The moment the manual and the product disagree, the user stops trusting the product. The manual says “press the POWER button.” The button is molded ON/OFF. The manual says the path is Settings > Wireless. The firmware says Settings > Network > Wireless. The user hunts for a menu that does not exist, then calls support.
None of that is a manual translation or technical document translation problem. The sentences can be flawless and the manual still fails, because the strings it quotes do not match the product. This guide covers the 15 checks that confirm a multilingual manual agrees with the shipping product: its UI, its physical buttons and labels, its on-screen messages, its part numbers, and its firmware version.
What is a UI matching check?
A UI matching check verifies, before release, that the product information quoted in a multilingual manual matches the product that actually ships. The subjects are the firmware version behind every screenshot, the molded text on buttons and ports, the exact strings of on-screen messages and menu paths, safety labels and certification marks, part numbers and barcodes, and document version data. Translation review judges sentences against the source. UI matching judges the document against the final firmware build and the physical product.

Rename one menu in firmware and the screenshots, menu paths, and quoted strings in 50 language editions go stale at once. A translation error stays inside one language. A UI mismatch propagates across every language you ship. That multiplier is why UI matching defects cost more to fix than defects in any other QA area.
- Who this is for Technical writers, manual owners, localization PMs, QA staff, product UI owners, and manufacturers producing manuals with AI tools.
- When to use it After the firmware is locked, right after per-language UI capture, whenever product labels, molded text, or part information change, and before final manual sign-off.
- Use with The companion UI Matching Checklist, which turns these 15 checks into a sign-off form.
This guide covers general checks that apply across industries and product types. Product-class certification, safety labeling, regulatory wording, and firmware change-control requirements need separate review. It does not replace a product specification or a release note.
1. Why UI matching is its own QA area
A translator is responsible for what a sentence means. Whether the button name, menu path, on-screen message, molded text, label, or part number quoted in that sentence agrees with the shipping product is settled by holding the document against the firmware and the physical unit. No amount of language expertise detects the mismatch.
In a multilingual program, one UI change lands on dozens of language editions at once. A single screenshot, one menu name, one molded label out of step produces higher support call volume, an impression of incomplete localization, the wrong manual in the box, and, at worst, a failed safety-critical operation.
2. Without a fixed baseline date, the check means nothing
The most common failure in UI matching is not a missed item. It is comparing artifacts from different dates. If the manuscript is written against the March build, the screenshots were captured from the April build, and the molded-text drawing is the February revision, then all 15 checks can pass and the manual still will not match what ships.
Before you start, pin five artifacts to a single baseline date. This is the precondition for the entire check.
| Reference artifact | What has to be locked | What it adjudicates |
|---|---|---|
| Final firmware build | Build number and build date | The source of every screenshot, menu path, and on-screen message |
| Per-language UI string resources | Firmware string files (.strings, .xml, .json) | The reference for character-level comparison of quoted text |
| Molded text and label drawings | Physical drawing or final artwork | The reference for button and port names, safety labels, certification marks |
| Part number and barcode list | Per-market box contents table | The reference for accessory lists and QR and barcode data |
| Manual manuscript | Document version and publication date | The artifact being checked against the four above |
Operating principle. Do not leave the reference artifacts scattered across departments. Keep the final firmware version, the per-language UI strings, the label drawings, the part number list, and the manual manuscript inside the same release package. That is what holds the baseline date together.
3. The five areas product and document synchronization has to cover
| Area | What has to be synchronized |
|---|---|
| Screenshots ↔ final UI | Shipping-build capture, per-language capture, how an English-only UI is disclosed, personal data scrubbed from screenshots |
| Molded text and physical labels | Button and port names, dual-marking convention where the molded text is fixed English, safety labels and certification marks |
| On-screen text ↔ quoted text | Character-level agreement of alerts, error codes, and menu paths, a shared firmware-manual glossary, and a consistent quoting convention |
| Part numbers and barcodes | Per-market box contents, part numbers, barcode and QR data verified with an actual scanner |
| Version and firmware sync | Document version, publication date, and a single tracked record of UI, molded text, and document changes |
4. The five categories of the 15 checks
| Category | What it covers | Checks |
|---|---|---|
| A. Screenshots ↔ final UI | Shipping build, per-language UI, English-UI disclosure, personal data in screenshots | 4 |
| B. Molded text and physical labels | Button, switch, and port names, dual-marking, safety labels, and certification marks | 4 |
| C. On-screen text ↔ quoted text | Alerts and menu paths, firmware strings, shared glossary, quoting convention | 4 |
| D. Part numbers and barcodes | Per-market box contents, part numbers, barcode and QR verification | 2 |
| E. Version and firmware sync | Change-control record for UI, molded text, and document changes | 1 |
Severity distribution: Critical (A) 8 · Major (B) 6 · Minor (C) 1
5. Severity grades and the sign-off rule
| Grade | Name | Definition and consequence if missed |
|---|---|---|
| A | Critical / mandatory | Leads directly to a failed user operation, inconsistent safety information, incorrect certification data, or a launch delay. Must be zero at sign-off. |
| B | Major / recommended | A miss produces customer claims, higher support call volume, rework, and erosion of brand trust. Target zero wherever possible. |
| C | Minor / supporting | Low impact individually. Accumulated, it drives quality variance between language editions and raises operating cost. |
6. The 15 checks in detail
The table below expands each checklist line into working guidance. In production, record every item twice: once by the document owner (Self), once by an independent reviewer (Reviewer).
