You are building a list of local shops to pitch a website to, and you are building it by hand. Map, trade, town, click a pin, look for the little globe icon, write the name down, click the next pin. Forty pins in, you have a handful of names, and a doubt you cannot shake: two of them almost certainly do have a site that the listing simply never showed, and one of them has a Facebook page you counted as a real website. Nobody selling you a scraper will tell you how often that happens, because the answer is not flattering to the tool.
So here it is, measured on our own extraction lot, with the date, the volume and the counting rule attached.
of Google Maps listings display no website field at all.
69,497 of 235,766 listingsof the links that are displayed do not lead to a site the business owns.
16,200 of 166,269 listings carrying a linkA displayed link is not a website
Start with the second number, because it is the one that quietly ruins a prospecting list. A naive filter keeps the rows where the website column is empty and throws away the rest. On our lot, that filter throws away 16,200 businesses whose displayed link goes somewhere else entirely. They sort into clean families.
Social pages
A Facebook page, an Instagram profile, sometimes a bare messaging link. Facebook and Instagram alone account for more than 5,500 listings in our lot.
Trade booking platforms
Appointment platforms for healthcare, booking platforms for hair and beauty. This is the heaviest family by far, and it is the whole explanation for the extreme rows further down this page.
Ordering and review platforms
Delivery pages, table reservation pages, online directory entries that the owner did not build and does not control.
Third party hosted pages
Free mini sites, link aggregator pages, subdomains handed out by a platform.
Network and franchise sites
The listing of an individual agent points at the national site of the network they belong to. Worth noting: the French statistics office counts that case as having a website, since its wording is "owned or shared".
Two practical consequences. Your addressable pool is wider than the headline rate, because a shop whose only shop window is a booking platform is a shop with no site of its own. And the opposite mistake costs you credibility: pitching a website to someone who already has a good one is the fastest way to be ignored. Telling the two apart means looking at what actually runs behind the link, which is the job of website technology detection. Its quieter cousin, the link that no longer answers at all, is handled by a dead URL check across a contact base. And when the shop window turns out to be a social page rather than a site, social profile extraction is what recovers a usable point of contact.
How the lot was measured
No modelling, no extrapolation, no sampling weights. We took the extractions we actually ran and counted.
- Lot
- 63 Google Maps extractions
- Window
- 18 Jun to 18 Sep 2026
- Area
- France
- After deduplication
- 235,766 businesses
- Counted on
- 21 Sep 2026
Listings were deduplicated on the business identifier carried by the Google Maps link, which is stable from one extraction to the next and removes businesses returned several times by different queries, map tiles or extraction runs, then filtered on a valid five digit French postcode. A listing counts as "no website" when its website field is empty. That is the whole rule, and it is deliberately blunt so that anyone can reproduce it.
The average is useless, the trade is not
29.5% describes the lot, not your evening. What changes your evening is the spread between trades, and it is very wide: from 7.7% to 75.3% of listings without a site across the thirteen trades that carry more than 2,000 listings each.
| Trade extracted | Listings | No site | Wrong link |
|---|---|---|---|
| Florist | 3,425 | 75.3% | 10.5% |
| Nurse | 4,032 | 75.3% | 36.1% |
| Physiotherapist | 4,353 | 58.7% | 67.2% |
| Hairdresser | 6,684 | 37.6% | 42.1% |
| Phone repair shop | 15,277 | 37.5% | 7.7% |
| Restaurant | 12,385 | 34.7% | 7.4% |
| Osteopath | 4,647 | 31.4% | 55.3% |
| Plumber | 6,880 | 31.3% | 6.2% |
| Beauty salon | 2,909 | 30.4% | 27.4% |
| Driving school | 2,417 | 26.6% | 4.7% |
| Real estate agency | 81,176 | 22.2% | 1.5% |
| Photographer | 39,231 | 18.2% | 4.1% |
| Digital marketing agency | 3,595 | 7.7% | 0.7% |
Read the two right hand columns together, never separately. Nurses and physiotherapists top the "no site" column, but they also top the "wrong link" column, because their real shop window is an appointment platform and their appetite for a site of their own is correspondingly small. Florists, hairdressers and restaurants are the interesting rows: high absence, low platform dependency, and a clear commercial reason to want their own pages. Rebuilding this exact breakdown for your county is what Google Maps scraping with CSV export produces, one query at a time.
Smaller town, denser list
The same lot, cut by the population of the town rather than by trade. Municipal populations come from the French statistics office reference populations for 2023. 221,906 listings could be attached to a town; the remaining 13,860 sit outside this one breakdown.
