts as it is about reading the translated text itself.

How does a translator break into running localisation projects as a PM?

A translator moves into localisation project management by stacking three capabilities on top of linguistic skill: CAT tool and TMS fluency, TM and term base governance, and vendor and budget tracking. The transition is achievable because translators already own the quality literacy that makes a strong LPM — they understand what a bad segment looks like, they can read a TM leverage report, and they know why a forbidden-term hit in the editor matters. What they add on the LPM desk is the coordination and commercial layer that sits above the linguistic work.

Which skills transfer from translation to project management?

CAT tool fluency, QA literacy, and style-guide discipline transfer directly from the translation seat to the PM desk — a translator who has worked inside a TMS already understands dispatch, segment status, and pre-delivery QA in a way that a generalist PM has to learn from scratch. Vendor negotiation, PO alignment, concurrent-job scheduling across 8–15 active projects simultaneously, and exception handling are learned on the LPM desk itself, not at the translation seat. The LPM role is defined by its breadth of contact — coordinating across engineering, client, and linguist teams at once — and that cross-team fluency is something translators build naturally by working with reviewers and engineers throughout their linguistic career.

  • Transfers directly: CAT tool fluency, TM leverage reading, term base governance, QA literacy, style guide discipline.
  • Learned on the LPM desk: vendor management, budget tracking, SLA enforcement, concurrent-job scheduling across 8–15 projects, exception handling, scope confirmation, and client-facing communication.

A translator targeting the LPM role should also understand the commercial structure of a project: translation is billed at $0.10–$0.25 per source word, review at $0.04–$0.08 per source word, LQA at $40–$90 per 1,000 words, and PM coordination at 10–15% of total project value. Knowing those numbers is essential for confirming scope, assigning work at the right tier, and tracking budget against delivery without surprises.

Common questions cover scope boundaries, SLA numbers, cost ranges, locale coding, and whether AI reduces the need for a human localisation project at all.

Frequently asked questions

How does localisation work?

Localisation works by systematically adapting source content — text, visuals, formats, and functionality — so it feels native to a target locale, not just translated. For a certified translation project specifically, this means the linguist or agency scopes the source document, assigns a qualified translator for the target language pair, produces a draft (increasingly via CAT tools or AI-assisted first-draft platforms), edits and proofreads against quality standards (ISO 17100 is the dominant benchmark), and attaches a signed certificate of accuracy before delivery. Turnaround for a certified document typically runs 24–72 hours depending on volume and language pair.

What does a localization project manager do?

A localisation project manager (LPM) coordinates every moving part of a translation or localisation project from intake to delivery. Day-to-day tasks include scoping source files, calculating word counts, assigning linguists and reviewers, setting deadlines, managing translation memories and termbases, tracking costs against quotes, and performing final quality checks before client handoff. On certified translation projects, the LPM also ensures the correct credentials and certification statements are in place for each target language. They act as the communication bridge between the client, the translator, and any DTP or desktop publishing specialists.

What is the average salary for a localization project manager in the US?

Entry-to-mid-level localisation project managers in the US typically earn between $45,000 and $75,000 per year, while senior or programme-manager roles at larger LSPs or tech companies can reach $90,000–$120,000+. Salary varies significantly by industry vertical (legal and medical tend to pay more), company size, and whether the role is in-house or agency-side. Certified translation agencies often staff LPMs at the lower end of that range, while global SaaS or life-sciences firms sit at the higher end.

What tools or workflows do localization PMs actually use?

Most localisation PMs rely on a combination of a translation management system (TMS), CAT tools, and project tracking software. Common TMS platforms include Phrase, Memsource, and XTM; CAT tools like SDL Trados, memoQ, and Wordfast handle segment-level translation with translation memory leverage. For certified translation workflows, AI-assisted first-draft platforms are increasingly used to accelerate the initial draft before a credentialed translator reviews and certifies. PMs also use standard PM tools — spreadsheets, Trello, Asana, or Jira — for deadline and resource tracking, plus QA tools like Xbench or Verifika for automated checks.

How do you evaluate whether a localization is good or bad?

A good localisation reads as if it were written natively in the target language — terminology is consistent, tone matches the register of the source, and locale-specific formats (dates, currencies, addresses) are correct. For certified translations, evaluators also check that meaning is complete and accurate against the source, with nothing omitted or added. Structured quality frameworks like ISO 17100 or the MQM (Multidimensional Quality Metrics) error taxonomy provide objective scoring across dimensions such as fluency, accuracy, and terminology. Practical red flags include calques (word-for-word constructions), untranslated strings, and mismatched legal terminology.

What makes a localization feel 'janky' vs polished?

A janky localisation feels like it was produced by someone who translated words rather than meaning — the result is awkward phrasing, inconsistent terminology, and cultural references that don't land. Common culprits include literal translation of idioms, failure to adapt number or date formats, ignoring text-expansion (German or Finnish strings can run 30–40% longer than English, breaking UI layouts), and mismatched formality levels. A polished localisation is invisible: the reader never senses a source document existed. On certified translations specifically, jank often shows up as inconsistent legal terms or missing obligatory phrasing required by the target jurisdiction.

How do translation jobs work?

Translation jobs are typically scoped by word count, language pair, subject matter, and required turnaround before being quoted and assigned to a qualified linguist. Agencies or in-house teams intake source files, run them through a CAT tool to calculate leveraged word counts (fuzzy matches and repetitions reduce the billable volume), assign a translator and reviewer, and deliver a final file plus — for certified work — a signed certificate of accuracy. Freelancers may work directly with end clients or through an LSP. Rates for certified translation commonly range from $0.10 to $0.25 per source word depending on language pair and specialisation.

Can you give me an example of localization?

A certified birth certificate translated from Spanish to English for a US immigration application is a clear example of localisation in a translation context. The translator doesn't just convert words — they render official names, dates in MM/DD/YYYY format, and institutional titles in a way that a USCIS officer will recognise, then attach a certification statement attesting to the translation's accuracy and completeness. Another example is adapting a software UI from US English to UK English, adjusting spellings, date formats, and currency symbols so the product feels locally native.

What is localisation in software?

Software localisation is the process of adapting a software product's interface, help content, and supporting documentation so it functions and feels native in a specific locale. This goes beyond translating UI strings — it includes adapting date/time formats, currency symbols, right-to-left text rendering, character encoding, and culturally appropriate imagery. For translation SaaS products in particular, localisation also means ensuring that any auto-generated certified translation output conforms to the legal and stylistic conventions of the target market, not just the source language.

Related