Mobile friendly website: can a phone visitor reach a Belgian small business?

In November 2026, 69.1 percent of Belgian small business homepages showed a phone visitor no button to call, email, book or get in touch in the first screen, even after the cookie notice was dismissed. We measured that in a browser on 594 sites. In the HTML of 21,973 homepages, 48.2 percent have no tap-to-call number.

The first screen is the shop window

Someone who opens the site of a restaurant, a hairdresser or a plumber on their phone usually wants one thing: a table, an appointment, the opening hours, or a phone call. On a common smartphone the first screen is 390 by 844 pixels, and for that visitor it is the whole shop window. If the button is not there, they have to scroll, open a menu or copy the number by hand.

Our question was simple: can a visitor on the homepage call, email, book or find the way straight away? We answered it twice. First we read the HTML of all 21,973 working small business homepages from our November count, looking for tap-to-call links, email links, WhatsApp links, booking tools, opening hours, addresses and maps. Then we opened 594 homepages in a real browser at phone size, dismissed the cookie notice without accepting anything, and recorded what was in the first screen.

We did not measure how many customers a business loses this way. What we did measure is how many steps stand between the visitor and what they came for. How fast these pages load is covered in our study of website speed in Belgium.

After the cookie notice, rarely a button

On 41.5 percent of homepages a notice appeared first, nearly always about cookies. At the median it covered 37.7 percent of the screen, and on 13.9 percent of sites half or more. After we rejected or closed it, 30.9 percent still showed a button to call, email, book, ask for a quote or go to the contact page. The other 69.1 percent offered no way to reach the business in the first screen.

Shares are weighted by each sector's weight among the 23,676 working sites; the last column is the raw count. A notice that does not leave without a yes has no reject or close button, or still covers at least a tenth of the screen after the click. On 21 of the 259 sites with a notice, a button only became visible once the notice was gone. Our cookie scanner shows what your own site loads before a visitor consents, and do I need a cookie banner explains when you need one.

A visible number is not a call button

The HTML of 72.6 percent of homepages contains a phone number. Only 51.8 percent turn it into a tap-to-call link, a tel: link that makes the phone dial straight away. On 24.5 percent the number is plain text, so a visitor who wants to call has to copy it or type it over.

According to Apple's own documentation, Safari on iOS turns any string that looks like a phone number into a link by itself, unless the site switches that off with format-detection telephone=no. That line was on 14.6 percent of homepages. On 2.2 percent of all homepages, 481 sites, both came together: a number without a link, and an iPhone told not to link it. Other browsers each behave differently, so a tel: link is the only route that works everywhere.

Base: 21,973 working homepages with readable HTML. Another 1,703 working sites only show content through JavaScript; they are outside this table but were part of the browser sample. WhatsApp is still a rare contact route on a homepage; what business messages through WhatsApp cost is covered in WhatsApp Business pricing.

The platform plays a part

The platform behind a site helps decide whether a number becomes a link. On Wix sites the number is rarely plain text; on WordPress and Squarespace it is much more often. One detail stands out: 98.9 percent of Wix homepages carry telephone=no in their code, and still they have a tap-to-call number more often than the other platforms, because the number sits in the page as a tel: link.

Shopify sites are mostly web shops, where ordering is the main action and calling matters less. To see which platform your own site runs on, try our CMS detector. A link with a platform does not prove the platform is the cause: whoever picks a platform often picks a builder and a theme with it.

By kind of business: does the main action work?

A restaurant has no use for a quote button, a contractor little use for opening hours. So we gave each kind of business a main action, the reason a phone visitor usually comes, and checked whether the homepage delivers it. For restaurants, accommodation, healthcare practices, hairdressers and sports venues that is a booking tool, a booking button, a tap-to-call number or WhatsApp. For cafés, shops, pharmacies and opticians it is opening hours together with an address, a map or a route link. For construction and business services it is a tap-to-call number, a quote button or WhatsApp; for garages a link to an appointment page also counts.

