Your client calls. They’ve seen an AI translation demo online, and they want to know why your MTPE quote isn’t 80% cheaper. “It’s just machine translation plus some fixes,” they say. “Why are we paying you so much?”
You have 30 seconds to not lose the deal.
Here’s the problem: they think the AI does 90% of the work and you do 10%. The reality is messier, and it depends entirely on what you’re actually translating. Your job isn’t to defend MTPE pricing—it’s to explain what MTPE actually costs you, so the client understands why the price is what it is.
This conversation happens dozens of times a year at translation agencies. Either you’ve learned to win it, or you’re stuck competing on price alone and bleeding money. Let’s fix that.
The Core Lie: MTPE Should Be Cheap¶
Here’s what clients think they understand:
- AI translation: free (or pennies)
- You add human expertise: small markup for QA
- Result: massive savings
Here’s what actually happens:
- Raw MT output: technically complete, quality is often 50-70% publishable
- Your post-editor: spends 4-8 hours fixing terminology, omissions, mistranslations, tone, formatting
- QA: another pass to catch errors the post-editor missed
- Result: project costs 30-60% of full human translation if the content is suitable for MTPE
The discount sounds good until you realize: post-editing a 2,000-word technical document might take the same 8 hours it would take to translate from scratch, just compressed into 4 hours of focused work. Speed isn’t the same as cost savings for you—it means your post-editor is working at higher cognitive intensity.
According to a 2025 survey of 500+ translators and post-editors, nearly 50% refuse to discount MTPE work at all. They’re not being difficult—they’re saying what the data shows: effort isn’t always lower.
Why Clients Think MTPE Should Be Dirt Cheap¶
Let’s be honest: clients see “machine translation” in the name and assume most of the work is automated. They compare it to Google Translate (free) or a rough draft (low-value). They don’t see what your post-editor actually does.
What they see: a fully translated document that “just needed a quick review.”
What they don’t see: - The MT system mistranslating a single term 47 times across the document, requiring individual fixes - The machine missing context and generating nonsense for a company name - The grammatical errors that look “close enough” but aren’t - The formatting that got flattened and needs manual reconstruction - The 30-minute research rabbit hole to find the right terminology
Clients think post-editing is a junior-level task—light proofreading. In reality, it’s often harder than translating from scratch, because you have to read both the source and the MT output simultaneously, spot what changed, decide if it’s wrong, and fix it. That cognitive load is brutal.
The Math: Why Your MTPE Rate Is What It Is¶
Let’s walk through actual numbers.
Scenario: A 2,000-word technical manual in English → German
Your normal full-translation rate: €0.15/word = €300 for the project
Time to translate from scratch: 8-10 hours (at ~200-250 words/hour)
Now with MTPE:
- Raw MT time: 30 minutes (basically instant)
- Post-editing time (depending on content difficulty):
- Light PE (simple, repetitive, strong MT): 2-3 hours
- Full PE (complex, technical, spotty MT): 5-8 hours
- QA check: 1 hour
Light MTPE total: 3-4 hours work If you charge 35% discount = €195/project, you’re earning €49-65/hour. Fair.
Full MTPE total: 6-9 hours work If you charge 50% discount = €150/project, you’re earning €17-25/hour. That’s garbage for skilled work.
Here’s the mistake: charging the same discount percentage for both. Light PE and full PE are completely different. One deserves a discount; the other deserves to be priced differently.
Light MTPE vs. Full MTPE: The Pricing Line¶
This is the core of your client conversation. You need to help them understand the difference—because if they get this, they’ll stop pushing back on price.
| Criterion | Light MTPE | Full MTPE |
|---|---|---|
| Content type | Repetitive, simple, technical specs, e-commerce, user manuals with glossaries | Legal, medical, creative, marketing, highly nuanced |
| MT output quality | Usually 70-85% publishable | Often 40-60% publishable |
| Editing depth | Fix terminology, obvious errors, tone polish | Rewrite sections, verify facts, rebuild sentences |
| Hours per 1,000 words | 2-4 hours | 4-8 hours |
| Fair discount | 25-40% off full rate | 50-80% of full rate (or hourly) |
| Productivity: words/day | 4,000-8,000+ | 2,000-5,000 |
See the difference? Light MTPE genuinely saves you work and time. Full MTPE often doesn’t—it just feels faster because the first draft already exists.
As one post-editor writes on SEAtongue: “Post-editing work can compress 80% of the cognitive labor into a shorter timeframe—a 2,000-word fresh translation takes 10 focused hours, while the same AI-generated 2,000 words takes 4 hours at a 50% rate. But complications like factual errors requiring research can extend the timeline.”
