APAC is not one market, and Google is not the default everywhere
In Australia, paid search and Google Ads are close to the same thing: Google holds roughly 94 percent of search here, so a Google Ads budget is very nearly the whole paid search opportunity. Carry that assumption north and it breaks. StatCounter has these figures for October 2026, with mainland China at September 2026:
- Japan: Google 59.34 percent and Bing 32.51 percent across all devices. On mobile, Google 85.08 percent and Yahoo 11.79 percent. On desktop, Bing actually leads at 51.83 percent against Google on 42.17 percent.
- South Korea: Google 48.48 percent and Naver 42.25 percent across all devices. On mobile, Naver takes 59.55 percent against Google on 34.13 percent.
- Mainland China: Baidu 46.65 percent, Bing 21.76 percent, Google 1.46 percent.
StatCounter counts page views across the sites in its network rather than people, so read those numbers as the shape of a market rather than a census. The shape is the point. The same regional budget buys near-total coverage in one market and a minority of mobile search in another, and a plan that funds Google Ads evenly across APAC is overfunding Korea on mobile and underfunding nothing in Australia.
One account per market, or one account for the region
This is the decision that is expensive to get wrong, because it is close to irreversible. Google Ads Help states that your currency is permanently set when you set up your account and cannot be updated, because it determines how you are billed, and that to pay in a different currency you have to create a new account. The time zone is the same: permanently set, and changes are no longer supported in standard serving accounts.
There is one narrow exception, and Australian advertisers mostly cannot use it. Google can reset the time zone on a manager account once, on a request from an admin user, and the time zone may be shifted only eastward. Google’s own example is GMT to India Standard Time. Sydney in daylight saving is UTC+11 and Singapore is UTC+8, so an Australian manager account wanting to move onto Singapore time is asking to go west, which the exception does not cover.
The time zone is not cosmetic. Reports segmented by time are generated on the account time zone, and ad schedules are set on it too: Google’s guidance is to adjust your ad schedule to account for any difference in your target customers’ time zone. A Sydney account day-parting a Singapore campaign to business hours is bidding hardest while Singapore is asleep, three hours out, and the campaign report will not say so.
When one account for the region is fine
- One entity is billed, in one currency, and nobody needs a per-market invoice.
- The markets sit within roughly two hours of each other, so one ad schedule is not wrong anywhere.
- Spend per market is small enough that the reporting overhead of separate accounts costs more than the precision buys.
When each market needs its own account
- A local entity is paying, or wants to be billed in its own currency.
- The time offset is wide enough to distort day-parting or daily reporting, which in practice means anything past about three hours.
- A market is large enough to need its own conversion actions, its own budget owner, or its own agency or in-house team with account access scoped to it.
Where you land, the campaign layer still gets split by market. One campaign per country keeps budget, bids and targets set per market instead of averaged across them, for the same reason laid out in Google Ads campaign structure for Australian businesses: a campaign is the smallest unit that owns a budget, so anything you want to control separately has to be a separate campaign.
The exchange rate is a reporting decision, not a detail
Once markets sit in separate accounts, somebody has to add them up. A Google Ads manager account will do it for you, and it is worth knowing exactly how. Google documents that cost data for accounts in different currencies is converted using the average monthly conversion rate for each month in the date range selected, and then says the quiet part out loud: because of this conversion method, the converted amounts you see for client accounts on a given day may not always be accurate. The converted view also refuses date ranges of All time or anything before May 2003.
So a rolled-up regional CPA is an average of averages. That is usually fine for a trend and not fine for a decision about one market in one week. Two habits fix it. Pick one currency of record for the business and one source for the rate, then state the rate in the report so the number can be reproduced. And make market-level decisions in the market’s own currency, in its own account, where no conversion has happened at all.
The same discipline applies on the way in. Conversion values sent to Google Ads carry an ISO 4217 currency code, and if that code is not the account currency Google converts it before it ever reaches a bid strategy. A feed quietly sending SGD into an AUD account is not a reporting problem, it is a bidding problem, because target ROAS is optimising against converted numbers. Worth a line in the checks described in the GA4 conversion tracking sanity check.
