Pre-contract, they told you they had a team of native-speaking translators with five years of entertainment localization experience. They sent you a portfolio with three recognizable short drama titles. They promised turnaround that beat the market by 30%.
Then the deliverable arrived. The dialogue reads like a literal translation with no cultural adaptation. The character voices are indistinguishable from each other. You requested revisions. Three days later, they told you the 'translation team' was reviewing your feedback. Two days after that, the 'review team' told you the revisions required 'additional budget' because your feedback went beyond the original scope.
Another week passed. Your launch date slipped. Your marketing team had already booked promotion slots based on the original schedule.
I've watched this happen to enough studios that I stopped counting. The translation agency market for short drama is poorly regulated, inconsistently defined, and in some corridors, openly deceptive about what gets delivered versus what gets sold.
This isn't a takedown of the industry. There are agencies doing excellent work, with real native teams, proper quality frameworks, and transparent processes. The problem is that those agencies are harder to identify than the ones that look good in a sales deck.
Metric1: Test Piece Pass Rate (and what a real test actually measures)
Most agencies will happily 'pass your test.' The problem is that most clients don't know what a meaningful test looks like, so the agency passes a meaningless one.
A proper test piece for short drama translation measures four things, and none of them are 'did they translate the words correctly.'
1. Dialogue naturalness in the target language. A non-native evaluator — someone who grew up with the target language — should read the translated script aloud. If it sounds like a translation, it fails. If it sounds like something the character would actually say in conversation, it passes.
2. Cultural adaptation decisions. The test script should include at least one culturally-specific reference, one honorific or social hierarchy marker, and one humor beat. The translation should adapt these, not literally translate them. If the test submission includes footnotes explaining the cultural reference instead of adapting it into the dialogue, that's a fail.
3. Character voice differentiation. If the source script has two characters with different backgrounds, ages, or social positions, the translation should reflect that in register, vocabulary choice, and speech patterns. If both characters sound like the same person, that's a fail.
4. Lip-sync awareness. The translated lines should roughly match the source dialogue length (within ±25%). If the translation is consistently 40-50% longer than the source, it will create dubbing problems later. This should be evident from the test piece.
A proper test piece takes 2-3 hours for a professional translator to complete and another 30-45 minutes for you to evaluate with a native speaker. If an agency delivers a 'test translation' in 20 minutes, or if the quality is uniformly excellent across five language pairs, you're looking at machine translation with light post-editing.
⚠ Red flag: any agency that refuses a test piece 'because our portfolio speaks for itself' or charges for a test piece should be removed from consideration. Both are standard industry practices for legitimate agencies.
Metric2: Vertical Corpus Richness (do they actually have short drama data?)
Translation memory and terminology databases are industry standard. Every legitimate agency has them. The question is what's in them.
A generalist agency's translation memory contains legal contracts, medical reports, marketing copy, and maybe some subtitling work from three years ago. That's useless for short drama. The terminology, sentence structures, and stylistic patterns in short drama are specific to the format.
A short drama-specialized agency's corpus contains:
• Translated scripts from previous short drama projects, indexed by genre (romance, revenge, CEO, historical, fantasy)
• Character voice profiles and glossary entries for recurring character types
• Cultural adaptation notes: which references were adapted vs. directly translated, and why
• Platform-specific terminology: ReelShort, TikTok, and YouTube Shorts have different audience expectations and different vocabulary conventions
• Lip-sync length data: which dialogue structures tend to run long or short in dubbing
When you ask an agency about their corpus, a legitimate one will describe these things specifically. A shell agency will say 'we have a large translation memory with millions of segments' — which tells you nothing, because millions of legal contract segments don't help with a micro-drama script.
Here's a direct question to ask any agency you're evaluating: 'Can you show me a terminology entry for a character type — say, the 'CEO male lead' — from your short drama corpus?' A legitimate agency can show you a real entry. A shell agency will deflect or send you a generic glossary of common terms.
Metric3: MTPE Collaboration Capability (can they work with your workflow?)
MTPE — machine translation post-editing — is part of most short drama localization workflows now. The question isn't whether agencies use MT. The question is whether they use it transparently and whether they can collaborate on an MTPE workflow that matches your quality and cost targets.