Your message to the client: “Light MTPE on your product catalog? Absolutely, we can discount that 30%. Legal contract? That’s full MTPE work—it takes nearly as long as translating from scratch, so we’re charging 80% of our standard rate. Or we can bill hourly and give you a time estimate upfront.”
Most clients will accept this if you explain the difference. They think it’s all the same “light AI touch-up”—because no one’s told them otherwise.
The Productivity Numbers (Show Your Work)¶
Clients love numbers. Give them numbers.
According to post-editing benchmarks from translation agencies worldwide:
Daily output: - Human translation alone: 2,000-2,500 words/day - Full MTPE (700 words/hour): 5,600 words/day (2.5x faster) - Light MTPE (1,000 words/hour): 8,000 words/day (3-4x faster)
But speed ≠ cost savings for the translator, because the compressed cognitive load is intense. A human translator can maintain quality for 2,500 words in a day. A post-editor at 5,600 words in a day is working much harder per hour—and that’s fine for short projects, but for long-term MTPE work, your post-editors will burn out at 50% rates.
One more number: According to the 2025 State of MTPE survey, 66% of post-editors describe machine output as “acceptable but requires significant edits,” and 21% report it as “poor, requiring extensive rework.” Only 12% say MT output is “usually high quality.”
Your talking point: “Our productivity on MTPE is 2-3x human translation. But the quality bar doesn’t drop—we still charge for expertise. Here’s how we price it…”
How to Price MTPE: Three Models¶
Clients are used to per-word rates. Give them options, because rigidity loses deals.
Model 1: Tiered Discounts (Easiest to Explain)¶
- Full human translation: €0.15/word (baseline)
- Light MTPE: 35% discount = €0.10/word
- Full MTPE: 20% discount = €0.12/word
- Raw MT (no PE): 70% discount = €0.045/word (but you won’t recommend this)
Pros: Simple, familiar, clients understand it instantly. Cons: Doesn’t reflect actual effort on hard projects. Post-editors resent it.
Model 2: Hourly Rate (Protects You on Hard Projects)¶
- €50-70/hour for experienced post-editors (depending on your market and language pair)
- Quote fixed time upfront: “This 2,000-word manual will take ~5 hours of post-editing work = €250”
- If the MT turns out worse than expected, you bill for actual time
Pros: Transparent, sustainable, protects you from unprofitable projects. Cons: Requires clients to trust your time estimates. Less “shoppable” for price comparison.
Model 3: Hybrid (Best for Long-Term Clients)¶
Light MTPE: per-word discount (e.g., 35% off) Full MTPE: switch to hourly if effort exceeds an estimated threshold Raw content unsuitable for MTPE: full human translation only
Pros: Flexible, rewards clients for choosing MTPE on suitable content, protects you on hard projects. Cons: More complex to explain initially, but clients get it once they see one project billed both ways.
My recommendation: Start with Model 2 (hourly) for new MTPE clients. Once you’ve done 2-3 projects together and they trust your estimates, you can offer Model 1 (tiered discounts) as a loyalty perk.
The Client Conversation: Five Key Points¶
When a client pushes back on MTPE pricing, you need five things in your arsenal.
1. Show Them Bad MT¶
“Here’s what untouched machine translation looks like.” Pull a sentence from their raw MT output and your post-edited version side-by-side. Make it visceral. Watch them realize the machine translated a proper name as a common noun, or missed an entire clause.
2. Define Light MTPE vs Full MTPE¶
“We can offer you two MTPE tiers. Light editing—for your product descriptions and manuals with consistent terminology—runs 35% below our standard rate and delivers in half the time. Full editing—for your contracts and technical specs—takes nearly as long as translating from scratch, so we charge 80% of standard. Which projects fit which category?”
Most clients will accept this immediately. They’re not stupid—once you explain the difference, they get it.
3. Mention the Alternative¶
“If MTPE doesn’t hit your budget, we can quote full human translation, which costs more but takes a different amount of time. Or we can split: human translation for the critical bits, MTPE for the repetitive stuff.” Gives them an out without dropping your MTPE rates.
4. Talk About Quality, Not Price¶
“You’re not paying us for the machine translation—you’re paying for a specialist who can read both the source and the machine output, spot what the machine missed, fix the terminology, and guarantee it won’t embarrass you. The cheaper the price, the less experience and care goes into that review.”
This reframes the conversation from “Why is MTPE so expensive?” to “Why would I trust cheap MTPE?”
