One thousand words, two translators, two very different amounts of work. The industry’s favorite pricing unit is simple, familiar and, increasingly, a poor description of what buyers are actually paying for.
THE SHORT VERSION
- Per-word pricing survives because it is easy to compare, not because it is accurate.
- Word count hides difficulty, research time and risk.
- Translation memory and AI have changed what a “word” costs to produce.
- A better quote shows the lines that protect quality: terminology, second review, QA and formatting.
Here is an uncomfortable question for both sides of the table. If two translators deliver the same 1,000 words, and one spent three hours researching terminology while the other spent thirty minutes, should the invoices be identical?
Under the pricing model that dominates the language industry, they would be. The unit is the word. The rate is the rate. What happened between the source text and the final file is, from the invoice’s point of view, invisible.
How a simple unit became the standard
Per-word pricing became the industry norm for good reasons. It is simple. A buyer can compare three quotes in a few seconds. A project manager can build a budget in a few minutes. A freelancer can estimate income without a long negotiation. When a file arrives, the count is objective: nobody has to argue about how many words it contains.
That simplicity built trust between strangers who had never worked together, often across time zones and currencies. It is the reason the model has lasted for decades. But simple is not the same as fair, or accurate.
What per-word pricing hides
Four things disappear when the word becomes the only number on the page.
First, effort is invisible. A legal clause, a medical leaflet and a product slogan can have exactly the same word count and completely different difficulty. The clause needs a second reading by someone who knows the jurisdiction. The leaflet needs verified terminology. The slogan may need five rewrites to land. All three are quoted at the same rate.
Second, it rewards speed over judgment. The fastest way to earn more per hour under a per-word rate is to think less per word. That is not a criticism of translators; it is the incentive the model creates. The professional who stops to research, query the client or rework a sentence is, financially, being penalised for doing the job well.
“The fastest way to earn more per hour is to think less per word.”
Third, translation memory and AI have changed the unit itself. When a large share of a file is repeated sentences or close matches to earlier work, “per word” stops describing the real work. Discount grids exist to correct for this, but they are often negotiated line by line and rarely explained to the end client. Add machine translation and post-editing to the workflow and the same word can involve very different amounts of human effort depending on the quality of the first draft.
Fourth, it turns every conversation into a price comparison. When the only number on the page is a rate, the cheapest rate wins the argument. Quality, process, security and accountability become things buyers have to take on faith, because nothing in the quote lets them check.
What per-word pricing gets right
It would be unfair to stop there. The great strength of the model is transparency. Buyers understand it. Nothing is buried in a vague “project fee” that can swell without explanation. Anyone who has received a quote that simply said “translation project: one lump sum” knows how unsettling that can be. Any alternative has to be at least as clear as the per-word rate, or it is a step backwards.
What a fairer quote could look like
Our position is simple: price the outcome, not only the word. For straightforward content, per-word remains a reasonable base. Around it, a good quote makes visible the lines that actually protect quality:
- Terminology work: building or checking a glossary so the same concept is translated the same way every time.
- Independent review: a second linguist who did not do the translation reads it before delivery.
- QA checks: numbers, tags, consistency and formatting verified with tools, not by eye alone.
- Formatting and layout: especially for right-to-left languages and unique scripts, where a broken layout can ruin a correct translation.
- Reuse: how earlier work is credited, so clients pay once for a sentence and benefit from it on every update.
Imagine two quotes for the same file. One shows a single rate and a total. The other shows the rate, plus a line for terminology, a line for review and a line for QA. The second is not necessarily more expensive. But it tells the buyer what they are getting, and it tells the next vendor what the bar is.
Where we might be wrong
Every alternative to per-word has its own risks. Hourly rates can reward slowness. Fixed project fees can hide risk or push cost onto the linguist. Outcome-based pricing is hard to define when the outcome is “good writing in another language”. We do not claim to have solved this. We think the industry is overdue for a more honest conversation about it.
Over to you
We want to hear from both sides. If you buy translation: what would make you trust a quote that is not per-word? What would you need to see on the page? If you translate: which part of your work does a per-word rate never pay for?
Comment below. We read and answer every comment.
Language Shift by Transinity: AI, localization, careers and culture from inside the language industry. Follow Transinity on LinkedIn and join the conversation.