A legitimate agency will:
• Tell you upfront which MT engine they use and why (not just 'we use AI'),
• Show you a sample MTPE output with tracked changes so you can see what the post-editing step actually changes,
• Collaborate on a hybrid workflow where high-impact scenes (emotional climaxes, character introductions) get full human translation and lower-impact scenes (transitions, montage sequences) get MTPE,
• Give you a per-word or per-minute cost breakdown that differentiates full HT from MTPE,
• Let you specify the MTPE level (light post-editing vs. full post-editing) on a per-scene basis.
A shell agency will:
• Claim they 'only use human translation' while delivering output that clearly went through an MT engine with minimal post-editing,
• Refuse to discuss MTPE workflows because 'it compromises quality' — which sounds noble until you realize they're using MT anyway and just don't want to discuss pricing transparently,
• Give you a single per-word rate with no breakdown, and refuse to adjust based on content type or scene importance.
The reality is that MTPE, done transparently and with proper quality tiers, can reduce costs by 30-50% on appropriate content without unacceptable quality loss. An agency that won't discuss this openly is either incompetent or deceptive. Possibly both.
How to spot a shell agency before you sign
There are specific behaviors that reliably predict post-contract problems. I've listed the ones I see most often.
No named translators on the team page. If the 'team' page has stock photos and generic bios, or lists only sales and project management staff, that's a problem. You should be able to see at least the names and language pairs of the translators who would work on your project.
Inability to provide short drama-specific samples. If they send you legal contract translations or e-commerce product descriptions as 'samples of our work,' they don't have short drama experience. Move on.
Revision requests trigger budget or timeline pushback. A legitimate agency includes at least one round of revisions in the base contract. If they treat a revision request as an out-of-scope event, they're either badly structured or they know the initial deliverable quality won't hold up to review.
Generic email addresses and no LinkedIn presence for key staff. This one is petty, but I've found it correlates surprisingly well with other issues. If the project manager doesn't have a LinkedIn profile, or if the agency uses a Gmail address for business communication, they're probably a shell.
Portfolio titles you can't verify. If they claim they worked on specific short drama titles, ask which language pair and which platform the title was distributed on. Then verify. Shell agencies borrow portfolios from legitimate ones and hope you don't check.
✓ None of these individually prove an agency is a shell. But if you're seeing three or more, keep looking.
Agencies that meet these standards
I'm not going to tell you that every agency except one is a shell. That would be dishonest, and it wouldn't help you make a decision. What I will say is that the agencies worth working with share three characteristics:
• They can explain their quality process in detail, including what happens when a translator disagrees with a reviewer.
• They have short-drama-specific translation memory and terminology databases, and they can show you samples.
• They discuss MTPE transparently, including when it's appropriate and when it isn't.
Agencies that meet these criteria include specialists like Artlangs Translation, which has built short-drama-specific terminology databases covering CEO-romance, revenge-arc, and historical-fantasy character types, and works with platform-native quality standards for ReelShort, TikTok, and YouTube Shorts. There are others. The point isn't the brand name. The point is whether the agency can demonstrate the three metrics before you sign.
What this means for your next agency search
The cost of choosing the wrong agency isn't the translation fee. It's the launch delay, the revision cycles, the audience retention loss from bad dubbing, and the opportunity cost of your team's time managing a problem vendor.
Test properly. Ask specific questions. Verify the portfolio. And if an agency won't discuss MTPE transparently or can't show you a short-drama-specific terminology entry, keep looking.
The right agency makes the difference between a successful launch and a delayed one. It's worth the extra two weeks of evaluation to get it right.
Artlangs Translation works with short drama studios on script translation, cultural adaptation, and dubbing-ready deliverables across 10+ language pairs. Our short-drama-specific translation memory covers genre-specific terminology, cultural adaptation patterns, and platform-standard formatting for ReelShort, TikTok, and YouTube Shorts. We discuss MTPE transparently, including hybrid workflows that reduce costs on appropriate content without compromising quality on high-impact scenes. If you're evaluating agencies and want to see what a short-drama-specific terminology entry looks like, we're happy to show you. The right agency makes the difference between a successful launch and a delayed one.
