.svg)
.svg)
.svg)
.svg)
.svg)
Reusing an inherited vendor evaluation when shopping for translation providers is standard practice, but it’s likely time to update that process. Call it a request for proposal (RFP), a vendor comparison, or a renewal review: it's the document you send to providers asking how they'd run your translation work and what it would cost, and starting from the last one saves time.
In many cases, the problem is that the market has changed faster than that document has. The rise of artificial intelligence (AI) has shifted what separates one provider from another, and a document written before that shift won't ask the questions that now matter most.
It has also stopped being one team's document. Marketing owns the campaigns and the brand voice that need to land the same way in every market, product owns what ships in the app, support owns the help center, and compliance owns anything carrying regulatory risk. Localization may run the process, but the answers in that document are important to many stakeholders.
For a sense of where to start, teams often turn to independent analyst research. The Forrester Wave™: Localization Services, Q3 2026 is one example. In it, Forrester notes that "AI translation has made multilingual content easier to generate, but harder to govern and trust." The report also finds that "the market is diverging between providers that use AI to redefine what an LSP can do from those primarily using it to improve established services." LSP is the industry's shorthand for a language services provider.
The results share how 11 providers were assessed against 27 criteria. This can be helpful in putting together questions to ask the providers on your own short list. In fact, the report notes: "We intend this evaluation to be a starting point only and encourage clients to view product evaluations and adapt the findings based on their priorities."
Here are five of those criteria we think are worth building into your next evaluation, along with questions we recommend asking.
1. Quality definition and assessment
Most evaluations ask providers to describe their quality process. However, the more revealing question is what happens before a single word moves.
Ask: How will we agree on what good looks like for each content type, and what happens when we disagree?
A landing page and a warranty disclaimer don't share the same quality bar. A provider that treats them the same may over-serve one and under-serve the other. You want the provider that shows up with a framework and details how to apply it.
2. Workflow orchestration and AI acceleration
A campaign that needs to launch in twelve languages at once is only as fast as its slowest handoff. The parts worth asking about aren't the ones that work fine when everything goes smoothly, but rather the ones that depend on someone remembering to move a file, approve a step, or notice something has stalled.
Ask: Walk us through what happens between content being ready and content being live in every market. Which of those steps run on their own, and which ones wait on a person?
The honest answer separates providers and helps you understand just what it takes to get a global campaign live on a deadline.
3. Model selection, testing, and customization
This is the one many teams can't answer about their own current setup. Which model translated last quarter's release notes, or the campaign that ran in Germany?
Ask: How do we know which model to use for each use case, and how can we make sure it's customized to sound like our brand everywhere?
A good provider doesn't just plug in a model and hope. They help you choose the right one for each use case, customize it to sound like your brand, catch issues like bias, hallucination, or terminology drift before they reach a reader, know what they do to fix it when it happens, and show you the data behind how well it's actually performing.
4. Data, model, and AI risk controls
Every provider will tell you they take AI risk seriously. What's harder to get is the paperwork that proves it: where your content is stored and for how long, which models are allowed to touch it, and who has to sign off before anything higher-risk goes out the door.
Ask: What documentation can you share with our compliance team today that backs this up?
That means retention, encryption, and access controls; isolation for sensitive or regulated content; limits on which models can touch your data, including whether any of it can be used to train them; defenses against threats like prompt injection; and a clear audit trail of what touched your content and who signed off on it.
5. Measurement and communication
Localization reporting tends to measure activity rather than outcome. Words translated, turnaround time, and cost per word are all easy to produce, and none of them answer the question of why the program is funded.
Ask: Six months in, what will you be able to show us that connects this work to something the business already measures?
The answer should name a metric that already exists somewhere else in the company, whether that's pipeline in a market you've just opened, support tickets deflected in a given language, or how long it takes a product update to reach customers in every language you support.
Two more considerations
Self-serve and distributed translation governance: Somewhere in your company a support team may be pasting text into a free AI tool, and a regional marketer may have given up waiting and translated the campaign themselves. None of it is in your evaluation, and all of it goes out under your brand. Ask how a provider serves the people who will never file a ticket.
Linguistic asset curation and governance: Your translation memory and glossary are the only things in this relationship that you own and that get more valuable over time. Ask who maintains them, how they improve and how they’re governed, and what condition they're in on the day you leave.
Now ask Smartling the same questions
We'd invite you to ask us the same five questions. Smartling was named a Leader in The Forrester Wave™: Localization Services, Q3 2026. The report notes that Smartling "excels at turning fragmented enterprise localization into an automated, measurable production model," and that its "strongest capabilities sit at the intersection of integrated workflows across decentralized systems, quality governance, testing, model control, linguistic asset management, and AI risk management, in which it shows greater maturity than most peers."
Access the report
Read The Forrester Wave™: Localization Services, Q3 2026 to see how all 11 providers were assessed, criterion by criterion.
And if you're conducting an evaluation right now, start with our sample translation RFP for the operational ground, then add the five questions above. Send them to every provider on your list, including Smartling.
Access The Forrester Wave™: Localization Services, Q3 2026 → https://www.smartling.com/ebooks/forrester-wave-localization-services
Objectivity statement
Forrester does not endorse any company, product, brand, or service included in its research publications and does not advise any person to select the products or services of any company or brand based on the ratings included in such publications. Information is based on the best available resources. Opinions reflect judgment at the time and are subject to change. This report is part of a broader collection of Forrester resources, including interactive models, frameworks, tools, data, and access to analyst guidance. For more information, read about Forrester's objectivity.
