A localization quality index is a score from 0 to 10 that every localized post earns on the same five checks. You average it per market, per brand and per month. The number tells you which market needs help before your audience tells you.
TLDR: Score each post on five checks worth two points each: meaning, tone, local fit, formatting and links. Set a pass mark of 8, review 10 posts per market each month, and fix the lowest market first.
Without a score, localization quality is just an opinion. Head office thinks the Spanish posts are fine because nobody complained, and the Madrid team knows they sound like a manual. A shared index turns that into a number you can actually discuss.
The five checks and what each one catches
Give each check two points: 2 for clean, 1 for a small problem, 0 for a miss. Keep the questions yes or no so two reviewers score the same post the same way.
| Check | The question | Typical failure |
|---|---|---|
| Meaning | Does it say what the source says, with nothing added or lost? | A claim softened or a number changed in translation |
| Tone | Does it sound like the brand in this language? | Formal where the brand is casual, or the reverse |
| Local fit | Would a local reader recognise the reference, date or example? | A US holiday, a pun that does not travel |
| Formatting | Are hashtags, line breaks, emoji and length right for the platform? | A caption cut off or over the character cap |
| Links | Do the link, UTM and CTA point to the right local page? | The English landing page in a French post |
Local fit is the check that machine output fails most often, and it is the one a native reviewer spots in seconds. Formatting is the one a tool should catch for you. Platform caps are real: the Instagram API rejects captions over 2,200 characters, Instagram caps posts and Reels at 5 hashtags, and X counts 280 weighted characters, so a translation that runs 20% longer can fail at publish time.
Set a pass mark per market, not one for everybody
Start with 8 out of 10 as the line for every market. After two months, look at the scores and move it. A new market with little brand history can run at 7 while the local team learns the voice. A flagship market should hold 9.
Write the pass mark in the same sheet as the scores. A posted standard is easier to hold than an unwritten one.
Review ten posts per market, chosen at random
You do not need to score everything. Pull 10 published posts per market each month, picked at random so the team cannot prepare. Score them in a shared sheet with these columns: post link, market, brand, five check scores, total, reviewer.
Use a native reviewer, not the person who produced the translation. If you work with an agency or freelancers, ask them to score a different team's posts. That keeps the score honest and takes about 20 minutes per market.
Read the index by brand and by language
An average hides the problem. Cut the same data two ways:
- By market. One language sits at 6.4 while the rest are above 8. That is a translator, a prompt or a process issue in one place.
- By check. If tone is the weak check everywhere, the brand voice guide is too thin. If links are the weak check, add a pre-publish step.
Pair the index with a brand consistency dashboard if you run several brands, so quality and consistency show up side by side.
Fix the lowest market first, then change the process
When a market falls under the mark, do three things in this order. Send the failing posts back with the check that failed. Add the recurring mistake to that market's style notes. Re-score ten posts the next month.
If the same check fails three months in a row, change the workflow, not the person. That might mean a glossary of banned and required terms, a short brief with local examples, or a required native review step before approval.
Build the review step into the publishing flow
The index only works if review happens before publish. In Mydrop, each brand can have its own workspace and approval flow, so a local reviewer approves the post before it is scheduled. AI-assisted captions can produce a first version in the local language, and the reviewer scores and fixes it instead of writing from scratch.
If you are deciding which markets to translate first, read which markets to localize first, then score the first ten posts you ship.
Score ten posts this week
Open a blank sheet, add the five checks, and score ten posts from your busiest non-English market. Write down the total and the weakest check. That one number is your baseline, and you can start improving it on Monday. To manage localized posts across brands in one place, start free with Mydrop.
Google review
Trustpilot review
Product Hunt comment