Shops score lowest: only 44.8 percent show their opening hours together with an address, a map or a route link, which is exactly what a customer on the street looks for. Restaurants let you reserve or call in 80.1 percent of cases, cafés show hours and route in 50.2 percent. Pharmacies lead with 83.5 percent showing hours together with an address or a map. In healthcare the number is plain text more often (31.1 percent) than in business services (18.3 percent). What a site for your trade can look like is shown on our pages on a restaurant website, hair salon website design and a website for contractors.

What the browser saw by sector

In the browser the sectors differed too, but with about 85 sites per sector the margin of error is too wide to rank them. We show the figures so you can place your own sector, not as a league table.

A direct button calls, opens WhatsApp or opens a booking screen. A booking widget that loads inside an embedded frame was not seen as a button by our browser, so for restaurants with such a widget the real share is somewhat higher. Which booking tool fits your business is covered in our guide to a booking system for your website.

By province

West Flanders has a tap-to-call number on the homepage most often (56.6 percent), Brussels least often (46.3 percent). Contact in one tap is lower in Brussels as well: 62.6 percent, against 72.6 percent in West Flanders. Between the other provinces the spread is small.

The main action works less often in the Walloon provinces than in the Flemish ones. We do not read that as a difference between business owners: our address detection misses street names that start with a lower-case letter, such as rue or chaussée, and so undercounts addresses in Wallonia and Brussels. How findable businesses in Ghent are is the subject of our Ghent business websites study.

Buttons too small for a thumb

A button you can see also has to be one you can hit. The accessibility guideline WCAG 2.2 asks for a target of at least 24 by 24 CSS pixels; Apple's design guidelines recommend 44 by 44 points. Of the 144 sites with a direct button in the first screen, the largest such button was under 24 pixels on its shortest side on 30 sites (20.8 percent), and at least 44 on 65 sites (45.1 percent). Call buttons are smaller: on the 79 sites with a call button in the first screen the median was 25 pixels, and 38.0 percent stayed under 24. Often it is a small icon in the top bar.

On top of that, 15.4 percent of homepages block zooming, with user-scalable=no or a maximum scale of 1. A visitor with poor eyesight cannot enlarge the text. Our accessibility checker flags that line and most undersized buttons. Whether the law expects an accessible site from your business is covered in does the European Accessibility Act apply to my website.

Opening hours and the way to your door

For a shop or a café, the route is the main action. 71.0 percent of homepages give an address, 43.2 percent the opening hours, and 39.0 percent both. A map appears on 17.8 percent, a link that opens directions on 22.2 percent. Local business schema, the structured data that tells Google which business, which address and which hours, was on 19.1 percent; the schema markup entry explains what it is.

Someone searching in Google Maps sees your business profile first and your site after that, so hours and address belong on both, and they should match. Setting up that profile is covered in create a Google Business Profile, and what to do next to be found locally in our local SEO tips. A Google reviews widget appears on only 3.1 percent of homepages.

Booking online: who already does

34.3 percent of homepages have a booking tool or a button to reserve or make an appointment. Looking at the right kind of tool or button for each business, you can reserve a table online at 63.0 percent of restaurants and book a room at 68.2 percent of accommodation sites. An appointment can be made online at 61.5 percent of healthcare practices and 65.6 percent of hair and beauty salons. The tool we found most was Zenchef, on 824 homepages, followed by Cubilis from Stardekk (338), Tablefever (207), Tablebooker (167) and Salonkee (123). Those are counts, not a verdict on the tools.

A booking tool only helps if the phone visitor can see the button. The website speed test shows whether the widget slows your page down.

