Only 11 of 147 F&B Results Were a Restaurant's Own Site

A diner types "korean bbq buffet singapore price". Twenty results come back. Not one of them is a restaurant.

That was the sharpest single result in a study we ran on 21 September 2026. We put eight ordinary Singapore food searches through our own metasearch, read 147 results, and counted how many were the website of the restaurant or café itself. The answer was 11.

Eleven of 147, or 7%. The other 140 slots belonged to food media, review platforms, video discovery pages, booking platforms, hotel chains, stock photo libraries and, in a few cases, places that were not in Singapore at all.

That is the outside view. The inside view is what we found when we opened the sites behind those eleven results. Three of nine measurable homepages showed a price. Five showed opening hours. One had a heading phrased as a question. Every single one had a way to book.

Key takeaways

Every figure below was collected on 21 September 2026. Counts are always shown with their base.

#	Finding	The number
1	Results that were a Singapore restaurant's or café's own website	11 of 147 (7%)
2	Measured homepages showing at least one price	3 of 9
3	Measured homepages showing opening hours	5 of 9
4	Measured homepages with a way to book or enquire	9 of 9
5	Measured homepages with a question heading or FAQ content	1 of 9
6	Measured homepages declaring a Restaurant or LocalBusiness entity	6 of 9
7	Homepage images with no alt text	128 of 261 (49%)
8	Results on "dim sum restaurant singapore menu price" that were third-party menu-price sites	4 of 15

Row 8 is the one we would read twice.

What we searched, and what we counted

Eight searches, run through SingRank's self-hosted SearXNG on 21 September 2026. We built them from the parts Google Autocomplete had returned for our F&B seeds the same morning — a cuisine, an area, and one of the five things buyers add: price, menu, hours, booking, near me. Two of the eight were autocomplete phrasings word for word. The other six were assembled, which is a choice you should weigh when you read the counts.

Search	Results read	Restaurant's or café's own site
japanese restaurant singapore near me	20	1
restaurant singapore with private room	20	2
korean bbq buffet singapore price	20	0
zi char restaurant singapore reservation	20	1
seafood restaurant east coast singapore menu	13	1
cafe singapore book a table	20	3
dim sum restaurant singapore menu price	15	1
halal cafe singapore opening hours	19	2
Total	147	11

We counted a result as the operator's own site only when the domain belonged to a Singapore food business: a restaurant, a café, or a Singapore restaurant group. We did not count a global hotel chain's Singapore dining page, a Facebook page, a guide site that reviews restaurants for a living, or a directory.

Those eleven results came from ten distinct websites. One of the ten would not serve a page to our crawler, and returned HTTP 503 to a second, unrelated fetch minutes later, so nine could be measured. A 503 at one moment is not proof a site is down for customers, and we are not treating it as one. It is one site we could not read.

Only 3 of 9 homepages showed a price

Three of the nine homepages showed at least one price. The other six showed dish names, menu sections or a link to a PDF, with no figure attached to anything.

Read that against the searches that led us there. Three of the eight phrases contained the word "price" or "menu". People were asking what it costs. Six of the nine sites we reached did not say.

There is an old reason for this, and it is not laziness. Publishing a price invites comparison, and in F&B the cost per head moves with supplier prices, service charge and whether it is a weekday lunch or a Saturday dinner. A printed number ages badly.

In our view the trade is still worth making, because the alternative is not silence. It is somebody else's number.

If you don't publish your price, who does?

On the search "dim sum restaurant singapore menu price", 15 results came back. One was a restaurant's own site. Four were third-party sites whose entire business is republishing Singapore restaurant menus with prices.

Three of those four carried a year stamp older than 2026 in their own title or snippet: one said its prices were updated in April 2024, one was titled for 2024, one for 2025. We did not open them to verify the figures, and we are not accusing anyone of anything. The date stamps are simply what the results showed.

So the question is not whether your prices are public. They already are. The question is whether the public version is yours and current, or someone else's and two years old. A buyer who sees a 2024 price and then gets a 2026 bill does not blame the aggregator.

