Per‑word pricing is supposedly passé. If you believe the conference chatter, we’re all meant to be billing by “value”, “impact units”, or whatever tomorrow’s slide‑deck buzzword will be. Hourly rates get wheeled out as the “grown‑up” alternative, and AI is held up as proof that churning out words no longer equates to effort. Lovely story—shame it crumbles under scrutiny. Let’s unpack why the oldest pricing model in localisation remains the steadiest, fairest, and most future‑proof deal on the table.
1. The Tech Tango: Why AI Doesn’t Kill the Word
Large language models can indeed spit out passable copy faster than an espresso shot. Yet raw MT is a confidence trick: polished surface, shaky foundations. Studies from TAUS and CSA Research show average factual error rates of 15–25 % in unedited MT output for general domains, and dramatically higher in regulated sectors like medical and legal. That “polished” sentence might swap milligrams for micrograms or invert a negative—mistakes with real‑world cost.
Fixing those stealthy errors is cognitive heavy lifting. Translators must run a mental diff on every sentence: What’s changed? Why? Is nuance lost? It’s line‑by‑line detective work, not a quick spell‑check. A blanket “AI discount” assumes the machine did 80 % of the job when, in reality, it often skipped the tough 20 %—terminology, stakeholder tone, regulatory nuance—the exact 20 % that guarantees brand safety.
Per‑word rates let you quote a fair baseline, then bolt on sensible MT‑post‑editing matrices or edit‑distance bands where appropriate. Everyone sees the calculation. Trust stays intact.
2. Hourly Rates: The Perverse‑Incentive Machine
Hourly pay feels adult—architects do it, consultants do it—so why not linguists? Because localisation isn’t architecture. In an hour, one translator might polish 500 Spanish marketing words to a mirror shine, while another inches through 150 German legal terms. Pay by the clock and you punish speed, reward dawdling, and turn every coffee break into a moral dilemma. Productivity tools (think memoQ productivity plug‑ins or Trados analytics) confirm that top‑tier linguists can be three times as fast as juniors without compromising quality. Per‑word pricing lets those veterans pocket the premium their mastery deserves—precisely the incentive that retains senior talent.
3. The Mirage of Value‑Based Pricing
“Price the outcome, not the output,” cry the thought‑leaders. Great on paper—lethal in practice. Genuine value‑based engagements do exist: a consultancy gig to slash support tickets by debugging multilingual UX, or a content‑design sprint to lift conversion in a new market. Those smell like management consulting, not translation. They’re scoped, KPI‑driven, and billed accordingly—usually with risk‑sharing baked in.
But the second you drag frontline translators into that model, the arithmetic implodes. Suppose your KPI is “boost German app‑store ratings by 0.3 stars in six months.” How do you apportion that uplift across seven linguists, a QA lead, and the product team tweaking UI strings? Suddenly everyone is hostage to variables they can’t control: release cadence, customer sentiment, even the weather (really—bad weather spikes mobile‑gaming downloads). Transparent? Hardly. Scalable? Never. The only predictable element in localisation is the volume of words (or characters or segments) that must change language. Price those accordingly, then quote consultancy separately. Clean lines, clean conscience.
4. Transparency, Planning, Sanity
Finance directors love per‑word. Procurement loves per‑word. Project managers love per‑word. It’s spreadsheet‑ready, and it de‑risks the budget conversation. “Here’s the source word count, here’s the discount grid for matches and repetitions—done.” You can cost a product launch into 18 languages before the devs have even frozen the string file. Try pulling that stunt with “value units” and watch the CFO’s eyebrows migrate north.
5. Mental‑Health Maths
Fast translators often mention an overlooked perk: predictable earnings per session. Instead of clock‑watching, they set a word target, race for the finish, then detach. That autonomy combats the chronic presenteeism and cognitive fatigue that plague creative knowledge work. When burnout already looms large in freelance circles, a model that lets professionals pace themselves without wage anxiety is worth its weight in wellness webinars.
6. Handling the Edge Cases
- Taglines and microcopy – Introduce a minimum charge or switch to a project fee. A 5‑word slogan can take two hours of cultural sleuthing; price it like design, not production.
- Interpreting – Charge by the day or half‑day. Per‑word would be comic (and cruel).
- Heavy MTPE with low‑quality input – Layer in edit‑distance coefficients. If you rewrite 60 % of the segment, you should earn 60 % (or more) of full‑rate. Most CAT tools now report that delta in seconds.
In every instance, per‑word remains the anchoring metric. You simply graft context‑sensitive rules onto it instead of ditching the foundation.
7. Who Really Wins from Scrapping the Word?
Follow the money. Platforms, consultants, and middle‑layer tech vendors want friction—new billing schemes mean training sessions, bespoke dashboards, subscription analytics. Translators? They inherit the risk. Clients? They face opaque invoices and budgeting headaches. The supposed “innovation” looks suspiciously like margin‑shuffling.
The Bottom Line
Per‑word pricing isn’t a dusty relic from the typewriter age. It’s the rare crossover where efficiency, fairness, and predictability actually align. Keep refining it—use smarter CAT analytics, introduce minimums, apply edit‑distance logic—but resist the siren call of philosophies that sound progressive while quietly eroding professional leverage.
Until AI learns to shoulder cultural nuance, legal liability, and brand reputation without human shepherding, the word will remain localisation’s most reliable currency. Long may it reign—one undisputed token of value in a marketplace obsessed with shiny but shaky alternatives.