Bilingual Welsh checkout testing for online stores selling into Wales

What to take away

Welsh language expectations for stores selling into Wales

Shoppers in Wales do not all speak Welsh, but a meaningful share expect to see it. The 2021 Census found 538,300 Welsh speakers aged three and over, about 17.8% of the population. In Cardiff, the proportion is lower than in west Wales, yet the city has a growing Welsh-medium education sector and widespread bilingual signage.

For an online store, that means a Welsh language option signals respect rather than novelty.

Language choice on a product page is not the same as a full translation. Many Welsh speakers browse in English and switch to Welsh for reassurance at checkout. Your job is to make that switch easy and obvious. If the toggle sits in a footer or behind a settings menu, most shoppers never find it.

ONS cultural identity data shows that Welsh identity is not tied purely to language ability. People who identify as Welsh but do not speak the language still respond to bilingual presentation.

That is why a bilingual checkout can build trust even among English-dominant shoppers in Cardiff and Newport. The cultural identity data records Welsh identity separately from language ability.

For stores selling into Wales, the practical question is where language choice belongs in the purchase path. Put it where the shopper feels risk: delivery, returns, payment and contact. Those are the moments when clarity matters most. A language toggle next to the basket total does more work than one on a homepage banner.

What Welsh language standards mean for online services

The Welsh language standards are a set of duties made under the Welsh Language (Wales) Measure 2011. They apply to named public bodies, such as local councils, health boards and Welsh Government sponsored bodies. A limited set of private and third sector organisations that deliver public services is also covered. Most private online stores are not covered.

The duties come from the Measure, not from other UK legislation such as the Parliamentary Voting System and Constituencies Act 2011.

For the organisations that are covered, the standards include duties to provide services and answer correspondence in Welsh. Websites, apps and self-service tools must also be available in Welsh, and Welsh must not be treated less favourably than English.

You do not need to be a public body to feel the commercial effect. Larger retailers and service providers often adopt bilingual checkouts to align with public sector procurement or to match competitor behaviour.

For an online store, the standards matter in three ways. First, they normalise bilingual service in Wales. Second, they set expectations for how language choice should be offered, not hidden. Third, they create a benchmark that shoppers may apply to private stores even when the law does not compel them.

If you sell into Wales and hold a public sector contract, check whether your checkout meets the same bilingual standard as your tender documents. A mismatch between a bilingual proposal and an English-only checkout undermines trust. The Competition and Markets Authority expects clear, accurate information for consumers, and language clarity is part of that.

The Information Commissioner's Office also expects fair processing of personal data. If you collect Welsh language preference, treat it as personal data under UK GDPR and the Data Protection Act 2018. Store it only as long as needed and explain why you are collecting it.

How language choice placement affects trust at checkout

Trust at checkout is fragile. Shoppers weigh delivery cost, returns policy and payment security in seconds. A Welsh language option placed early tells the shopper that the store understands Wales. A toggle placed late, after the payment step, tells them the opposite.

Test three placements. First, a persistent header toggle visible on every page. Second, a language step inside the checkout flow, before address entry. Third, a contextual link near the delivery estimate.

Each placement changes perceived effort. The header toggle is easy to find but easy to ignore. The checkout step forces a decision but can add friction. The contextual link feels helpful but may be missed.

Language choice placement also affects how shoppers read your trust signals. A Welsh returns policy next to an English payment form can look unfinished. A fully bilingual checkout, with consistent terminology for delivery and returns, reads as competent. That competence carries over to perceptions of card security and data handling.

Use plain Welsh for transactional language. Terms like "ychwanegu at y fasged" and "talu nawr" are widely understood. Avoid literal translations of marketing slogans. A clumsy translation damages trust more than no translation at all. If you cannot review the Welsh with a first-language speaker, keep the Welsh to functional strings and leave persuasive copy in English.

To understand which trust signals matter most to UK shoppers, see our guide to trust signals, which covers the evidence you can gather without inventing conversion lifts.

Test designs for bilingual product pages

A bilingual product page test must isolate language from layout. If you change the language and the button colour at the same time, you learn nothing. Start with a controlled test that varies only the language of the product title, description and calls to action.

