Switching localization vendors is the decision that keeps getting postponed. Not because the current provider is good enough, but because the risk feels unclear. This page documents how other localization managers made the move, what transferred, and what happened to quality during the transition.
See who made the switch ↓Every one of these is legitimate. Here is how teams that have already switched addressed each one.
Translation memories, glossaries, and terminology databases contain years of accumulated brand knowledge. The fear: something gets lost, corrupted, or left behind in a proprietary format. The reality: TMs and glossaries are standard formats (TMX, TBX, CSV). Your current LSP is contractually required to export them. If they are making this difficult, that is the portal lock-in problem in action.
Standard formats. Your data. Your right.The nightmare scenario: you switch, the new partner is not ready, and a deadline arrives with no one to handle it. The parallel model prevents this entirely. Your current LSP continues on production work while the new partner runs on a separate content stream. At no point are you without coverage. This is not a switchover. It is a gradual handoff.
Parallel model. Zero gap.Your internal team has workflows built around the current LSP's portal, file formats, and communication channels. Switching means relearning. Unless the new partner works inside your existing tools. A tech-agnostic partner integrates with Phrase, Lokalise, Slack, Notion, or whatever your team already uses. No new platform to learn. No workflow disruption.
Your tools stay. Partner adapts.A new partner does not know your brand voice, your terminology preferences, or your quality bar. Quality will drop before it stabilizes. Unless you run a pilot first. The benchmark phase (week 1-2) gives you hard data on revision rates before any commitment. Your TMs and glossaries transfer context from day one. A dedicated team retains that context because they stay on your account for years, not months.
Pilot data before commitment.Leadership wants to know why switching is worth the disruption. "Our current provider is not great" is not a business case. Parallel benchmark data is. Run the same content through both providers. Compare revision rate, turnaround time, coordination hours, and total cost of ownership. Present numbers, not opinions. The data makes the argument.
Numbers, not opinions.No leap of faith. No gap in coverage. A controlled experiment with data at every step.
Send the same content to both your current LSP and the new partner. Compare revision rate, turnaround time, coordination overhead, and how many clarification questions each one asks. No commitment. No contract. Just data.
Assign specific content types to the new partner while your existing LSP continues on everything else. Measure against your actual KPIs. This is a controlled experiment, not a leap of faith. If the data does not justify continuing, stop here.
Migrate content types one at a time. TMs, glossaries, and style guides transfer. They are your assets, not your LSP's. The new partner's dedicated team is already familiar with your brand from the pilot phase. Continuity is built in.
The same team that ran your pilot is the team that runs your production work. Brand knowledge compounds instead of resetting every 14 months. Quality metrics are published and auditable. You have one point of contact, not a portal ticket queue.
Not a case study written by marketing. A real transition story, shared with the client's permission.
Grupo Mango evaluated enterprise LSPs including TransPerfect before choosing Kobalt. Their feedback was direct: working with an enterprise provider felt like being an account number. Working with Kobalt felt like a red carpet experience. The same linguistic team has been on their account for over 12 years to date. Not because of a contract clause. Because the team is 15 people who know the brand, not thousands who rotate through it.
"Generalist LSPs will continue to disappear. Specialisation moves from aspiration to requirement."CSA Research, 10 Predictions for 2026
Your linguistic assets are yours. Your vendor's proprietary portal data stays with them. Here is the breakdown.
| Asset | Status | Notes |
|---|---|---|
| Translation memories | Yours. Transfers. | Standard TMX format. Your LSP is required to export on request. Contains years of brand-specific translation context. |
| Glossaries / termbases | Yours. Transfers. | TBX or CSV format. Industry-standard export. Brand terminology, forbidden terms, preferred phrasing. |
| Style guides | Yours. Transfers. | Documents you authored. Not vendor property. Transfer as-is to the new partner. |
| Reference materials | Yours. Transfers. | Product docs, brand guidelines, past translations. All your intellectual property. |
| Portal project history | Vendor's portal. Does not transfer. | Job IDs, internal workflow data, portal configurations. Specific to their system. You do not need it. |
| Vendor QA reports | Vendor's methodology. Does not transfer. | LQA scores, internal QA records. Based on their framework. Your new partner will establish their own auditable metrics. |
| Vendor-hosted TMS config | Vendor's platform. Does not transfer. | Workflow automations, user permissions, project templates inside their TMS. Rebuilt in your own tools (if you use a tech-agnostic partner). |
If your current LSP is making data export difficult or charging for it, that is the portal lock-in problem. Your TMs and glossaries are your intellectual property, not theirs.
Talk to a localization manager who already made the transition.
Talk to a Client Who SwitchedTypically 3-6 months from first benchmark to full operations, depending on the number of content types and language pairs. The parallel model means your existing LSP continues handling production during the entire transition. There is no moment where you are uncovered.
No. Translation memories are your intellectual property, stored in standard TMX format. Your current LSP is contractually required to export them on request. If they are making this difficult, that is the portal lock-in problem, not a technical limitation.
The parallel model prevents this. You run the new partner on a separate content stream while your current LSP handles production. You compare revision rates on the same content. If the pilot data does not meet your quality bar, you stop. There is no quality gap because there is no switchover gap.
Run the parallel benchmark first. You will have hard data: revision rate comparison, turnaround time, coordination hours saved, and total cost of ownership. Present the pilot results, not a sales pitch. The data makes the case for you.
Not from zero. Your TMs, glossaries, and style guides carry years of brand context and transfer on day one. A specialist partner with a dedicated team will internalize your brand voice during the pilot phase. The difference is that they retain it, because the same team stays on your account for years instead of rotating every 14 months.
Yes. Most enterprise LSP contracts have 30-90 day exit clauses. Start the pilot during the notice period. By the time the contract ends, you have months of performance data from the new partner. You can also run both vendors in parallel indefinitely if the contract requires it.
Ask about specific coverage, not theoretical reach. Kobalt operates across 20+ markets and handles 80,000+ requests per month with a 98.7% on-time delivery rate to date. The question is not whether a partner can cover your languages, but whether they maintain quality and team consistency across those languages.
Yes. Grupo Mango evaluated TransPerfect and other enterprise LSPs before choosing Kobalt. Their feedback: working with an enterprise provider felt like being an account number, while working with a specialist felt like a red carpet experience. Same team for over 12 years to date. Brand name shared with permission.
Not a sales call. A conversation with a localization manager who went through the same evaluation you are doing now. We will connect you directly.
Prefer email? ricard@kobaltlanguages.com