Last updated: 28 August 2026. Written by Adrian Tan, Singapore Digital Marketing (SDM).
Two numbers describe the state of Google Ads for food businesses better than any amount of strategy talk. In LocaliQ’s 2026 benchmark data, Restaurants & Food has an average cost per click of US$2.05 — the cheapest of any category measured, against an all-industry average of US$5.42. In the same dataset, the category’s conversion rate fell 18.69% year on year, one of the steepest declines of any industry.
Cheap traffic that converts worse every year is a specific kind of trap. It is easy to spend into, easy to justify on a cost-per-click report, and easy to lose money on. The F&B operators we see doing well on Google are not the ones bidding hardest. They are the ones who worked out which of the three distinct kinds of food search they were buying, and stopped paying for the other two.
This guide covers that structural question first, then the brand-defence problem that is now the single highest-return campaign in most Singapore restaurant accounts, then the Nutri-Grade mechanics that are specific to Search and Performance Max, then what you can honestly measure. For the paid social side, see our companion piece on Meta Ads for F&B in Singapore, which covers the Nutri-Grade content rules in full; this article deals with the Google-specific mechanisms. The broader picture is in our F&B digital marketing guide.
Three kinds of food search, and only one of them is yours
Almost every wasted dollar in an F&B Google Ads account comes from treating food search as one thing. It is three, and they behave completely differently.
Discovery search is a trap for single outlets. “Best ramen in Singapore” is a query owned by listicles and aggregators, and a searcher typing it has no attachment to you. You can buy the click for two dollars, but you are buying a coin flip against every other ramen shop in the country, and the page they land on has one chance to overcome the fact that they were not looking for you. Unless you are a chain with the coverage to make the odds work, this is the demand type to reach through local surfaces and organic presence rather than broad keyword bidding — which is what our note on near-me searches in Singapore is about.
Occasion search is under-served and under-bought. “Private dining room Singapore”, “halal buffet for 30 pax”, “corporate lunch catering Raffles Place”, “same-day birthday cake delivery”. These are specific, commercially serious, low-competition, and almost nobody builds a proper landing page for them. One enquiry can be worth two hundred walk-ins. If your venue can service any of these, this is where the first dollar should go.
Brand search is where the money actually is, and it needs its own section.
Brand defence: the campaign that pays for the account
Search your own restaurant’s name in Singapore. In most cases you will find GrabFood and foodpanda listings above or alongside your own, sometimes as ads, usually as organic results with more domain authority than you will ever have. Someone who searched specifically for you — the warmest traffic that exists — clicks the aggregator, and 15% to 30% of that order goes to the platform, depending on the plan you are on.
The market got more concentrated this year, which matters for this calculation. Deliveroo went offline in Singapore after 4 March 2026, when DoorDash wound down its operations here alongside three other markets, ending an eleven-year presence. Singapore delivery is now effectively a two-platform market, which does not improve anyone’s negotiating position.
The arithmetic on brand defence is unusually clean. Take a restaurant with a $45 average delivery order on a 25% commission. Every order that comes through the platform costs $11.25 in commission. If a brand-term click costs $0.60 — brand terms are cheap because your Quality Score on your own name should be excellent — and one in five clicks converts to a direct order, your cost per direct order is $3.00. You spent $3.00 to avoid $11.25. That is the whole argument.
| Order via aggregator | Order via brand-defence ad | |
|---|---|---|
| Order value | $45.00 | $45.00 |
| Platform commission (25%) | −$11.25 | $0.00 |
| Ad cost (5 clicks at $0.60, 20% CVR) | $0.00 | −$3.00 |
| Own-delivery or collection cost | Included | Varies — model yours |
| Retained per order, before fulfilment | $33.75 | $42.00 |
| Customer data | Platform’s | Yours |
Two honest caveats. The fulfilment column is not zero — if you have to arrange your own delivery, the gap narrows, and for some operators it closes entirely, so run your own numbers rather than ours. And you are not capturing every order: some people prefer the aggregator’s interface, its vouchers, or its stored payment details, and no ad will change that. Brand defence is a margin-recovery play on the subset of customers who would have gone direct if going direct had been easy.