Use a simple A/B design with four arms if traffic allows:

Segment results by device and by whether the shopper arrived from a .wales or .cymru domain.

For the bilingual arms, keep the toggle in the same position across both. The only difference should be which language loads first. That tells you whether Welsh-first presentation changes engagement or completion.

Before you test, review your product copy for translation quality. Our guide to conversion copywriting explains how to write product pages that sell in the UK market. Apply the same principles to Welsh, but keep sentences shorter and avoid idioms.

If you sell a wide catalogue, test on a representative sample of ten to twenty products. Choose items with steady traffic and similar price points. Testing on bestsellers only will skew results toward shoppers who already know your brand.

Measuring completion differences between Welsh and English checkouts

Completion rate measurement starts with a clear definition. For this test, completion means a paid order, not a checkout start. Track these events separately: product view, add to basket, checkout start, address submitted, payment submitted, order complete. Without that funnel, you cannot tell whether Welsh language affects interest or only the final step.

Set up your analytics to record the language variant as a user property. That lets you compare Welsh-first and English-first cohorts without relying on session-level joins. If you use a consent management platform, make sure language preference is not blocked by analytics consent, or you will lose the very data you need.

Calculate completion rate as orders divided by checkout starts for each variant. Report the absolute difference and the confidence interval. A two percentage point difference on a small sample is noise.

Work out the sample before you launch and stop only when you reach it. To detect a two percentage point difference from a 3% baseline, at 95% confidence and 80% power, you need roughly 1,500 orders per variant.

That figure rises as the baseline falls or as the effect you want to detect gets smaller. On a mid-traffic store, expect four to six weeks.

Watch for interaction effects. Shoppers who choose Welsh may also be more likely to use a specific payment method, such as a local card or a digital wallet. Segment by payment method to see whether the language effect is real or a proxy for something else.

ONS population and community data can help you size the Welsh-speaking market before you commit to a test. The people and community figures cover local authority breakdowns.

For a deeper look at how to design research that starts with the shop visit, read our guide to audience research.

Cardiff store examples and what to copy

Cardiff has a mix of independent retailers and national chains. The independents that handle bilingual checkout well tend to share three habits. They place the language toggle in the header on every page. They keep Welsh and English strings in the same template, so prices and delivery dates never break. They train staff to answer queries in either language.

Copy the header toggle first. It is the cheapest change with the clearest effect on perceived effort. Next, translate the checkout error messages. An error message in English after switching to Welsh feels like a broken promise. Then translate the order confirmation email. That email is the last trust signal before delivery.

One Cardiff store selling homeware found that Welsh-first product pages increased add-to-basket rate but not completed orders. The reason was delivery cost clarity. The Welsh page did not explain that delivery to west Wales took longer. Once the store added a bilingual delivery estimate, completion improved. The lesson is that language alone does not fix a weak delivery promise.

Another Cardiff example: a bike shop added a language toggle to its basket page. Checkout starts rose, but completed orders did not change, because the payment step was still English only.

Once the store translated the payment button and the card error messages, completion followed. The size of that change depends on how much of the funnel is still English and on how many shoppers are Welsh-first.

In Cardiff the Welsh-speaking share sits below the national 17.8%, so the measurable effect is smaller than in west Wales. The pattern is consistent: translate the whole critical path, not just the top of the funnel.

If you need to survey Welsh shoppers about their preferences, our guide to survey tools sets out four routes for store research.

Common mistakes in bilingual checkout testing

Testing language and design at the same time. If you change the language and the layout, you cannot attribute the result. Keep the design constant and vary only the language.

Hiding the language toggle. A toggle in the footer is invisible to most shoppers. Put it in the header or at the start of checkout.

Translating only the homepage. The checkout is where trust is won or lost.

Using machine translation without review. Awkward Welsh undermines trust. Have a first-language speaker review transactional strings.

Ignoring sample size. Set the sample before launch, using the thresholds in the measurement section.

Forgetting consent. Language preference is personal data. Record it lawfully and do not block analytics consent for the variant.

Assuming all Welsh speakers want Welsh. Some prefer English for commerce. The point is choice, not compulsion.

Common questions