Client name withheld under NDA.
Casual dining searches in Singapore are local, quick and mostly on a phone. People type a cuisine and an area, or an occasion such as a birthday dinner or a group booking, and they decide within minutes. Demand peaks from Thursday to Sunday evenings and on weekend lunch. That makes Google Ads management for a restaurant group a matter of structure: each outlet has its own catchment, its own busy hours and its own booking habits, and one shared setup rarely fits all four.
This group ran four restaurants, in a mall, two heartland estates and the CBD. The owner's marketing executive had managed Google Ads in-house for two years on S$6,000 a month, using Smart campaigns and broad match. The account was not bad. It brought guests in. But it was a single pile of keywords with no view of which outlet or which search was working. Booking button clicks were not tracked, calls were not tracked, and a review of the search terms showed 31% of spend going to searches that had nothing to do with dining. We rebuilt the before figures from call logs and the booking platform for the four weeks before we started.
We went through two years of search terms, location settings and ad text. Location targeting was set to people interested in Singapore, so searches from overseas and Johor were using budget. We kept what the executive had built that worked, including the brand campaign and two strong ad headlines. Then we set up call tracking and booking button tracking, and checked them against the booking platform for three weeks.
Without tracking, nobody could say which search made a table booking. Keeping the good parts of the in-house work also saved time and gave the team a base to build on, rather than a blank page.
We replaced the Smart campaigns with a campaign for each outlet. Inside each one, ad groups split brand searches, cuisine and area searches, and occasion searches such as birthday dinner, group dinner and set menu. We moved to exact and phrase match, and kept one small broad match test with tight limits. Each outlet got its own budget line inside the S$6,000.
Guests choose a restaurant by area first. Splitting by outlet let us set budgets by real demand and see which searches led to a table, instead of reading one blended number.
We built shared negative lists for recipes, jobs, cooking classes, free items and other cities, and added a weekly search term review. Location targeting changed to presence only, and each outlet was limited to the neighbourhoods its guests actually travel from. Searches from tourists staying in the CBD stayed on for the one outlet that serves them.
Most of the waste sat in a few hundred search terms. Cutting those and tightening the radius moved money to people who could realistically walk in that evening.
We rewrote ads around what a diner checks before booking: set menu, opening hours, group seating and a tap-to-call button. We raised bids from Thursday to Sunday between 5pm and 9pm and on weekend lunch, and reduced them on weekday afternoons. The mall outlet gained the least, because most of its guests walk in from the mall, so we moved budget from it to the heartland outlets.
Search demand for dinner is lumpy. Spending when people decide, and writing ads that answer the booking questions, raised the share of clicks that became a call or a booking.
We added a small remarketing audience of people who viewed a menu page but did not book, kept inside the same monthly budget. Each month we sent a short report by outlet: bookings, calls, cost per booking or call, and the search terms added as negatives. We also showed the executive how to read the report, so the in-house knowledge stayed with the team.
A guest who looked at a menu has already shown interest. Reminding them cheaply, and keeping reports simple enough for the owner to read in five minutes, kept the account useful after the rebuild.
After four months on the same S$6,000 a month, bookings and calls from Google Ads rose from 120 to 192 a month. Cost per booking or call fell from S$50.00 to S$31.25 and conversion rate rose from 3.1% to 4.6%. Spend on searches unrelated to dining dropped from 31% to 9%, and average Quality Score moved from 5.1 to 7.4. The mall outlet improved least, and we cut its budget share rather than chase it.
Search terms and location leaks found, call and booking tracking live and checked against the booking platform.
Outlet campaigns launched, negative lists added, location tightened, waste fell sharply.
Day-parting, new ad copy and menu-viewer remarketing in place, monthly report by outlet.
The group did not need a bigger budget. It needed to know which outlet and which search earned a booking. The same S$6,000 produced 192 bookings and calls instead of 120 once the account was split by outlet and intent. If your account feels expensive, look at structure before you add spend.
The brand campaign and a few headlines written in-house were worth keeping. Rebuilding around them was faster, and the executive stayed involved. An in-house account usually has real knowledge in it, and an audit should separate that from the waste.
Before tracking, a click on a menu page looked the same as a click on the booking button. Once calls and bookings were counted, cost per booking or call became the number to manage, and it fell by 38%. Restaurants should measure what fills tables.
How we plan, build and run Google Ads accounts for Singapore businesses.
Search campaigns that reach people at the moment they look for you.
Bringing back visitors who looked but did not act.
Another Google Ads account, for a firm where each enquiry is worth far more.
Tell us about your outlets and what you spend today. We will review your account and show you where the money is going.