Tu or Lei? How DACH Brands Address Italian Users

Data and reports · Version 1.0 · Published October 8, 2026 · Updated October 8, 2026 · by Alessio Di Rubbo

Italian has two ways to address a reader: the informal tu and the formal Lei, a bit like the French tu and vous or the German du and Sie. When a brand based in Vienna, Berlin or Zurich publishes its Italian version, it has to pick one. German makes the same distinction, but the conventions do not line up with the Italian ones, and a wrong choice is heard at once. The tu where Lei is expected sounds brisk; the Lei where the market uses tu sounds distant. We set out to measure what DACH brands actually do, rather than rely on impressions.

The method

We analysed 20 websites of brands based in Austria, Germany and Switzerland, all in their Italian version, on 8 October 2026. For each site we looked at seven user touchpoints: cookie banner, homepage, product page, cart, 404 page, newsletter form and FAQ or support section.

Collection is automated, with Playwright and Chromium, so every point is read the same way across all sites. Brand names appear neither in this article nor in the downloadable data [names not published]: what matters is the aggregate behaviour, not a report card for any single brand.

There are 140 touchpoints. Ten do not exist on the site in question (a SaaS product, for instance, has no conventional cart), which leaves 130. The tool read 90 of them, or 69 percent. In 73 cases, 56 percent, the text allowed a register class to be assigned; in the other 17 it was too thin, and we marked them as uncertain rather than force a classification. The remaining 40 points were not read, for technical reasons described below.

What the tu says

The main result is clear-cut. Across the 90 points read there is not a single case of Lei or voi. Of the 73 points with a class assigned, 58 are in tu, 13 are impersonal and 2 are mixed. Nineteen homepages out of twenty address the user with tu; the twentieth is served in English and was not classified. Among the cookie banners, eight of the fourteen that were read use tu.

The consistency finding is just as clear: zero sites waver. No site switches between tu and Lei from one page to the next. Those that choose tu, and they all do, keep it at every readable touchpoint. This holds for the 13 e-commerce sites as much as for the 7 SaaS or app sites: there is no register difference between the two groups.

The table gives the counts per touchpoint. No row contains a Lei or a voi.

A file with the same counts can be downloaded from the sources section below. It contains aggregate data per touchpoint only.

The cookie banner: the weak spot

If consistency breaks anywhere, it breaks in the cookie banner, and the cause is not the brand but the banner vendor. In two cases the body text is impersonal, with "the user" or the brand's "we" as subject, while the button is an imperative, "Accetta tutti i cookie". Both banners run on the same consent management platform, OneTrust, the most common system in the sample with six of the fourteen banners identified. This is not a register shift the brand decided on: it is the vendor template placing a neutral body next to a default button label. The reader, though, perceives it as text that changes tone halfway through.

Alongside these, three cookie banners on three other e-commerce sites build the sentence with the brand's "we" or with the brand in the third person, using neither tu nor Lei. On all three, the homepage of the same site is in tu: the jump is not a register choice by the brand but a consent text written under different rules. A fourth case is less clear-cut. In the newsletter form of another e-commerce site the buttons are short and carry no pronoun, for example "Iscriviti". We classified it as impersonal, but as an imperative it can also be read as tu.

In total, 13 occurrences are classified as impersonal. Nine concern pages served in a language other than Italian, and the method classifies them that way by rule. Only four are genuinely impersonal: the three banners and the newsletter case just described.

The cart and the FAQ: missing Italian, not a register choice

There is a second pattern that looks like a register problem at first and is in fact incomplete localisation. Thirteen touchpoints, spread across seven sites, serve no Italian text despite an Italian address or setting. The cart is the most frequent case, with four points. Next come four 404 pages in mixed or non-Italian language, two FAQ pages, two product pages and one homepage. Here there is neither tu nor Lei: there is a page speaking the wrong language.

The recurring pattern is that cart and checkout sit on a separate domain or a local path, often a separate shop, that has no Italian version. The browser shows an Italian address up to the last step, and right there, before payment, the language changes.

The FAQ results need caution. Five of the support URLs we tried return a 404 and two more lead to pages in German or English. The addresses were tried according to a pattern and not reached by navigating the site's menus, so a 404 can mean a missing page as much as an address different from the one we tried. The finding points to a weak spot to check case by case, not to a measurement.

Limits of the survey

The figures should be read with these limits in mind.

• Newsletter: 12 sites out of 20 do not expose the form in a way that can be captured automatically, because of popups, widgets or loading after scroll. The sample is small and not representative.
• Cart: on 7 of the 13 e-commerce sites the cart cannot be opened without first choosing a variant, such as size or colour. This is a technical limit of the tool, not a register choice.
• Cookie banner: 6 sites out of 20 do not expose a readable banner with standard selectors.
• 404 pages: 11 out of 20 are marked uncertain, because the text is too thin to assign a register.
• Classification: where the banner body is impersonal and the button is an imperative, we used mixed only if the button was clearly tu. Another reviewer might move a few rows between impersonal and mixed.

What it means for localisers

For a localisation manager or content strategist working on a DACH site into Italian, these figures turn into four practical points.

• Tu is the standard, not a risk. For a DACH brand, using tu in Italian is the market norm, not an exception to justify. The question "tu or Lei?" already has a de facto answer for e-commerce and digital products.
• Check the cart and the checkout, not just the shop window. That is where Italian goes missing most often, and where the customer decides whether to complete the purchase. One pass through the flow to payment is worth more than ten re-reads of the homepage.
• Treat the cookie banner as part of the copy, not as technical trimming. The vendor template can introduce a tone shift no one wrote. Check that the body and the buttons of the CMP use the same register as the rest of the site.
• Do not assume the glossary is enough. A consistent terminology glossary covers neither the pages left in German nor the default consent labels. Register consistency is checked on the published site, point by point, not only in the translation files.

Conclusion

The most useful finding here is not that tu beats Lei: it is that it does so without exception and without wavering, across twenty sites and two different business models. For a DACH brand entering Italy, the real question is not which register to choose, but whether that register holds at every touchpoint, including the ones no one re-reads: the cart, the consent banner, the page that appears when something goes wrong. That is where Italian localisation is completed, or interrupted.

Sources and further reading

• Aggregate survey data from 8 October 2026 (CSV)
• How I assess the quality of Italian text: audit methodology
• Do AI systems cite different sources in Italian and German?
• How much does an audit of your Italian website cost?