What changed since October

In October, the state of Belgian business websites 2026 measured 586 Belgian small businesses, 427 of them with a working site, plus 178 businesses in Ghent. That study looked at the basics for Google; this one looks at what a phone visitor can do. Four figures overlap:

• WhatsApp link4.4 percent in October, 4.2 percent now. About the same.
• Local business schema15.0 percent in October, 19.1 percent now. The gap is under 5 percentage points, so about the same.
• Zoom blocked15.5 percent in October, 15.4 percent now. The same.
• Fits a phonein October 92.0 percent fitted, with a viewport line on 96.5 percent. Now that line is on 89.6 percent. The gap comes from the method: in October we read the page after it loaded in a browser, now the HTML as the server sends it. A line added later by JavaScript does not count now, and the 1,703 sites that only load through JavaScript fall outside. In the browser, 1.5 percent of sites made you scroll sideways.

The October sample was small and drawn per province; this measurement covers every working site. Where the figures agree, they back each other up.

What does this mean for your business?

Take your own phone, open your homepage and look at the first screen. Five things you can check right away:

1. Can visitors tap your number? Tap it. If your phone does nothing, the tel: link is missing, which is a small change to your site.
2. Is your main action at the top? A restaurant shows a button to reserve, a hairdresser a button to book, a shop the hours and address. A button at least 44 pixels high is easy to hit with a thumb.
3. Can the cookie notice be rejected? Click reject and see whether the notice really goes, and whether it covered half your screen. Our cookie scanner shows what loads before consent.
4. Are hours and address right? Put them on your homepage and in your business profile, with local business schema. Our SEO checklist covers the rest.
5. Do customers find your reviews? A Google review link generator gives you one link to share after a visit.

Building a new site? Put these requirements in your brief; our website brief template has a field for them. If you want someone to sort it out, we build a custom website where the first phone screen carries the main action, or help with local SEO so the right customers find that page.

How we measured

The frame is the list of 31,123 domains of Belgian businesses with their own website in OpenStreetMap, without governments, associations and chains. Round 7's count, in the first week of November 2026, fetched the homepage of every domain and found 23,676 that work (76.1 percent). 532 sites showed a bot check; we did not get around it, and they count as dropouts.

Static measurement. On the HTML of the 21,973 working homepages that show content without JavaScript, we ran detectors for tel: links, phone numbers in text, mailto, WhatsApp, links to a contact page, forms, 94 booking tools, booking and quote buttons, opening hours, addresses, maps, route links, local business schema and its type, review widgets and the viewport line. A detector only counts what is visibly in the page, not what sits inside a script.

Browser measurement. From the working sites we drew 86 homepages per sector, 602 in all, from the provisional base of the same count; none of them dropped out of the final base. A Chromium browser opened them at 390 by 844 pixels, at most four at a time. If a notice appeared, we clicked reject, necessary cookies only, or close, never accept. We measured the screen before and after that click. 2 sites returned an error, and 6 showed an error page after the rejection; those 8 are left out, leaving 594. The figures are weighted by each sector's share of the 23,676 working sites.

Checked by hand. For each static detector we drew 50 sites, 25 where it said yes and 25 where it said no, and read the HTML. Tel link, WhatsApp, map, form, quote, local schema, zoom block and format detection scored 50 out of 50, as did phone number in text; booking tool, booking button, mailto, review widget and route link 49; opening hours 48, contact page 47 and address 46 out of 50. A chat widget detector missed the bar and was dropped. We checked the browser measurement on 51 sites with before and after screenshots: a first version was 82 percent right on buttons. After fixes for slow pages and hidden buttons we measured everything again; the second version scored 51 of 51 on the notice, 50 of 51 on what happened after the click, 24 of 25 on coverage and 46 of 51 (90.2 percent) on buttons.

Limits. A booking widget or chat button inside an embedded frame does not count as a button. A header that appears after a few seconds may be missed. A notice that only offered an accept button was left in place. The address detection misses street names in lower case and so undercounts in French-speaking areas. We give percentages from 30 sites; we only call a gap between two groups a difference when both have at least 100 sites and the gap is at least 5 percentage points. A correlation is not a cause.

Measured politely. robots.txt first, GET only, at most one request per second per server, homepage only, our own user agent with a contact address. No form submitted, no login, no cookies accepted. We publish counts only: no domains, no names. The raw HTML and the screenshots stay with us.