An agency reaches out: “MTPE, 50% of your normal rate. The machine’s already done half the work — doesn’t that make sense?” You’re tempted. Then you dive into the project and realize the machine’s output is grammatically correct but semantically nonsensical. Every sentence needs checking. Half of them need rewriting. By the end, you’ve done almost as much work for half the pay.
The question isn’t whether you can charge less for MTPE. The question is whether you should. And more importantly, at what threshold does it stop making financial sense.
What MTPE Actually Costs on the Market (2025-2026)¶
Let’s start with the numbers without the marketing spin.
Today’s market rate for MTPE is $0.05-$0.15 per word — that’s 30-50% off standard translation rates. If you normally charge $0.20/word for business text, agencies are quoting you $0.06-$0.14 for MTPE. On the surface, that looks reasonable. Let’s do the math.
Comparison on a 10,000-word project:
| Scenario | Rate per word | Volume | Total | Time (industry standard) | Hourly rate |
|---|---|---|---|---|---|
| Full translation | $0.20 | 10,000 | $2,000 | 40 hours (250 words/hr) | $50/hr |
| MTPE (50% rate) | $0.10 | 10,000 | $1,000 | 14-15 hours (700 words/hr) | $67-71/hr |
| MTPE (poor MT quality) | $0.10 | 10,000 | $1,000 | 30-35 hours (300 words/hr) | $29-33/hr ⚠️ |
Here’s the paradox: if you’re truly editing 700 words per hour, your $0.10/word rate gives you $70/hour — more than full translation. But that’s only if the machine output is actually decent.
The real constraint is one number: 50% of post-editors work at 1000-3000 words per hour, but only if the machine translation is usable. When MT quality is poor, you’re down to 300-500 words/hour, and suddenly your rate drops below minimum wage.
Why MTPE Often Takes Just as Long as Full Translation¶
Neural machine translation (the current standard) does one thing remarkably well: it produces text that’s grammatically correct but semantically wrong. You read it and think “okay, this is a translation.” You read it again and realize half the meaning is gone.
Take this example:
Original: “The company decided to shelve the project indefinitely.”
MT output: “The company decided to place the project on the shelf indefinitely.” (technically accurate, completely bizarre)
What you actually need: “The company postponed the project with no set timeline” or “The company put the project on hold”
Full translation: you start from scratch, knowing exactly what needs to happen on the target side. MTPE: you think the machine did 80% of the work, then discover you need to check almost everything.
Research shows the quality of machine translation is the key variable. For every 1% improvement in MT quality, average post-editing time drops by 0.16 seconds per word. That sounds tiny, but on a 10,000-word project, it’s 27 minutes of difference. On poor MT, those 27 minutes can become 5-7 hours of extra work.
Industry Standards: How Fast Should MTPE Really Be?¶
Here’s what the research says about post-editing productivity:
- Light post-editing (minor fixes, straightforward text): 1000 words/hour
- Full post-editing (comprehensive corrections, complex content): 700 words/hour
- Industry average (across all project types): 700-1000 words/hour
For comparison: full translation from scratch — 200-300 words/hour on clean material, 100-150 on complex stuff (legal, technical). The math:
700 words/hour × $0.10/word = $70/hour 250 words/hour (full) × $0.20/word = $50/hour
On paper, MTPE looks better. But only if all of these are true: - The machine output is actually coherent - You can trust most of the machine’s suggestions - You can spend 30-60 seconds per word (that’s 60-120 words per minute)
In practice, usually none of those are guaranteed.
Three MTPE Pricing Models (And Which One’s Actually Fair)¶
Model 1: Percentage of Your Full Rate¶
Charge 50-70% of your standard per-word rate. If you normally charge $0.20, MTPE would be $0.10-$0.14.
Pros: Easy to calculate, looks fair.
Cons: Ignores MT quality. On poor machine output, you’re doing near-full translation work for 50% of the money.
