Localization Pricing

SaaS Localization Pricing

Per-word pricing was designed for documents, not product strings.
What pricing model actually works for continuous localization?

Compare per-word, per-string, and retainer pricing models for SaaS localization. Find the model that matches your release cadence, not your vendor's billing preference.

See why per-word fails ↓

Five reasons per-word pricing fails for SaaS products

Per-word pricing works for documents with a beginning and an end. Product strings are living content that changes with every sprint. The pricing model needs to match.

Problem 01

String updates cost the same as new strings

Change one word in a UI string and you are charged for the full string again. In continuous delivery, strings get updated constantly. Per-word pricing penalizes iteration.

Iteration tax
Problem 02

Context switching is not priced

A translator switching between your onboarding flow, error messages, and settings page wastes time re-establishing context. Per-word pricing does not account for this fragmentation. You pay the same rate whether the translator has context or not.

Hidden inefficiency
Problem 03

Platform fees eat into the budget

Your TMS vendor charges per-seat, per-word, or per-project. Your translation vendor charges per-word. You are paying two vendors for overlapping services. The platform fee alone can equal 20 to 30% of translation cost.

Double billing
Problem 04

Sprint-aligned delivery does not exist

Per-word vendors batch work. Your sprints do not wait. The gap between "strings ready" and "translations delivered" creates a release bottleneck. You either delay the release or ship in English only.

Release bottleneck
Problem 05

No one owns the end-to-end

The TMS vendor owns the platform. The translation vendor owns the linguistic work. Your developer owns the integration. When something breaks, three parties point at each other. Coordination cost is invisible but real.

Ownership gap

Three pricing models for SaaS localization

Each model makes trade-offs between predictability, flexibility, and alignment with continuous delivery. The right choice depends on your release cadence and string volume.

Per-word + TMS platform fee

Designed for documents, not products.

Pay the TMS vendor for the platform, pay the translation vendor per word. Double billing for overlapping functions. String updates charged as new work. No alignment with sprint cadence.

Works for documentation with a clear scope and deadline. Fails for product strings that change with every release. The more you iterate on your product, the more you pay.

Works for: static documentation and one-time translation projects.
Per-string flat rate

Better alignment, limited flexibility.

Better alignment with product development than per-word. But all strings are treated equally: a 3-word button label costs the same as a 50-word tooltip. No quality differentiation between critical UI copy and internal strings. Limited flexibility for content type variations.

Simplifies budgeting compared to per-word, but does not solve the coordination, context, or delivery cadence problems. Still typically paired with a separate TMS fee.

Works for: teams with stable string counts and infrequent updates.

Total Cost of Ownership: what continuous localization actually costs

The per-word rate is never the full cost. Platform fees, coordination time, and delivery gaps add up. Here is what the total picture looks like.

The platform tax

Most SaaS teams pay for the same service twice.

The typical stack: TMS license (annual fee) + translation vendor (per-word) + developer integration time + PM coordination time. Four cost lines for one outcome: translated strings in production.

With Kobalt: one partner handles translation, tool integration, and coordination. You keep your TMS if you want it, or we work directly with your repo. No forced platform. No duplicate billing.

One partner No platform fees No forced migration
Predictable spend, continuous delivery

Retainer model means no invoice surprises.

Predictable monthly cost. Strings translated as they are merged, not batched. Same team who knows your product, no re-onboarding. Integration with your actual tools: GitHub, Slack, Phrase.

80,000+ requests per month across all clients. <60 seconds request-to-production. The retainer absorbs the natural rhythm of product development: quiet weeks and launch surges alike.

80K+ requests/month <60s to production Predictable cost
<60s
Request to production
80K+
Requests/month
98.7%
On-time delivery to date
<1%
Revision rate to date
0
Platform fees
12+
Years, longest client to date

What's always included in the price

No line-item surprises. No technology access fees. No rush surcharges. Everything your team needs to ship localized product continuously.

Always included

Dedicated product team

Same linguists who know your product, your UI patterns, and your users. Sprint-aligned delivery. Sub-minute response as standard. Your team builds product context over years, not months.

Always included

Tool integration

GitHub/GitLab PR workflow. Slack communication. Your TMS (Phrase, Lokalise, or none). No "technology access" fees. No forced platform migration. We plug into your stack, not the other way around.

Always included

Quality and operations

<1% revision rate. Terminology management for product consistency. Performance reporting. Scalable for launch surges. ISO 9001 and ISO 17100 certified processes behind every string.

Want pricing built for continuous deployment?

Get a proposal that scales with your release cadence.

Request a Custom Pricing Proposal

Frequently asked questions

How does the retainer model work for SaaS?

A monthly retainer covers your dedicated linguistic team, sprint-aligned delivery, tool integration, and coordination. A variable component adjusts for actual string volume. You get predictable monthly costs instead of invoice surprises, and strings are translated as they are merged rather than batched. The same team learns your product context over years.

Do we need to keep our TMS or can you replace it?

You choose. Kobalt integrates with your existing TMS (Phrase, Lokalise, Crowdin, or others) or works directly with your code repository via GitHub or GitLab PR workflows. There is no forced platform migration and no technology access fees. If your current TMS works, we plug into it. If you want to simplify, we can work without one.

How do you handle continuous string updates without per-word inflation?

Under the retainer model, string updates are part of the continuous delivery workflow. Changing one word in a UI string does not trigger a full re-translation charge. Your team iterates freely on product copy without worrying about translation cost inflation. The retainer accounts for the natural rhythm of product development.

What happens during a big launch with 3x normal volume?

Volume surges are handled within the retainer structure with a variable component for actual volume. Your dedicated team scales for launch periods because they already know your product context. There are no rush fees or emergency surcharges. Kobalt processes 80,000+ requests per month across all clients, so capacity is never the bottleneck.

How does pricing change when we add a new language?

Adding a new language adjusts the retainer to reflect the additional linguistic team and volume. There is no per-language platform fee or setup charge. The integration work (connecting your repo, TMS, or CI/CD pipeline) is done once and applies to all languages. Scaling to new markets is incremental, not exponential.

Can we start with a pilot sprint before committing?

Yes. A pilot sprint typically runs 2 to 4 weeks with a defined scope of strings and languages. You see the full workflow: repo integration, translation delivery, PR merging, and quality. The pilot uses real product content so you can evaluate speed, quality, and team fit before any long-term commitment.

Do you charge for string updates to previously translated content?

Under the retainer model, string updates are included in the continuous delivery workflow. You are not charged per word for updating a button label or revising a tooltip. This is the fundamental difference from per-word pricing: iteration is expected and accounted for, not penalized.

How quickly can you provide pricing?

Share your tech stack, string volume, target languages, and release cadence. Kobalt delivers a transparent pricing proposal within 48 hours, structured for continuous delivery rather than batch translation. The proposal includes a breakdown of what is covered under the retainer and what falls under the variable component.

Request a custom pricing proposal

Tell us your tech stack, string volume, and release cadence. We will send you a transparent pricing proposal within 48 hours — structured for continuous delivery, not batch translation.

Prefer email? ricard@kobaltlanguages.com