5. Set Expectations About Content¶
“Not all of your content is suitable for MTPE. The more repetitive and technical, the better MTPE works—and the lower the cost. The more creative or legal, the worse MTPE works—and the higher the cost because the effort is almost the same as starting from scratch.”
This gives you permission to say “no” to MTPE on certain projects.
When NOT to Discount MTPE (Or Not to Offer It)¶
This is the conversation you’re probably avoiding, but it’ll save you thousands in margin loss.
Don’t offer MTPE when:
- The content is legal or certified. A lawyer or notary has to review MTPE output anyway, so you’re not saving them work. Charge full rate or don’t take the job.
- The language pair has weak MT. English to Swedish MTPE is 7% slower than human translation (per the data above). If your MT engine for a language pair is mediocre, post-editing takes longer and is more painful. Don’t discount it.
- Your client only cares about price. If they’re shopping MTPE purely to pay less, they’ll be miserable. They’ll compare your MTPE rate to some offshore human translator’s rate and feel cheated. Better to only offer full human translation to price-sensitive clients.
- You don’t have a skilled post-editor for that content type. MTPE requires more expertise than routine human translation, not less. A weak post-editor will miss the machine’s errors. If you don’t have a specialist, translate it yourself.
Do offer MTPE when:
- High volume, repetitive content (e-commerce, technical manuals, newsletters)
- Client has tight deadlines and is willing to accept “good enough” instead of “perfect”
- Your MT engine for that language pair is strong
- You have a skilled post-editor on staff
If none of these apply, politely pass on the MTPE pitch and quote full translation.
Three Sample Pricing Conversations¶
Let’s get concrete. Here’s how to handle the most common client objections.
Conversation 1: “Your MTPE Price Isn’t Much Cheaper Than Full Translation”¶
Client: “You’re charging 50% of your full rate for MTPE. That’s not much of a discount. Why don’t you charge 30%?”
You: “Good question. Here’s the reality: if we’re doing full MTPE on your content—catching the machine’s errors, fixing terminology, rebuilding weak sentences—it takes nearly as long as translating from scratch. Charging 30% would mean we’re working for poverty wages, and we’d burn out.
What I can do is look at your project and see if some of it qualifies for light MTPE—quick product descriptions or repetitive sections that the machine usually gets right. Those we can do at 35% discount. The technical stuff? That’s 20% discount or hourly billing, because the effort is real.
Let me review your content and break it down by section. I bet 60% of it is light-PE territory.”
Outcome: Client sees you’re not being lazy—you’re actually looking at the work and pricing fairly.
Conversation 2: “We Can Get Cheaper MTPE Elsewhere”¶
Client: “I found another vendor offering MTPE at 40% of full translation cost. You’re at 50%. Why shouldn’t I go with them?”
You: “You should if that’s your only criterion. Here’s what I’d ask them: Which MT engine are you using? How is your QA structured? How many revisions are included? What’s your SLA if the output is unusable?
We use [specific engine], run two-pass QA with our most experienced post-editors, and include up to two revision rounds. If the machine output is bad, we either re-do it or charge hourly for the extra time, so you’re not stuck with garbage. Cheaper vendors sometimes just run the MT through and light-skim it—and your next client will notice the quality dip.
Price is one factor. But if the cheapest MTPE destroys your brand, it cost you way more than the discount saved.”
Outcome: You’re not competing on price—you’re competing on quality and sustainability. The right client will stay.
Conversation 3: “Raw AI Translation is Free, Why Am I Paying You?”¶
Client: “Google Translate or ChatGPT will do this for free. Why are we paying you 50% of the full price?”
You: “You’re right—raw AI is free and it shows. Let me show you what we’d get:
[Paste a paragraph of untouched MT output]
That’s serviceable gibberish. It reads like a machine wrote it. Your customer or partner will notice immediately. Now here’s what we deliver:
[Paste your post-edited version]
Readable. Correct. Professional. The cost of that transformation is not ‘small’—it’s the core of what we do. You’re paying for expertise, not convenience.
If you want to ship raw AI translation, that’s a business decision I’ll respect. But then you’re not paying us anything—you’re competing on a completely different value proposition, and you need to accept the quality risk that comes with it.”
Outcome: Client understands you’re not in the “cheaper AI” business. You’re in the “professional translation” business. If they want cheaper, they’ll go elsewhere and you’ll move on to someone who values quality.