| No | Check and acceptance criterion | Grade |
|---|---|---|
| 1 | A. Screenshots ↔ final UI Was every screenshot in the manual captured from the shipping firmware? Look for captures left over from interim development builds or earlier versions. Comparing screenshot capture dates against the shipping build date surfaces stale captures quickly. | A |
| 2 | A. Screenshots ↔ final UI Is the text inside each screenshot captured from that language’s UI build? Confirm that English UI screenshots have not been reused in non-English manuals. Capture from the per-language build. | A |
| 3 | A. Screenshots ↔ final UI If the product has no multilingual UI, does the manual state that the UI is displayed in English? Confirm the manual says so plainly, so readers do not assume they can change the display language. The wording and its placement stay consistent across every language edition. | B |
| 4 | A. Screenshots ↔ final UI Has personal and test data inside screenshots been scrubbed? Prevent exposure of real user names, email addresses, phone numbers, account IDs, and location data. Confirm dummy data was used. | B |
| 5 | B. Molded text and physical labels Do the button, switch, and port names in the manual match the molded text on the product exactly? Catch mismatches such as the manual saying POWER while the product is molded ON/OFF. Check icon markings the same way. | A |
| 6 | B. Molded text and physical labels Is the “product marking plus local-language gloss” convention applied consistently? Where molded text is fixed English or a symbol and a gloss is needed, confirm the convention, for example POWER (Encendido), is uniform throughout the document. | C |
| 7 | B. Molded text and physical labels Do the position, wording, and pictograms of warning and safety labels match the manual’s safety section? Confirm that pictograms and signal words under ANSI Z535.6, ISO 3864, and ISO 7010 are synchronized between the physical label and the manual. | A |
| 8 | B. Molded text and physical labels Do the certifications the product actually holds match the manual’s certification page? Confirm the mark types, certification numbers (including the FCC ID where applicable), and marking format agree. Reconcile against the approved mark list from Part 3 (Regulatory). | A |
| 9 | C. On-screen text ↔ quoted text Do alerts and menu paths match the firmware display character for character? Once the firmware strings are locked, compare every quoted string in the manual. Taking the firmware string resource files directly lets you verify down to words, symbols, and spaces. | A |
| 10 | C. On-screen text ↔ quoted text Are on-screen messages and the manual’s quotations written from the same translation basis? Confirm a single shared glossary governs both, so firmware translation and manual translation do not render the same message two different ways. | A |
| 11 | C. On-screen text ↔ quoted text Do menu path notations (for example, Settings > Network > Connect) match the per-language UI menu names? Confirm the per-language menu translation matches the manual, and that the path separator is used consistently. | A |
| 12 | C. On-screen text ↔ quoted text Is the quoting convention for buttons and menus (quotation marks, bold, brackets) consistent? Confirm the same convention is applied throughout the manual. | B |
| 13 | D. Part numbers and barcodes Do the part numbers in the accessory list match what is actually in the box? Confirm per-market differences in box contents are reflected, and that the manual describes the regional configuration accurately. | B |
| 14 | D. Part numbers and barcodes Do the part numbers, barcodes, and QR data agree between product and manual? Verify by scanning with an actual scanner and confirming the same data is returned. Visual comparison does not detect these errors. | B |
| 15 | E. Version and firmware sync Is there a change-control record covering UI, molded text, and manual changes? Confirm the product-to-document change history is traceable in a single system. Record the requester, the date applied, and the list of affected language editions. | B |
7. Five errors found most often at incoming inspection
- v2.3 captures survive into the v2.5 release If it ships The user hunts for a menu that is not on screen. Support call volume rises, and the affected pages have to be recaptured across every language edition.
How to block it Compare screenshot capture dates against the shipping build date. Adopt a rule that a firmware build change invalidates the entire capture set. - The manual says POWER, the product is molded ON/OFF If it ships The user cannot find the button they were told to press. When it appears in a safety-critical instruction, the operation fails.
How to block it Take the molded-text drawing (physical drawing or final artwork) as the single source, and reconcile every button and port reference in the manual against it. - Firmware and manual translate the same message differently If it ships The reader takes one item for two different functions. When it happens in an error message, troubleshooting becomes impossible.
How to block it Apply one shared glossary to firmware translation and manual translation. Fix the UI strings as the upstream source for the manual. - English UI screens survive in non-English manuals If it ships The localization reads as unfinished, and non-English users lose confidence in the brand.
