- Google publishes 30 conversions, and 50 for Target ROAS, as the sample needed to evaluate results accurately. [1]
- It is an evaluation window, not an activation threshold. The industry uses it as the second thing. [1]
- Google says Smart Bidding can optimize on data from all your campaigns, so a new one is not starting blind. [1]
- Portfolio bid strategies optimize across a group of campaigns, with a cross-account version at manager level. [2][3]
- Portfolio strategies are not available for Performance Max. [2]
- One Merchant Center data source can target several countries, and currency converts automatically. [4][5]
- A different primary language is what forces a second data source. Currency alone does not. [4]
- Split markets for control you will use. Pool them so the result stays readable.
The meeting goes the same way in every company that crosses its first border. Somebody wants eleven campaigns because eleven markets. Somebody else wants one campaign because the algorithm needs volume. Both cite the same number, neither has read the page it comes from, and the structure that results is usually a compromise nobody can defend six months later.
So start with the page. Google’s Smart Bidding documentation says that to evaluate results accurately, it recommends measuring performance over longer time periods that have at least 30 conversions, such as a month or longer, and 50 conversions for Target ROAS. [1]
Read the verb. Evaluate. Google is blunter elsewhere on the same page: you can turn on Smart Bidding without prior data, and the conversion baselines it recommends exist to help the system calibrate and to let you evaluate accurately. [1] The industry only ever repeats the first half of that job description.
Google does acknowledge a volume dependency elsewhere in the same place, noting that for best results some Smart Bidding strategies rely on a minimum volume of historical conversion data depending on the campaign type, and sending you to each strategy’s own page rather than publishing one universal figure. [1] That is a much weaker and more honest claim than the folk version, and it is the one the industry chose not to repeat.
Why the distinction changes the structure decision
If 30 conversions were an activation threshold, consolidation would be a technical necessity. Small markets would have to be merged or they would simply not work, and the argument would end there.
It is not a threshold, and the reason sits one sentence earlier in the same paragraph. Google states that Smart Bidding can optimize based on data from all of your campaigns, so even new campaigns without data of their own may notice increased performance. [1] A campaign with no history is not excluded. It borrows. Campaign borders are an organizational choice you impose, and the model is not sitting inside one waiting for it to fill up.
What a thin market really costs you, then, is not learning. It is legibility. Eight conversions in Slovakia last month is a number you cannot act on, cannot defend in a review, and cannot use to justify either increasing or cutting the budget. The campaign may well be running fine. You just have no way to know, and that is a reporting problem wearing a bidding costume.
Separate the two things everyone bundles
The useful move is to stop treating structure as one decision. It is two, and they have different right answers.
Budget control wants separation. One campaign spanning several countries will spend where conversions come cheapest, which is nearly always the biggest, lowest-CPC market. The small markets get starved while the blended result looks healthy, and nobody notices because the campaign is hitting its target. That is not the bidder misbehaving. It is a single blended goal being obeyed exactly. It is also our read of the mechanism rather than Google’s. What is documented is that structure changes where budget lands. [7]
Learning wants pooling. Which is what a portfolio bid strategy is for. Google defines a portfolio bid strategy as an AI-powered, goal-driven strategy that groups together multiple campaigns, ad groups, and keywords, and offers a cross-account version at manager level. [2][3] One caveat worth knowing before you plan around it: portfolio bid strategies are not available for Performance Max. [2]
Put those together and the default structure for a multi-market account writes itself. Separate campaigns per market, so budget and reporting belong to the market. A shared portfolio strategy across them, so the bidding is not trying to learn eleven times from one-eleventh of the data. Different targets only where the unit economics genuinely differ, which is a real thing in Europe once shipping and returns vary by country and a single blended target misprices the smaller markets.
How many conversions does this market produce in a month?
Do the unit economics differ materially from your main market?
Does this market need its own primary language?
Does anyone outside the media team control this budget?
Consolidate it
Same language, same economics, one pooled budget, and not enough volume to read on its own. Fold this market into a multi-country campaign and keep the country as a reporting dimension rather than a structure.
Own campaign, shared portfolio strategy
You need the budget separated and the learning pooled. Give it a campaign so the money is ring-fenced and attach it to a portfolio strategy with your other markets so the bidding is not starting from nothing.
Fully separate, with its own target
Different economics and someone accountable for the number. This market has earned its own campaign and its own target. Do not average it into anything, and if the volume is thin, expect to wait before the read means much.
