
Hear the discussion
212% ROI, built on a composite that blends extremes
Forrester's headline model is one composite organization. The interviews behind it describe very different companies.
In October 2026, Forrester Consulting published a Total Economic Impact (TEI) study commissioned by RWS and authored by Diane Deng, projecting a 212% ROI, a $14.7 million net present value ($14,722,281), $21.7 million in present-value benefits against $6.9 million in costs over three years, and a payback period under six months.
- 212%ROI over three years
- $14.7Mnet present value ($14,722,281)
- $21.7Mpresent-value benefits
- $6.9Mpresent-value costs
- < 6 monthspayback period
Forrester built those totals around a single composite global technology enterprise with over $30 billion in annual revenue, 60,000 employees, and localization across 24 to 32 languages. Yet the four RWS-nominated interviewees behind that composite range from a hardware engineering team translating 2.0 billion words a year down to a cloud infrastructure and security team translating 1.5 million words a year.
Annual translation volume, log scale
Interviewee Volume Spread vs. $30B Composite
Each strip is one interviewee's yearly word volume. Every step along the axis is ten times the last. Select a strip, or move between them with the arrow keys, to compare it with the composite.
Select an interviewee to compare it with the composite's Year 1 volume of 90.0 million words.
| Interviewee industry | Role, where reported | Words per year |
|---|---|---|
| Technology hardware | Director of content engineering, translation, and DevOps | 2.0 billion |
| Home improvement retail | Not stated | 13.0 million |
| Professional services | Translation management lead | 8.4 million |
| Cloud infrastructure and security | Localization manager | 1.5 million |
Blending an automated two-billion-word API pipeline with a specialized 1.5-million-word security documentation workflow skews the financial model far above mid-market reality. While composite aggregation protects interviewee anonymity, adding a sub-$1 billion cohort or volume sensitivity chart would let smaller buyers right-size the payback curve.
The engine: translation memory, machine translation and a clean six months
Translation memory and machine translation do the heavy lifting in the model. The study treats getting there as a tidy six-month build.
The composite's largest benefit is $15.2 million PV in translation cost savings across 350.7 million modeled words (90.0 million in Year 1, 115.5 million in Year 2, and 145.2 million in Year 3 at $0.09 per word). Translation memory scales from 40% to 60% of words, while Language Weaver MT scales from 20% to 40% of remaining new and fuzzy words, avoiding human translation on 76% of words by Year 3.
Three-year automation funnel
TM plus Language Weaver MT, year by year
Choose a year to see how much of the text is handled before a human translator is needed.
Translation memoryLanguage Weaver MTStill translated by people
Year 3: 145.2 million words. Translation memory covers 60% of words. Language Weaver MT handles 40% of the remaining new and fuzzy words, which is 16% of all words. Human translation is avoided on 76% of words.
| Year | Words | TM share of words | MT share of remaining words | Human translation avoided |
|---|---|---|---|---|
| Year 1 | 90.0 million | 40% | 20% | 52% |
| Year 2 | 115.5 million | 50% | 30% | 65% |
| Year 3 | 145.2 million | 60% | 40% | 76% |
The $15.2 million present-value saving covers all three years combined ($18.8 million before discounting, after a 10% downward risk adjustment). The "share of all words" figures are simple arithmetic on the study's percentages.
On the cost side, Forrester models $6.9 million PV total: $5.77 million PV for RWS products and HITL services and $1.18 million PV for a six-month implementation and ongoing collaboration (one FTE software engineer at $240,000/year plus 64 PM hours per language/year at $80/hour). That clean timeline overlooks the heavy cleanup required when migrating fragmented legacy TMs and omits quality variance across low-resource languages.
Present-value costs, $6.9M in total. Dark: RWS products and human-in-the-loop services. Apricot: implementation and ongoing collaboration.
Forrester also notes that the financial model reflects data collected prior to the March 2026 release of Language Weaver Pro, which RWS developed with Cohere to add generative AI capabilities. Adding an independent audit of pre-RWS baselines and a post-Language Weaver Pro addendum would ground the workflow savings in current buyer reality.
Downstream savings and the multipliers holding them up
Three smaller benefits rest on two flat assumptions: a 75% productivity recapture rate and a 10% launch profit margin.