Build it as its own campaign, not an ad group inside a bigger one. Exact match on your name and its common misspellings, phrase match on your name plus “delivery”, “menu”, “opening hours”, “reservation”, “book”. Sitelinks straight to the ordering page and the reservation page. A small, ring-fenced budget so it never competes with your prospecting campaigns for spend. And a landing page that makes ordering direct genuinely easier than the app — if your own ordering flow is worse than GrabFood’s, the campaign will fail and it will be the page’s fault, not the ad’s.
Nutri-Grade, on the Search side
The substantive Nutri-Grade rules are set out in our Meta Ads for F&B piece and we will not repeat them. The short version: advertisements for Nutri-Grade beverages graded D are prohibited across broadcast, print, out-of-home, on-ground and online platforms, with a narrow point-of-sale exception. The measures have applied to pre-packaged beverages and non-customisable dispensers since 30 December 2022, and to freshly prepared beverages and customisable dispensers since 30 December 2023.
What is different about Google is the mechanism by which you breach it. Three of them, none of which exist on Meta.
Mechanisms 1 and 2 are ad-side; mechanism 3 is destination-side. Auditing only the ad account leaves the job half done.
1. Keyword triggering. On Meta you choose the creative and it runs as written. On Search, a broad-match keyword decides which of your products a query gets matched to. A restaurant running broad match on “bubble tea [area]” with a responsive search ad pulling headlines about its signature milk tea has, if that drink grades D, built an advertisement for a Grade D beverage without ever intending to. The fix is unglamorous: know the grade of every beverage on your menu, and treat Grade D SKUs as negative keywords at the account level. Not just their exact names — their common variants and abbreviations too.
2. Auto-generated assets. Performance Max and responsive search ads will generate headlines and descriptions from your landing page and your feed unless you turn that off. If your menu page lists a Grade D drink prominently, Google’s asset generation may well surface it. This is the single most common way we see compliant intentions produce non-compliant ads. Switch off automatically created assets in any campaign where beverage grading is in play, and review the asset report rather than trusting the setting.
3. The landing page is a menu. This is the duty most easily missed, because it is not an advertising rule at all. Where a freshly prepared or pre-packaged beverage is graded C or D, the Nutri-Grade mark must appear next to the beverage where it is listed for sale — and MOH’s implementation notice names physical and online menus, with digital platforms explicitly listed among the covered retail settings. Toppings that can be added — pearls, jellies, ice cream, whipped cream — must carry a declaration of their sugar content. So the page your ad sends traffic to has its own labelling obligation, independent of the ad. Auditing the ads and forgetting the destination is the classic half-done job.
There is a genuine exemption worth knowing. MOH’s notice provides that smaller food businesses — those with annual revenue not exceeding S$1 million and operating fewer than ten premises — are exempt from the freshly prepared beverage requirements, though they must still comply for pre-packaged beverages and automated dispensers. A single-outlet café below that revenue line has less to do here than a chain does. Check your own position rather than assuming either way, and note that the exemption is about freshly prepared drinks only.
Looking ahead: from mid-2027 the Nutri-Grade framework extends to sodium and saturated fat across 23 sub-categories of pre-packed salt, sauces, seasonings, instant noodles and cooking oils in retail settings, with the same structure — front-of-pack labelling for Grade C and D, advertising prohibited for Grade D. If you manufacture or retail any of those, that is a 2027 campaign-planning problem you should be scoping in 2026.