How to block it Run a full screenshot inventory to confirm each was captured from the per-language build. If the product has no multilingual UI, say so explicitly in the text. - Per-market model variants are not mapped to manual variants If it ships Distribution receives the wrong stock, and the wrong manual ends up in the box.
How to block it Map the per-market box contents table to the manual variants one-to-one, and verify part numbers and barcodes with an actual scanner.
8. Error or correct: three worked examples
| Screenshot version | |
|---|---|
| ✗ Error | Captures from the v2.3 development build are still sitting in the v2.5 release manual. |
| ✓ Correct | Once firmware is locked, recapture from the shipping build, per language. A build change invalidates the existing captures. |
| Molded product text | |
|---|---|
| ✗ Error | The manual says POWER while the product is molded ON/OFF. |
| ✓ Correct | Take the molded-text drawing as the single source and synchronize the manual to it. Where a gloss is needed, apply the dual-marking convention consistently. |
| Menu path | |
|---|---|
| ✗ Error | The manual says Settings > Wireless. The actual UI is Settings > Network > Wireless. |
| ✓ Correct | Take the firmware string resources and compare the per-language menu paths against the manual’s quotations character for character, then update. |
9. Running it as a two-stage cross-check
UI matching has to be revisited whenever firmware moves. Final sign-off, though, happens once, against a single baseline date fixed just before release. If the product and the document sit on different baseline dates, every check can pass and the manual can still miss the shipping build.
| Stage | Who performs it | What they do | Sign-off criterion |
|---|---|---|---|
| Stage 1: author self-check | The technical writer or manual owner | Assemble the final firmware build, per-language UI captures, molded-text and label drawings, part number list, and change history on one baseline date, then walk the 15 checks. | Grade A at zero. Baseline date and build number recorded. |
| Stage 2: independent cross-check | A reviewer outside the manual work, plus firmware/UI and product label owners | Re-verify screenshots, menu paths, molded text, labels, part numbers, and the change record against the physical product or the final build. | Approve for release, or reject. A rejected file is corrected and re-verified. |
10. How a manufacturer documentation team can adopt this
- Lock the firmware build number and the baseline date first. Every comparison that follows sits on that one date.
- Get the per-language UI string resource files directly from the firmware team. Transcribing from screenshots cannot guarantee character-level agreement.
- Obtain the molded-text drawing (physical drawing or final artwork) as a single source, and reconcile every button and port reference in the manual against it.
- Recapture screenshots from the shipping build, per language. Adopt a rule that a build change invalidates the entire existing capture set.
- Verify part numbers and barcodes with an actual scanner. Visual comparison does not work for this item.
- Record the firmware build number and the baseline date alongside the check results. When firmware updates later, that record tells you which items have to be rechecked.
Frequently asked questions
- How is UI matching different from translation review? The subject and the reference differ. Translation review judges meaning and expression against the source text. UI matching judges the product information the manual quotes against the final firmware and the physical unit. A perfectly translated sentence that quotes “Settings > Wireless” passes translation review and fails the product, if the actual path is “Settings > Network > Wireless.”
- Can we use captures taken before the firmware is locked? For draft review, yes. For the shipping manual, recapture against the final firmware. Screens from an interim development build put the user in front of a UI the product does not have. Set the rule in advance: a build change invalidates the existing captures.
- Is checking menu paths during translation review enough? A menu path is not a translation quality question. It is a synchronization question between the firmware strings and the document’s quotations. However accurate the translation, one extra level in the firmware menu hierarchy makes the quotation wrong. Compare the per-language UI strings against the manual text character by character.
- If the product UI is English-only, do we still need per-language screenshots? If the product has no multilingual UI, per-language capture is not possible. What the manual owes the reader instead is a clear statement that the UI is displayed in English. Keep that wording and its placement consistent across every language edition.
- Who should own UI matching sign-off? This is not something the documentation team can close alone. The safest arrangement puts the documentation team, the localization PM, the firmware and UI owners, and the product label and certification owners in front of the same baseline-dated artifacts. Working alone, the documentation team has no way to verify that its own reference material is current.
Pre-Launch QA Series for Multilingual Manuals (8 parts)
The series follows a multilingual manual from translation intake to market release, one gate at a time. Each part pairs a guide with a self-check checklist.
- Part 1 · Translation integrity Catching defects in translated files before editing
- Part 2 · DTP Multilingual typesetting and layout integrity
- Part 3 · Regulatory Market-specific regulatory and standards marking
- Part 4 · UI matching Synchronizing product screens with the document (this guide)
- Part 5 · Accessibility EAA, WCAG, PDF/UA, and Section 508 conformance
- Part 6 · Multi-channel output Consistency across PDF, HTML, and mobile
- Part 7 · Print Bleed, CMYK, font embedding, and print readiness
- Part 8 · Data management Turning source files and TM into reusable assets