The Transparency Angle: Show Clients Where Their Cost Goes¶
This works especially well with objections about MTPE price:
“Here’s where your €300 MTPE cost breaks down: - MT engine licensing & API cost: €20 (infrastructure you’re sharing across clients) - Post-editor hourly rate, 4 hours @ €50/hour: €200 (the person who actually fixes the machine’s mistakes) - QA specialist, 1 hour @ €40: €40 (catches errors your post-editor missed) - Project management, invoicing, payment processing: €20 - Margin to keep the lights on: €20
Total: €300 (50% of your €600 full-translation rate)
If we charged 30%, we’d have to fire the QA person, use a less-experienced post-editor, or not exist. You can try it, but you’ll get what you pay for.”
Clients respond to transparency. Most of them genuinely don’t know where your costs go. Once they do, price pushback often stops.
The Real Shift: From Per-Word to Value¶
Here’s the uncomfortable truth: the per-word model is dying for MTPE. It doesn’t work.
Per-word assumes translation is always the same effort per word. It’s not. One 1,000-word document might take 3 hours to post-edit; another might take 8 hours. Charging the same rate per word for both is either unfair to you or unprofitable.
The agencies winning on MTPE are shifting to: - Hourly rates (transparent effort-based pricing) - Project rates (fixed price upfront, based on estimated hours) - Tiered service levels (light PE vs full PE vs human translation, with clear effort/cost differences)
If you’re still quoting MTPE as a flat 40% discount, you’re leaving money on the table and frustrating your post-editors.
Your new positioning: “We price MTPE based on the actual effort, not an arbitrary percentage. Light MTPE on simple content costs less. Full MTPE on complex content costs almost as much as translation because the effort is real. We’ll do a quick review of your project and give you an honest estimate—and you’ll know exactly what you’re getting.”
Clients respect honesty. Most will accept that.
FAQ¶
Q: Should I always discount MTPE, or can I charge the same as human translation?
A: Neither. The discount depends on actual effort. Light MTPE on suitable content (product descriptions, repetitive docs) deserves 30-40% off because the effort is genuinely lower. Full MTPE on complex text (legal, technical) often takes as long as translating from scratch, so charge 80-100% of your full rate or switch to hourly billing. The worst mistake is a flat 50% discount across all content types—that’s how you destroy margins and burn out your post-editors.
Q: How do I respond when a client says ‘AI can translate for free, why am I paying you 50% of the price’?
A: “Because raw AI is garbage. Show them an example of untouched MT output versus your post-edited version—the difference is night and day. You’re not selling the AI; you’re selling the skilled human who fixes its mistakes, catches omissions, and guarantees it won’t embarrass them. Frame it: ‘Raw machine translation costs you nothing and delivers nothing publishable. Post-editing costs half, delivers professional quality.’ A client who doesn’t value that distinction isn’t a good MTPE fit.”
Q: My translators refuse to post-edit at 50% off because they say it takes as long as translating. Are they right?
A: Yes, sometimes. MTPE is not inherently faster—it depends on content and MT engine quality. Simple, repetitive content (e-commerce descriptions, manuals) legitimately goes faster at light-PE rates. Complex text (legal, medical, creative) can take just as long or longer. If your post-editors tell you a 2,000-word legal document takes 8 hours to fix when they’d translate it from scratch in 8 hours, believe them. Either charge accordingly or don’t offer MTPE for that work.
Q: How do I position MTPE to cost-conscious clients without looking desperate?
A: Stop leading with cost. Lead with speed and scale. “For your 50,000-word monthly library, MTPE gets it to you in 2 weeks instead of 6, at a cost saving of 30%.” Clients adopt MTPE because they have volume and deadlines, not because they want to pay less. If price is their only concern and they don’t have volume or speed pressure, MTPE isn’t the right fit.
Q: Can I charge hourly for MTPE instead of per-word to avoid the discount trap?
A: Yes, and it’s smarter. Per-word rates lock you into discounts (30-50% off) that don’t reflect effort. Hourly rates (€40-70/hour for experienced post-editors) are clearer: a client pays for time, period. You quote a fixed estimate upfront, and the client knows what they’re paying. This also protects you when projects turn out harder than expected.
Q: What percentage discount do agencies typically offer for MTPE in 2025-2026?
A: It varies: 30% offer 25-40% discounts, 20% offer 40-60%, and roughly 50% don’t discount at all. The median is around 30-40% for full MTPE and 25-35% for light MTPE. But the best agencies have stopped thinking in percentages and started thinking in effort-based pricing or hourly rates. Blind percentage discounts destroy margins and breed resentment.
Q: What should I do if a client absolutely refuses to accept my MTPE price?
A: Offer them full human translation at your standard rate. If they balk at that too, they’re price-shopping and not a good fit. You can’t compete on price—you compete on quality, reliability, and expertise. Losing a race-to-the-bottom deal frees up time for clients who value what you actually do.