Machine translation is a handy starting point for quick mock-up demonstrations. But it can jeopardise acceptance in non-English-speaking markets such as Germany. We often observe that the audience uses language quality as a proxy for trustworthiness, and many have developed a keen eye for conspicuous machine translations.

A recurring issue when translating from English into other languages is deciding between formal and informal forms of address, which English doesn’t have. Many translations take the supposedly safe route and use the formal form of address, but this sometimes came across as too distant and felt off in our studies.

What to do: Have your UI translated by human professionals who understand the service and its field well. Give them some freedom to make the translation sound natural and adapted to local standards (e.g., button labelling). Have the translation checked through user testing with real users from the target audience. Include examples of campaign messages and off-platform communications, such as emails.

You might have heard that Germans are especially sensitive when it comes to data protection, or that they tend to be late adopters of technology. The German market boasts 70 million adults and is fairly diverse. Attitudes vary a great deal depending on how you define your audience and addressable market. Among other things, this influences how the audience prefers to be addressed (“du” or “Sie”), how wary they are about data protection (this might be lower than expected), and how much explanation they need to understand the product.

What to do: Define your audience well. Which criteria are essential for someone to be a potential buyer or user? A propensity to use the service category, for example, is often a better criterion than income-related criteria. Get first-hand feedback from exactly these people in Germany, e.g., through a user test or survey.

All user testing in Germany (or in any non-English-speaking market, for that matter) should be conducted in the local language. Even participants who are highly proficient in English sometimes struggle when asked to verbalise their thoughts and feelings in a foreign language. However, getting as close as possible to the perceived user experience is so essential that cutting the budget for local moderators is a bad idea. The only exception to this rule is business software for audiences that use English as their main business language.

International teams can observe interviews conducted in the local language in one of these ways:

• Live stream with live translation into English: This is good for observing the first interviews of a study, so moderators can get feedback and the discussion guide can be adapted if necessary. Observers can switch between the original language and the translation. Tip: Make sure the translators are very good and that working hours overlap across time zones. Since live translation is rather expensive, the main part of the translation can be done with…
• Video recordings with English subtitles: These are less expensive and preserve the local flavour, as the original audio remains available. In many projects, not all interviews are observed live (because of time-zone differences or personal schedules), so watching them on demand is a good solution.

Not recommended for now: Live machine-generated subtitles (e.g. from Zoom, Teams or Google Meet). We often noticed that you can grasp the gist of the conversation, but you’ll lose too much context because of the still poor translation quality.

The screener and moderation guide should not only be translated but also adapted to local needs, e.g., to ensure participants understand the tasks and to account for different ways of asking questions in local languages.

It is essential that test materials, stimuli and mock-up prototypes be translated properly. Machine translations should always be checked by local professionals to avoid biased results caused by poor cultural or linguistic fit.

The main factors affecting the timing and cost of user research in Germany are:

• Incidence of the target group: The more niche the target group, the more time and budget are needed.
• Number of different user profiles: Sample sizes must be larger when there are more profiles. Plan for 6 to 8 participants per profile.
• In-person sessions are more expensive than video calls.
• Live translation is fairly expensive. Therefore, it makes sense to use it when live observation is really needed, e.g., during the first interviews, to give the moderator feedback.
• Reporting format: The more detailed the report, the more budget is needed.
• A standard user test in Germany will take 4 to 6 weeks from start to finish. It can be sped up if you opt for a results workshop instead of a fully-fledged PowerPoint report.

There are some misconceptions about the GDPR and what is allowed in a user study in Europe. In a nutshell, the GDPR requires participants to be informed about what you do with their data. Simply speaking, if they consent, you are free to process whatever data was agreed. Therefore, we as the vendor undertaking the study look after data protection rights and consents. Usually, this includes personally identifiable information (PII), such as first names, video recordings, selfies or cultural probes like screenshots. This means the study vendor can send artefacts such as interview recordings, private photos, screenshots, etc., to the client. It may be the client’s internal policy not to process certain types of PII (e.g., video recordings of participants). This should be clarified before setting up the study. In some cases, it makes sense to pixelate information such as credit card numbers that become visible in a recording of a participant’s shared screen. If customer contact details from a database are transferred to the vendor, a data processing agreement is necessary.

Which tasks you outsource to a local partner depends on how involved you want to be and on your available resources. You can usually ‘mix and match’ the services you need, from selective support to full-service.

Full service: bemorrow handles the design, recruitment, delivery, analysis and reporting.

Joint project: The client’s team and bemorrow share tasks and decision-making.

Selective support: bemorrow handles individual elements such as local recruitment, facilitation, translation, analysis or interpretation.

Qualitative, exploratory user tests are not primarily designed to compare outcomes between samples, but to explore audience responses and identify usage problems. However, it is possible to compare reactions by trend, e.g., between user tests in the US, the EU and Germany.

• Test participants in Germany are in our 25+ year experience usually straightforward and open with their reactions, both positive and negative, and don’t tend to sugar-coat their feedback.
• In our cross-cultural studies, general usage patterns, usage habits and usability heuristics didn't differ significantly between Western markets.
• Provided that the study methodology remains identical (e.g., how the stimuli are presented, the order of questions and the test setting), differences might be observed in the acceptance of the product, the perceived value of features, the need for understanding and specific UX preferences. These are what we are looking for in a cross-cultural test.
• When interpreting the outcomes, context differences must be considered, including the local competitive landscape, existing similar services, the tech landscape, cultural norms, expected tone of voice, etc.

Even though German participants usually say what they think and mean what they say, the researchers interpreting the findings should be aware of the local context, e.g., social norms, the competitive landscape in the category and cultural references. This usually requires being embedded in the local context. You can achieve this through a written or verbal debrief with the local UX analysts or by having the local partner run the analysis and prepare the report.