Model 2: Editing Distance (The Fairest Approach)¶
Your CAT tool measures how much each segment was actually changed. You charge based on that:
- Segment rewritten >70% = full rate
- Segment modified 30-70% = 70% of rate
- Segment minimally changed (<30%) = 30% of rate
Example: On 10,000 words: - 4,000 words completely rewritten = 4000 × $0.20 = $800 - 3,000 words modified 50% = 3000 × $0.14 = $420 - 3,000 words touched up lightly = 3000 × $0.06 = $180 - Total: $1,400 (instead of $1,000 at flat 50% or $2,000 for full rate)
Pros: Genuinely fair to both sides. Good machine output means lower pay; bad output means full rate for what you rewrote.
Cons: Requires a CAT tool that can calculate this (Trados, memoQ, etc). Not all agencies are set up for it.
Model 3: Hourly Rate¶
Charge $35-50/hour for MTPE work, depending on complexity and language pair.
Pros: On terrible MT, you’re not underpaid. On easy MT, you move fast and that’s your advantage.
Cons: Harder for agencies to budget. They often want a fixed per-project price.
Best practice: Experienced freelancers in Europe often charge €30-50/hour for MTPE (higher in Switzerland and Germany). That’s on good material. On poor MT, they often bump to full rate or walk away.
The MTPE Trap: How Agencies Squeeze Your Rates¶
Here’s how the math looks from the agency side:
- Client pays $20,000 for a translation project
- Agency pitches MTPE + post-editing as an alternative to full translation
- Cost to client drops to $10,000 (50% discount)
- Agency pays you $5,000 (50% of $10,000)
- Agency keeps $5,000 (50% margin)
Result: you’re doing 80% of the work for 50% of the money, and the agency takes home 50% margin. That’s the paradox.
Common Mistake #1: Accepting 50% MTPE rates just because “there’s less work”
There isn’t. You still need to: - Read the entire text (same as full translation) - Verify every sentence for meaning (often replacing full translation) - Rewrite 30-60% of it (worst cases: 80-90%)
Common Mistake #2: Accepting a low hourly rate for MTPE
If your normal rate is $50/hour but you take MTPE at $25/hour, that’s brutal. 700 words/hour × $0.05/word = $35/hour. Brutal.
Common Mistake #3: Not asking to see a sample of the MT before quoting
Best practice: tell the agency “Send me a sample segment so I can see the MT quality. If it’s below [your threshold], my rate is 70% of normal, not 50%.”
When MTPE is Actually Worth It (and When It Isn’t)¶
MTPE Makes Sense:¶
✅ High volume (20,000+ words of one material) ✅ Good MT quality (neural translation for major language pairs) ✅ Easy content (IT, marketing, not legal/medical) ✅ You can realistically hit 700+ words/hour ✅ Rate is at least 70% of your full rate
Example: 30,000 words of IT documentation at $0.12/word = $3,600 for 40 hours = $90/hour. That works.
MTPE Doesn’t Make Sense:¶
❌ Low volume (under 5,000 words) ❌ Unknown or poor MT quality ❌ Complex content (legal, medical, technical with no context) ❌ Rate below 60% of your normal
Example: 5,000 words of legal text at $0.06/word = $300 for 15 hours = $20/hour. That’s not viable.
How to Negotiate MTPE Rates with Agencies¶
Step 1: Get the per-word rate upfront, before committing
Make clear that MTPE isn’t automatically 50% of your full rate — it depends on quality.
Step 2: Ask for a sample of the MT
Send them a paragraph from the source, ask them to run it through their machine translation, evaluate it. If it’s already incoherent, tell them honestly: “At this quality level, I’m charging 70% of my normal rate, or we do this hourly.”
Step 3: Push for editing distance if they’re open to it
“I’ll take 60% for minimal edits, 85% for significant ones, 100% for complete rewrites. Let me use your CAT tool to measure it.”