Share of listings with no website field, by municipal population. outsend measurement of 21 September 2026, on 221,906 attached listings.
The gradient is modest but it holds across the whole scale, on tens of thousands of listings. It says something a map cannot: target density rises as you move away from city centres, exactly where the number of web freelancers competing for the same shops falls. That makes it a routing argument as much as a list argument. If you plan an actual round of visits or calls, the extra phone numbers recoverable from listings are usually worth more than a guessed email address, because a shop owner with no website answers the phone.
The only official figure we could verify stops short
There is one public statistic on this, and it is worth knowing precisely what it covers. In 2023, 69% of French enterprises had a website, owned or shared, according to the French national statistics office in its October 2025 report on business websites and social media. The figure is sound. The line underneath it is what matters and is almost never quoted: the survey scope is enterprises with 10 employees or more.
The single florist, the four table restaurant, the hairdresser with one apprentice are not under-represented in that number. They are outside it. And the internal trend gives away what lies beyond the boundary: in the detailed results of the 2023 ICT survey, published on 21 December 2023, 59,009 of the 94,858 enterprises under 20 persons employed report a website, that is 62.2%, against 71.3% between 20 and 49, 85.9% between 50 and 249, and 94.4% at 250 and above. The line falls steadily as firms get smaller, and then the data simply stops at the edge of the scope. What happens below ten employees is where your prospect list lives, and it is why the measurement above had to be made rather than looked up.
What this measurement does not say
The lot is not a representative sample of any economy. It is the set of queries that were actually run in the period, and two of them, real estate agencies and photographers, account for half the listings on their own. The headline 29.5% therefore depends on that mix. The per trade rates stand on their own and should be read individually.
An empty field means the listing shows no website, not that the business has none: a site that exists but was never added to the listing is counted as absent. In the other direction, the platform list used to count wrong links is a closed list, so 9.7% is a floor and not a ceiling.
The whole lot is French, and it was counted on 21 September 2026. Listings move. We will run it again.
Running it on your own area
outsend is a Google Maps scraping and qualification tool, in public alpha. In practice you describe a trade query and an area, the extraction returns each business with its name, address, phone, website field, reviews and opening hours, and you get the result as CSV, JSON or XLSX. Filtering for "no site" is then one column, and checking what the remaining links are really worth uses the qualification modules linked above, inside the same tool rather than across four separate subscriptions. The full module list sits on the outsend features page, and the vocabulary is defined in the prospecting glossary. If you are weighing tools against each other first, the comparison hub lays out the alternatives.
Access is by application and free: Request free alpha access.
Frequently asked questions
How do I find businesses with no website for free?
Opening Google Maps and clicking listings one at a time costs nothing and does work. Our measurement gives you the price in clicks: fewer than two listings per hit on florists, close to thirteen on digital marketing agencies. A Google Maps scraper returns the same field as a spreadsheet column, which moves the effort from clicking to sorting. outsend's alpha is free on application.
Which trades should a web freelancer approach first?
On the lot measured on 21 September 2026, florists sit at 75.3% of listings without a site and hairdressers at 37.6%, against 18.2% for photographers and 7.7% for digital marketing agencies. But a high absence rate alone does not make a good target. Nurses and physiotherapists show the highest absence rates and the highest wrong link rates at the same time, because an appointment platform already does the job a website would do for them. Florists, hairdressers and restaurants combine high absence with low platform dependency.
What exactly is a wrong link?
A listing that displays a link, and that a naive filter would therefore discard, where the link leads to a Facebook page, an Instagram profile, a booking platform or the national site of a franchise network, rather than to a site the business owns. On our lot, 9.7% of displayed links fall into that category, rising to 67.2% in some healthcare trades.
Does the official 69% contradict your 29.5%?
No. The two numbers describe different populations and different objects. The statistics office asks enterprises with ten employees or more whether they have a website, owned or shared, and records the answer. We look at what a Google Maps listing displays, for businesses the vast majority of which sit below that ten employee threshold. One measures a declaration, the other measures a public shop window.
Do these French numbers travel to other countries?
The counting rule does, the rates almost certainly do not. The rule, count the empty website field and then check the domain behind the links that are filled, is country agnostic and takes an afternoon to reproduce anywhere Google Maps has coverage. The rates depend on which booking platforms dominate a given market, and those differ by country. Treat the French figures as an order of magnitude and as a warning about wrong links, then measure your own market.