Structure for a single outlet, and for a chain
F&B accounts fail more often from structure than from bidding. A workable shape:
| Campaign | Match / type | Budget share | Primary conversion |
|---|---|---|---|
| Brand defence | Exact + phrase on brand terms | 10–15%, ring-fenced | Direct order / reservation |
| Occasion & enquiry | Exact + phrase, one ad group per occasion | 35–45% | Form submission or call |
| Local / store goals | PMax for store goals, GBP linked | 30–40% | Local actions; store visits if eligible |
| Category / discovery | Phrase, tightly negated | 0–15%, last to fund | Direct order |
For a chain, add location-level segmentation before you add budget. Nothing wastes money in Singapore faster than one national campaign serving an outlet in Jurong to a searcher in Tampines. For a single outlet, radius targeting around the venue plus the surrounding office and residential catchments will outperform an island-wide setting almost every time, and it costs nothing to implement. Our guide to building your first Google Ads campaign in Singapore covers the setup mechanics; Google Ads costs in Singapore covers the budgeting side.
The local surfaces, and what “store visits” really requires
Performance Max for store goals promotes your locations across Search, Maps, YouTube, Display, Gmail and Discover, optimising toward store visits, store sales and local actions such as call clicks and direction clicks. It draws locations, hours and contact details from your linked Business Profile, which makes Google Business Profile optimisation a prerequisite rather than a nice-to-have: incomplete hours or an unverified location will quietly degrade the campaign.
Set expectations correctly on store visits reporting. Google’s own eligibility guidance requires active location assets or affiliate location assets, verified Business Profile locations, and enough ad clicks or impressions and enough foot traffic to clear its privacy thresholds — with no published minimums, and the note that the more locations you have, the likelier you are to have enough data. In practice, a single-outlet Singapore restaurant frequently will not qualify.
That is not a reason to avoid the campaign type. It is a reason to define your conversions around what you can actually count: direction clicks, call clicks, menu views, reservation-platform handoffs, order-page starts. Assign each a considered value based on what proportion historically turns into a paying customer, and optimise to value rather than to a volume number you cannot audit.
Measuring an industry that transacts offline
F&B has the hardest measurement problem in local marketing, because the conversion happens at a table. Four things are worth doing, in order of return.
- Call tracking with a duration threshold. A thirty-second call to a restaurant is usually a booking or a large order enquiry. A five-second call is a wrong number. Counting all calls equally teaches the bidding algorithm to chase noise.
- Reservation and ordering platform handoffs. If you use Chope, Oddle, a POS-linked ordering page or similar, get the confirmation event back into Google Ads. Our guide to conversion tracking in Singapore covers the implementation, including the server-side route where a third-party checkout blocks the tag.
- A promo code that only exists in your ads. Crude, unglamorous and the only offline attribution most independents will ever get. Redemption count is a floor on ad-driven walk-ins, not a complete picture, but a floor you can trust beats a modelled number you cannot.
- A weekly reconciliation against covers. Compare ad spend and lead volume against actual covers and average spend per head. If the ad account is improving and the covers are not, the problem is downstream — the page, the menu, the wait time, the price — and more budget will not fix it. That diagnostic sequence is what our note on Google Ads not converting works through in detail.
On the benchmark numbers: the LocaliQ 2026 dataset puts Restaurants & Food at a 6.83% click-through rate against a 6.64% all-industry average, an 8.05% conversion rate against 8.18%, and a cost per lead of US$30.57 against US$66.69. Read together, that is a category where people click readily, convert about averagely, and produce leads at less than half the cross-industry cost — but where the conversion trend is deteriorating sharply. The implication is not to spend less. It is that the differences between operators now come from the landing experience and the offer, not from the bid.
A realistic 90 days
If you are starting from nothing:
- Weeks 1–2. Grade every beverage on the menu and add Grade D SKUs as account-level negatives. Audit the online menu for Nutri-Grade marks and topping sugar declarations. Verify and complete the Business Profile. Fix the direct ordering flow before you buy a single click.
- Weeks 2–3. Launch brand defence, ring-fenced. This is the fastest positive-return campaign you will run and it validates your tracking cheaply.
- Weeks 3–6. Build one occasion landing page — whichever occasion your venue genuinely serves best — and a tight exact/phrase campaign against it. One page done properly beats five thin ones.