Beyond direct word-cost savings, the study models $3.3 million PV in avoided international revenue loss (cutting a 15-day SLA by 30% to 50% across 20% of delayed launches at a 10% profit margin), $744,540 PV in PM efficiency (automating 25% to 55% of 9,000 to 18,150 annual projects via a self-service portal at $80/hour), and $2.5 million PV in reduced quality validation costs.
Both the PM efficiency model and the in-country linguist review savings (40% to 65% efficiency across 45,000 to 72,600 baseline hours at $50/hour) rely on a flat 75% productivity recapture rate. In practice, freed hours often absorb backlog or reduce burnout rather than generating hard ROI. Modeling tiered recapture rates and a 5% to 15% profit margin sensitivity range would make the financial case far more durable.
ROI Sensitivity and Recapture Stress-Test
Move the two assumptions and watch the benefit strip and the ROI respond. Forrester's own case sits at 10% margin and 75% recapture.
- Translation savings: $15.2M
- Avoided revenue loss: $3.3M
- PM efficiency: $0.74M
- Quality validation: $2.5M
The dashed line marks Forrester's $21.7M in present-value benefits.
| Scenario | Margin | Recapture | Benefits PV | NPV | ROI |
|---|---|---|---|---|---|
| Forrester baseline | 10% | 75% | $21.7M | $14.7M | 212% |
| Conservative | 5% | 35% | $18.3M | $11.4M | 164% |
| Optimistic | 15% | 75% | $23.3M | $16.3M | 235% |
| Low margin only | 5% | 75% | $20.0M | $13.1M | 188% |
| Low recapture only | 10% | 35% | $19.9M | $13.0M | 187% |
How this works: the avoided revenue loss scales in proportion to margin, the PM efficiency and quality validation benefits scale in proportion to recapture, and every other figure stays as Forrester reports it. The small third-party sampling saving inside quality validation is not separated out, so recapture effects are slightly overstated. These are LOCANUCU illustrations built from the study's published present values, not measured results.
Governance gains, and the missing "compared to what"
A 212% return needs a comparison. This study measures RWS against each buyer's own legacy setup and nothing else.
The report highlights five unquantified benefits: brand voice consistency, enterprise security and legal compliance readiness, spend transparency, consultative partnership, and cloud usability. Yet Forrester explicitly discloses that the study is commissioned by RWS, relies on RWS-nominated interviewees, and is not meant to be used as a competitive analysis.
- Brand voice consistency
- Security and legal compliance readiness
- Spend transparency
- Consultative partnership
- Cloud usability
Concept sketch, no data
What the study compares, and what it does not
Because the analysis compares RWS only against each buyer's fragmented legacy state, enterprise IT leaders get zero head-to-head comparison against rival translation management systems or standalone LLM pipelines. Pairing the TEI model with external BLEU or COMET quality benchmarks and turnaround comparisons would give procurement teams the competitive anchor they need.
Four questions before you sign
Treat a vendor-commissioned TEI study as a framework, not a guarantee. Test it against your own numbers.
Buyer Audit Checklist
Mark each question once you have answered it with your own data.
Does your annual word volume match the 90M to 145M composite, or the 1.5M to 13M interviewees?
The fix to ask for: a sub-$1 billion cohort and volume sensitivity model.
Have you budgeted for legacy TM cleanup and post-March 2026 Language Weaver Pro testing?
The fix to ask for: quantified migration friction, low-resource language variance and post-Language Weaver Pro performance.
What happens to payback if PM and linguist recapture is 35% instead of 75%?
The fix to ask for: tiered recapture rates and a variable margin range in place of flat multipliers.
How do RWS turnaround and BLEU or COMET scores compare against rival TMS platforms or LLM pipelines?
The fix to ask for: external competitive benchmarks alongside the proprietary TEI math.
0 of 4 questions checked.
We would welcome an updated edition from Diane Deng, Forrester Consulting, and RWS that puts these refinements to the test. For localization and IT leaders evaluating language AI platforms today, the takeaway is clear: never accept a pristine ROI spreadsheet at face value. Question the composite scale, stress-test the recapture multipliers, and demand competitive context before you buy.