Location targeting is where regional budget quietly leaks
Google Ads Help describes the default advanced location option as using both physical location, meaning where a user is located or regularly located, and location of interest, meaning locations the user has shown interest in. Left on the default, a campaign targeted at Singapore can serve to someone sitting in Perth who has been reading about Singapore. Across six markets that is six campaigns each buying a slice of everybody else’s audience, and every market’s CPA is slightly wrong in a direction nobody can see.
For market-level accountability, set presence only, so each campaign buys the market it is being measured on. The exception is real cross-border demand: inbound travel, relocation, study, expat services. There, interest targeting is the whole point, and it belongs in its own campaign with its own budget so its very different CPA cannot contaminate the in-market one.
Keywords are researched natively, not translated
The literal translation of a high-volume English term is frequently not the phrase people in that market type, and a translated keyword list reads as plausible right up until it returns no volume. Research each language natively, with somebody who searches in it, and treat the English list as a starting prompt rather than a source. The landing page then has to be in the language of the ad, or both Quality Score and conversion rate pay for the mismatch. This is the practical half of what our paid search glossary defines as international and multilingual paid search.
Running English-only campaigns across Singapore, Malaysia, the Philippines and India is a legitimate strategy, particularly for B2B, and it is cheaper to stand up than four localised builds. Just be explicit that it is a reach decision rather than a coverage one, and size the market on English search volume rather than on population.
Thin data per market breaks bidding before it breaks the budget
Splitting one account’s conversions across six markets does not give you six signals. It gives you six thin ones. Target CPA and target ROAS both need enough conversions in a recent window to model anything, and a market contributing four conversions a month is not going to supply it. Portfolio bid strategies help, because they pool several campaigns under one shared target, but the pooling is also the catch: one target gets applied to markets with genuinely different economics. A target that is right for Australia at a AUD 120 cost per lead is wrong for the Philippines in both directions at once.
A workable order: start the thin markets on manual or maximise clicks with a cost per click ceiling while conversions accumulate, pool only markets with comparable economics into a portfolio, and let a market graduate to its own target once its own data supports one. The reasoning is the same as in smart bidding without enough conversion data, applied per market instead of per account.
Micro conversions are the other lever, and they travel badly. A phone call is a primary conversion in some markets and a WhatsApp message is the real enquiry in others, so the conversion set is a per-market design question rather than something to copy across. Google Ads conversion tracking in Australia covers how to decide what counts as a conversion in the first place.
Ads in AI search reached APAC before most of the world
This is the part of an APAC paid search plan most likely to be out of date. Google Ads Help currently lists ads in AI Overviews and AI Mode as available in the United States, Australia, Canada, India, Indonesia for AI Overviews only, Kenya, Malaysia, New Zealand, Nigeria, Pakistan for text ads only, the Philippines and Singapore, in English. Eight of those twelve markets are in APAC.
Nothing has to be built for it, which is exactly why it needs attention. Google states that advertisers do not need to create new campaigns or set up separate budgets, and that existing Search, Shopping and Performance Max campaigns are automatically eligible to serve there through the usual auction. So if you are running paid search in Singapore, Manila or Sydney today, some of that spend is already answering an AI generated response rather than sitting above ten blue links.
Two consequences. The landing page is now being reached by someone who has already read a summary of the answer, so a page that restates the summary converts worse than one that moves straight to the specifics. And the surface is English-only for now, which makes the English-language markets in the region the ones where this is live and the localised ones where it is not yet. How to appear in Google AI Overviews covers the organic side of the same surface.
A build order that holds up
- Establish each market’s real search platform mix before allocating any budget.
- Decide the account topology on billing currency and time offset, because neither can be changed afterwards.
- One campaign per market, always, whatever the account topology.
- Set presence-only location targeting, and give genuine cross-border intent its own campaign.
- Research keywords natively, and ship a landing page in the ad’s language before the campaign goes live.
- Fix one currency of record and one rate source for reporting, and make market decisions in the market’s own currency.
- Hold thin markets off conversion-based bidding until their own data supports a target.
- Re-check the AI search surface each quarter, because the market list is still moving.
None of this is exotic. It is the same discipline as a single-market account, applied to the three things that only break when there is more than one market: the currency, the clock, and the question of which market a click actually came from. Get those right and an APAC account behaves like several well-run local accounts. Get them wrong and it behaves like one account with an unexplained average. If you want this looked at properly, our paid search work starts with the account structure rather than the ad copy.