- Weeks 6–10. Add PMax for store goals with automatically created assets switched off and a full negative list applied. Review the asset report weekly for the first month.
- Weeks 10–13. Reconcile against covers, assign real values to local actions, and only then decide whether category and discovery search deserves any budget at all.
If you are weighing this against organic investment, our comparison of SEO versus Google Ads in Singapore sets out the trade-off, and the complete Google Ads guide for Singapore is the wider context this sits inside.
Cheap clicks are not the same as cheap customers
F&B gets the cheapest traffic on Google and one of the fastest-falling conversion rates, and those two facts are related. Low barriers to entry mean everyone is buying, undifferentiated landing pages mean nobody converts, and the aggregators sit in the middle collecting a quarter of every order that the restaurant’s own marketing generated.
The operators who make this channel work do three unfashionable things. They defend their own name first. They build for the specific occasion rather than the general appetite. And they fix the ordering page before they raise the budget, because in a category with a 6.83% click-through rate and a falling conversion rate, the constraint was never traffic.
If you want a read on your current account, see how we approach Google Ads management in Singapore, look at the F&B work in our client case studies, or send us your account and we will tell you where the waste is.
Frequently asked questions
How much do Google Ads cost for a restaurant in Singapore?
Clicks are cheap by cross-industry standards. LocaliQ’s 2026 benchmark data puts Restaurants & Food at an average cost per click of US$2.05 against an all-industry average of US$5.42, with an average cost per lead of US$30.57 against US$66.69. Singapore-specific costs vary with your area and cuisine competition, and brand terms typically cost a fraction of the category average. The more useful budgeting question is your cost per booking or per direct order, not per click.
Should a restaurant bid on its own brand name?
In Singapore, usually yes, because delivery aggregators frequently occupy the top of the results for restaurant names and take 15–30% of any order that goes through them. If a brand click costs around $0.60 and one in five converts to a direct order, you spend about $3 to avoid roughly $11 of commission on a $45 order. The caveats are that your own ordering flow must be genuinely easy, and that some customers will use the app regardless.
Does Nutri-Grade apply to Google Ads?
Yes. The prohibition on advertising Nutri-Grade beverages graded D covers online platforms, which includes Google Search, Display and Performance Max. Three Google-specific mechanisms cause accidental breaches: broad-match keywords pulling queries toward a Grade D product, automatically created assets generating headlines from a menu page that features one, and the landing page itself being an online menu, which must carry the Nutri-Grade mark for Grade C and D beverages and sugar declarations for toppings.
Is my café exempt from the Nutri-Grade rules?
Possibly, in part. MOH’s implementation notice exempts smaller food businesses — annual revenue not exceeding S$1 million and fewer than ten premises — from the freshly prepared beverage requirements, while still requiring compliance for pre-packaged beverages and automated dispensers. Confirm your own position against the current guidance rather than assuming, and note the exemption covers freshly prepared drinks only.
Can a single-outlet restaurant use store visits as a conversion?
Often not. Google’s eligibility guidance requires active location assets, verified Business Profile locations, and enough ad clicks or impressions plus enough foot traffic to pass its privacy thresholds, with no published minimum and a note that more locations make sufficient data likelier. A single Singapore outlet frequently will not qualify. Use local actions instead — direction clicks, call clicks, menu views, order-page starts — and assign each a considered value.
Which delivery platforms matter in Singapore now?
Two. Deliveroo went offline in Singapore after 4 March 2026, when parent company DoorDash wound down operations here along with three other markets, ending an eleven-year presence. That leaves GrabFood and foodpanda, which makes commission negotiation harder and raises the value of owning your own ordering channel. It also simplifies brand-defence targeting, since there are fewer competing listings on your own name.
Related
The same Nutri-Grade rules land differently on an organic feed, where the brand-advertisement carve-out is a content strategy rather than a constraint, and where your menu carries a labelling duty every time you post it. That side is covered in social media for F&B in Singapore.