The same pattern shows up on opening hours. On "halal cafe singapore opening hours", 19 results came back and two were a café's own site. The rest were directories, listicles and food media, several of which reproduce hours they collected at some point and have no reason to revisit.

Five of nine showed opening hours, and two put them in structured data

Five of the nine homepages showed opening hours in the visible text. Two of the nine also declared them as structured data, using the OpeningHoursSpecification type that Google's local business documentation describes as "Hours during which the business location is open."

That gap matters more in F&B than in most trades, because the question "are they open now" is asked at the moment of the decision, usually on a phone, usually by someone standing somewhere. Autocomplete knows it: among the suggestions Google returned for our F&B seeds were "halal food singapore near me open now" and "cafe 24 hours singapore".

Menu markup was rarer still. Across every page we sampled, two of the nine sites carried Menu or MenuItem structured data anywhere. Google's documentation defines the menu property for a restaurant as "For food establishments, the fully-qualified URL of the menu" — a small, specific field that most of the sample left empty.

Do these sites tell a machine they are restaurants?

Six of the nine homepages declared themselves as a Restaurant or LocalBusiness in structured data. Two declared only a generic Organization, which tells a machine that an organisation exists but not that it serves food at an address. One had no structured data at all.

Before anyone adds markup as a first move, Google is explicit that it is not an entry requirement. Its guidance on AI features says "You don't need to create new machine readable files, AI text files, or markup to appear in these features" and "There's also no special schema.org structured data that you need to add."

What the same page does ask for is duller and harder: "Making sure your structured data matches the visible text on the page." In a restaurant that means the hours in your markup, the hours on your homepage and the hours on your Google Business Profile agreeing with each other. We did not test that agreement here, and it is the check we would run next. Our guide to schema markup for AI search covers the types.

One of nine homepages asked a question

One homepage in nine had a heading phrased as a question or a block of question-and-answer content on the page itself. Two more linked out to a separate FAQ page.

This is the cheapest miss in the whole study. A diner arrives with a question already formed — can we book for twelve, is there a private room, is it halal, what time do you close on Sunday — and the page answers none of them in the shape it was asked.

Our study of 109 Singapore local pages found question headings on 57 of 109, or 52%. F&B in this smaller sample sat far below that. The two samples are not directly comparable — different searches, different services, much smaller numbers here — but the direction is hard to miss.

Is the gap technical, or is it the words?

It is the words. On the plumbing, this sample did well.

All nine served HTTPS and had a robots.txt that did not block our crawler.
Eight of nine homepages returned in under 400 milliseconds. The slowest took 2,912 milliseconds, which is server response time, not a full page load.
Eight of nine had a title tag of 65 characters or fewer; seven of nine had a meta description between 70 and 160 characters.
Six of nine had exactly one H1. Two had none. One had three.

The one exception was images. Across the nine homepages we counted 261 images, of which 128 (49%) had no alt text. On two sites, not one homepage image was labelled — 56 of 56 on one, 40 of 40 on the other.

For a restaurant that is a specific loss, not a generic one. The food photograph is the product. An unlabelled image is a photograph that no search system, and no customer using a screen reader, can tell anything about.

What we would fix first

If you run a Singapore restaurant or café and you want one afternoon's work to count, this is the order we would take it in.

Put a price on the page. A range is enough: lunch sets from S$X, dinner à la carte S$X to S$X per head, buffet S$X++ on weekdays. Date it, and change it when you change the menu. Six of nine sites in this sample left that field to somebody else.
Put opening hours in the visible text, then match them everywhere. Homepage, Google Business Profile, structured data. Different hours in three places is worse than hours in one.
Turn four real enquiries into four headings. Use the questions your staff answer on the phone every week, with the answer directly underneath. That is the single most common gap we measured.
Label the food photographs. Alt text on the dishes, not on the decorative background images.
Check what the aggregators say about you. Search your own name plus "menu" and plus "price", and read the first page as a customer would. If a 2024 price is sitting there, that is the page doing your pricing.