The feed side, where the rule is the opposite of what people assume
Structure conversations tend to assume that every new country means a new everything. On the Shopping side that is measurably not true, and the actual dividing line is worth memorizing because it is not the obvious one.
| What changes in the new market | Do you need a second data source? |
|---|---|
| Same language, same currency | No. Products serve both markets |
| Same language, different currency | No. Currency conversion happens automatically |
| Different primary language | Yes. A separate data source, translated |
You select additional countries by editing the Countries field of your product data under data source setup, and currency conversions take place automatically so the shopper sees their local price. [4][5] A different primary language is what forces a separate, translated data source. [4]
That is a genuinely useful asymmetry. Austria and Germany do not need two feeds. Belgium might need two, for reasons that have nothing to do with money. And because the feed decides which auctions you enter, getting this wrong does not produce a small inefficiency. It produces products that are simply not eligible.
Localization does not end at the feed either. Google’s own advice is to localize fully: translate landing pages and the checkout flow into the target country’s language, and get price, tax, and shipping right for the destination. [8][6] Translated ad copy pointing at a checkout in the wrong language does not break a policy. It breaks the conversion rate, and nothing sends you an email about that.
The honest summary
Split a market when you will use the control that splitting gives you. A budget someone else owns. Economics that genuinely differ. A language of its own. Pool the learning underneath that split, because the platform was already pooling it and pretending otherwise costs you.
And treat the 30 as what the documentation actually calls it. Not a gate the algorithm has to clear, but the amount of evidence you need before you are allowed to have an opinion. Most multi-market accounts have eleven opinions and evidence for three.
Sources
- Google Ads Help · About Smart BiddingThe 30 and 50 conversion guidance, stated as an evaluation window, the line that Smart Bidding can optimize on data from all your campaigns, and the FAQ on turning it on without prior data.
- Google Ads Help · Portfolio bid strategy: definitionOptimizing across multiple campaigns, and the Performance Max exclusion.
- Google Ads Help · About cross-account bid strategies
- Google Merchant Center Help · Show products in multiple target countriesAdding countries to one data source, and the language rule that forces a second one.
- Google Merchant Center Help · About currency conversion
- Google Merchant Center Help · Set up Shopping ads and free listings in multiple countriesLanguage, price, tax, and shipping expectations per destination country.
- Search Engine Land · How campaign structure shapes Google Ads performanceHow structure moves budget. This piece also repeats the 30-to-50 learning-phase claim that this article disputes.
- Google Merchant Center Help · Supported languages and currenciesThe recommendation to translate landing pages and the checkout flow, and that any supported language may be used in any market.
Frequently asked questions
Does Google require 30 conversions a month for Smart Bidding?
Not as a condition of running. Google’s Smart Bidding help page recommends measuring performance over periods with at least 30 conversions, 50 for Target ROAS, in order to evaluate results accurately. It separately notes that some strategies rely on a minimum volume of historical conversion data depending on campaign type, and points to the individual strategy pages rather than publishing one universal figure.
Does each campaign learn only from its own conversions?
No. Google states that Smart Bidding can optimize based on data from all of your campaigns, so even new campaigns without data of their own may notice increased performance. It also notes that relevant keywords can be added to low volume campaigns to expand targeting and increase conversions. Campaign borders are an organizational choice, not a wall the model cannot see across.
When should I split a market into its own campaign?
When something about that market genuinely differs and you need to act on it: materially different unit economics, a different language, a budget that finance controls separately, or a regulatory constraint. Split for control you will actually use. Splitting purely to give a market its own bidding target usually buys nothing and costs you readable data.
What is a portfolio bid strategy?
A goal-driven bid strategy applied across a group of campaigns rather than one. Google describes it as optimizing bids across multiple campaigns with a single place to change bidding settings for all of them. There is a cross-account version at manager level, and it is not available for Performance Max.
Can one Merchant Center data source serve several countries?
Yes. You select additional countries by editing the Countries field of your product data under data source setup. Currency conversion happens automatically, so a shopper sees the price in their local currency without you creating a second feed.
When do I need a second data source?
When the language changes. Google’s guidance is that if a new country uses a different primary language, you should create a separate data source with the product data translated into that language. Currency alone does not force a split. Language does.
What is budget cannibalization across markets?
One campaign covering several countries spends where conversions are cheapest, which is usually the largest and lowest-CPC market. The smaller markets quietly receive almost nothing while the campaign as a whole reports a healthy blended result. It is not a bug in the bidding. It is the bidding doing exactly what a single blended target asked for.
How do I keep small markets from being starved without splitting everything?
Give the small markets their own campaigns so budget is theirs, then attach a shared portfolio strategy so the bidding still learns across the set. That combination separates the two things people conflate: budget control, which you want per market, and learning, which you want pooled.
Is translating the ads enough to enter a market?
No. Google’s advice for Shopping is to translate landing pages and the checkout flow into the target country’s language, and to get price, tax, and shipping right for the destination. Translated ad copy pointing at an untranslated checkout does not break a policy, it breaks the conversion rate.
Mikołaj Salecki
Writes about media, tech, and AI business for people who actually run digital. Former agency lead. Skeptic of frameworks that read better than they perform.
