Restaurants Without a Website: The Signal That Predicts Them
Of 975 Madrid restaurants, 75.0% of those with fewer than 50 reviews had no website. Among those above a thousand reviews, the figure dropped to 8.5%.
That is the signal to use. The municipal average of 28.2% mixes the two groups and is useless for prospecting. The review count sits on the same listing, costs nothing to read, and splits the market better than the neighbourhood or the cuisine.
Download the review band data and the category data.
Reviews predict the website better than anything else
| Reviews | Venues | No website link | 95% CI | Median reviews |
|---|---|---|---|---|
| 1 to 49 | 56 | 75.0% | 62.3 – 84.5 | 26.5 |
| 50 to 199 | 124 | 52.4% | 43.7 – 61.0 | 115 |
| 200 to 999 | 385 | 33.8% | 29.2 – 38.6 | 542 |
| 1000 or more | 399 | 8.5% | 6.2 – 11.7 | 2081 |
| None | 11 | 36.4% | 15.2 – 64.6 | 0 |
The drop is ninefold between the first band and the last, and the intervals do not touch: the 1 to 49 band starts at 62.3% and the thousand-plus band ends at 11.7%. This is not sampling noise.
The band with no reviews at all breaks the staircase and sits at 36.4%, but that is eleven venues and its interval runs from 15.2% to 64.6%. At that width there is little to conclude. It most likely mixes recent openings with abandoned listings.
What the signal does not tell you
Fewer reviews also means fewer customers and a smaller budget. The 1 to 49 band has a median of 26.5 reviews: small venues, many of them newly opened, and the odds of them paying for a custom site are low.
There is a second problem in that band, and it is about reachability. One venue in four in the 1 to 49 band had no phone number on the listing either, against 1.0% of those above a thousand reviews. The band with the most missing websites is also the hardest to reach.
The middle band is the one that closes. Between 50 and 199 reviews there are 124 venues, slightly more than half without a website, with a median of 115 reviews. They already turn over enough to justify the spend and still have nobody running their digital side.
The thousand-plus band looks disposable at 8.5%, but that is 399 venues, and the percentage works out to 34 restaurants with heavy volume and no site of their own. It is the shortest list and the one with the largest ticket.
The exact category, and why it shifts by market
| Google category | Venues | No website link | 95% CI |
|---|---|---|---|
| Bar | 20 | 65.0% | 43.3 – 81.9 |
| Bar & grill | 84 | 50.0% | 39.5 – 60.5 |
| Tapas bar | 19 | 36.8% | 19.1 – 59.0 |
| Restaurant | 456 | 31.8% | 27.7 – 36.2 |
| Peruvian restaurant | 25 | 24.0% | 11.5 – 43.4 |
| Spanish restaurant | 31 | 12.9% | 5.1 – 28.9 |
| Mexican restaurant | 14 | 7.1% | 1.3 – 31.5 |
| Italian restaurant | 28 | 3.6% | 0.6 – 17.7 |
| Mediterranean restaurant | 24 | 0.0% | 0.0 – 13.8 |
One detail is worth pausing on. In Madrid, Mexican restaurants came in at 7.1% with no website. In our international sample, the same label reached 50.0%.
That is not a contradiction, it is the label meaning different things. A Mexican restaurant in Madrid is a niche concept with investment and marketing behind it. In Mexico it is the default category for any neighbourhood eatery. A category benchmark is only valid inside the market where it was measured.
Hence the rule: run a hundred-listing test in your own city before applying anyone's table, this one included.
Counting the social-only venues too
An empty field is the strict definition. If your offer is a site the owner controls, the relevant group is larger, because it includes the venues linking Instagram, Facebook or a booking platform.
| Reviews | Field empty | No site of their own |
|---|---|---|
| 1 to 49 | 75.0% | 80.4% |
| 50 to 199 | 52.4% | 62.9% |
| 200 to 999 | 33.8% | 42.6% |
| 1000 or more | 8.5% | 10.5% |
The difference between the two columns is your expanded market. In the middle band it is ten points, roughly twelve extra venues per hundred reviewed.
Watch the reverse case too: an empty field does not prove a site is absent. The restaurant may have one and simply never linked it. A quick name search before you call is worth the seconds it costs.
How to build the list
- Pick the city and the exact category. "Bar" and "Bar & grill" return far more than "Italian restaurant".
- Run a small test search and check the real ratio before you spend the whole budget.
- Export to CSV with the website, phone and review count columns.
- Filter on empty website and sort by reviews, highest first.
- Cut the list into the three bands and treat each as a separate campaign.
To narrow down to one district instead of the whole city, draw a polygon on the map. There is no postal-code filter and no radius filter.
The Madrid district breakdown is in the Madrid restaurant study, and the general workflow is in the guide to finding businesses without a website.
What to offer each band
Under 50 reviews. Small budget and a hurry to exist. A one-screen page with the menu, the hours and the phone number works. Fixed price, delivered in days.
Between 50 and 199 reviews. The band where the money is. The venue already works and its problem is operational: bookings, orders, an out-of-date menu. Here you argue with the commission they hand to the delivery platforms.
Above a thousand reviews. Few, large, and with a lot to lose. You are not selling them a website, you are selling them independence from a profile they do not control. Longer conversation, bigger contract.
96.0% of the restaurants in the Madrid sample had a phone number, so the channel is available across almost the whole list. Emails will not be: the field does not exist in Google Maps, and enrichment pulls them from the website, which on this list is precisely what is missing.
FAQ
How many restaurants have no website?
It depends on the review band far more than on the city. In our sample of 975 Madrid restaurants, 75.0% of venues with fewer than 50 reviews had an empty website field, against only 8.5% of those above a thousand. The municipal average of 28.2% hides that difference.
Which signal works for filtering restaurants without a website?
The review count. It is the one listing field that separates the bands cleanly without overlap: the confidence interval for the 1 to 49 band runs from 62.3% to 84.5%, and the one for a thousand or more runs from 6.2% to 11.7%. Sorting the export by reviews beats filtering by neighbourhood.
Does a restaurant with an Instagram page count as a restaurant without a website?
That depends on what you sell. If your offer is a site the owner controls, then yes: in the 1 to 49 review band, 80.4% had no site of their own, against 75.0% with the field outright empty. The gap between those two numbers is the venues linking a social profile or a platform.
Does the same category behave the same in every market?
No. The label means different things in different countries. Mexican restaurants in Madrid came in at just 7.1% with no website, because there the concept is a niche one. In the international sample the same label reached 50.0%. Measure the category in your own market before trusting anyone else's benchmark.
Can I get the email addresses of those restaurants?
Not from a list of restaurants without a website. A Google Maps listing has no email field, and enrichment pulls the address from the business website. No site means no email. The channel for that group is the phone, and 96.0% of the Madrid restaurants in the sample had one.
Getting started
Export a hundred restaurants from your own city with the review column included, cut them into the same bands and compare. If your market resembles Madrid, the 50 to 199 band will hand you more than half the prospects for half the effort.
Create an account and run the test with 100 free credits every month, no card required.