What we would not start with is the website at all, for many operators. The Google Business Profile is where the "open now" and "near me" decision is settled, and it is free. Our Google Business Profile guide covers that surface, and the fuller structure for an F&B site is in our restaurant SEO guide.

The part that costs us something

Here is the honest reading of our own headline number. If food media, platforms and directories hold 140 of 147 slots, then for most restaurants, paying an agency to rank your own site for "best korean bbq buffet singapore" is a poor use of money. Those pages are held by publishers with years of accumulated authority and a business model built on holding them. We are not going to pretend a retainer changes that quickly, or at all.

The money in F&B search is in the searches where you are already the answer: your name, your area, your menu, your hours, your booking. Those you can win, because nobody else is better placed to answer them than you. We measured the same crowding problem from the other side in best-of lists hold 28 of 60 slots.

Method and limits — read these before quoting

Metasearch is not Google. Results came from SingRank's self-hosted SearXNG on 21 September 2026. It blends engines and its ordering is not Google's ranking, not Google Maps, and not the local pack that most food searches actually land in. Several engines returned errors on the day, which is normal for this tool and logged.
Small numbers. Eight searches, 147 results, ten own-sites found, nine measured. A single different search would move the percentages.
One day, one snapshot.
Homepages, not whole sites. The visible-content checks describe each site's homepage. A price or an FAQ living on an inner page would not be counted, and on several of these sites it very likely does. We say so deliberately, because a homepage is where these searches landed.
The content checks were read, not parsed. Address, price, hours, booking path and question headings were assessed by reading each page. Structured data, titles, meta descriptions, H1 counts, image alt text and response times were read by machine.
What we did not measure. JavaScript-rendered content, CDN or firewall blocking, Google Business Profile accuracy, review counts, or whether any of these businesses get bookings. None of that is in these numbers.
No business is named. This page rates no company and identifies no site.

Want these checks run on your page?

Send us the one page that should answer "menu", "price" or "book" for your outlet. We will run the same checks on it and tell you what to fix first, in order. That starts as a free visibility check through our contact page.

If you would rather we did the work, our SEO, AEO and GEO programme starts from S$500/month on the live page today, with a three-month minimum. We will not promise you a ranking, a citation or a booking. AI platforms and Google choose their own answers, and a full restaurant is decided by the food long before it is decided by a website.

These findings describe eight searches and nine homepages sampled on 21 September 2026. Counts will change. This page names no business and is not a rating of any company.

FAQ

Why do food media sites hold so many Singapore restaurant search results?

Because listicles answer the question most food searches ask. "Best korean bbq buffet" wants a comparison of many places, and a single restaurant's site cannot honestly provide one. Publishers who make those comparisons full time accumulate authority on exactly those phrases, and they hold the slots.

Should a restaurant publish prices on its website?

In our view yes, as a range with a date. In this sample only three of nine homepages showed any price, while third-party menu sites filled the gap — three of the four we saw on one search carried a 2024 or 2025 date stamp. A stale public price is still your price to the customer.

Does a restaurant website need schema markup to appear in AI answers?

No. Google's documentation states you do not need new machine-readable files, AI text files or markup to appear in its AI features, and that no special schema.org data is required. What it does ask is that any structured data you do use matches the visible text on the page.

What is the fastest fix for a Singapore café website?

Put opening hours and a price range in the visible text on the page a customer lands on, then make those match your Google Business Profile. Five of nine homepages in this sample showed hours and three showed a price, so both are common gaps and neither needs a developer.

How was this study measured?

Eight Singapore food searches were run through SingRank's self-hosted SearXNG on 21 September 2026 and 147 results read by hand. Ten belonged to Singapore food businesses; nine could be fetched. Each homepage was then checked for address, price, hours, booking path, question headings and structured data.

Can I cite these figures?

Yes. Please cite "SingRank, Singapore Restaurant Website Study, 21 September 2026" with a link to this page. Quote every percentage with its count, such as 11 of 147, and carry the metasearch caveat. Corrections and challenges to any number here are welcome.