Step 4: If they won’t budge, walk
Not every project is worth taking. A $20/hour project on your time is a time investment you don’t need. You’re not running a charity.
The Alternative: Hybrid Online Services¶
If an agency pushes cheap MTPE but you’re not interested, consider how hybrid online services approach it. Tools like ChatsControl offer a middle ground: customers upload a document, AI generates a draft, a human translator (sworn translator, in some cases) reviews and certifies it. Price is typically higher than raw MTPE but lower than full translation. For end customers, it’s a compromise. For you, it’s a way to show clients a more transparent pricing structure than just “we’ll charge you 50% for MTPE.”
The machine translation alone isn’t full translation (it breaks on context), but with human review the quality becomes usable for many cases. Plus the document comes back in the original format (DOCX, PDF), which matters for business documents.
Comparison Table: MTPE vs Full Translation vs Hybrid¶
| Scenario | Per-word rate | Time on 10k words | Hourly | Quality | Best for |
|---|---|---|---|---|---|
| Full translation | $0.15-$0.30 | 40-50 hours | $50-75/hr | High | When quality is critical |
| MTPE (50% rate) | $0.06-$0.15 | 14-25 hours | $35-67/hr | Depends on MT | Only with good machine output |
| MTPE (70% rate) | $0.10-$0.20 | 14-25 hours | $50-100/hr | Depends on MT | Reasonable compromise |
| Hybrid (AI + human) | $0.10-$0.20 | 2-4 hours (for customer) | N/A | Medium-high | Budget-conscious customers |
Checklist: Before You Accept an MTPE Project¶
Run through this before saying yes:
- [ ] Do I know the per-word rate or hourly rate? (Minimum: $0.07/word or $35/hour)
- [ ] Have I seen a sample of their MT output and assessed the quality?
- [ ] Do they have CAT access for editing distance? (If not, I need 70% of full rate minimum)
- [ ] What’s my realistic hourly rate on this? Multiply rate × 800 words/hour (conservative estimate). If less than $40/hour, I’m passing.
- [ ] Have I discussed edge cases with the agency? What if the MT is garbage? Is there a rate adjustment?
- [ ] Have I set a time boundary? If this actually takes 3x as long as expected, I’m renegotiating.
- [ ] Am I promising delivery faster than I can actually complete? (Better to ask for extra time on MTPE than to burn out.)
FAQ on MTPE Pricing¶
Is it fair that agencies take 50% margin on MTPE?
Technically they’re absorbing some risk. But if the client’s paying close to full price and you’re getting 50% of the discounted rate, the margin is aggressive. Try to build it into your contract: if the final output is poor, the agency bears the risk, not you — so they don’t get to dictate a low rate upfront.
What do I do if an agency offers MTPE below my minimum?
Decline. Seriously. That project will steal your time and money, damage your reputation (the client will buy it, read it, and be disappointed), and pull your market rates down in the process.
What do I suggest if they insist on cheap MTPE?
“This MTPE rate isn’t viable for me. Here’s what I can do instead: 1. Normal MTPE rate (70% of full) 2. I do the MTPE work, you control quality 3. You pay me full rate for every segment I had to rewrite completely (editing distance) 4. This isn’t the right project for me — find another post-editor”
Most agencies will move to option 1 or 3, because they know cheap MTPE doesn’t actually exist.
How much time should I request for an MTPE project so I don’t underestimate?
Conservative estimate: 1 hour per 1000 words on normal MT. On poor MT, 1 hour per 500 words. Always ask for more time than you think you’ll need. It’s better to deliver early than to ask for an extension.
Bottom line: MTPE is profitable if you’re charging 70%+ of your full rate, the machine output is good, and the volume is substantial. On anything else, it’s a trap agencies invented to shift pricing expectations down across the board — but without shifting your actual workload. Don’t accept 50% just because it’s what they offered. Calculate your hourly rate. That’s the real number.