You wanted to find out whether customers in three countries would buy. By the end of the test, most of the money had gone to one. You have a spending pattern, but you may still have no answer about the other two markets.
That is a campaign-design problem before it is a verdict on any country. If learning about each market matters, reserve an allocation for that learning. A global targeting list does not, by itself, promise every country a fair hearing.
Decide what the budget is supposed to discover
There are two different jobs here. One is to buy the best business outcomes available within a shared budget. The other is to learn whether a particular market deserves investment. Sometimes those jobs fit together. Do not assume they always do.
Google describes Maximize conversions as seeking the most conversions within the budget, distinct from maximizing conversion value. Neither objective is a promise to distribute a research budget across your target countries.
If the board wants an answer about Canada, an account that spends elsewhere has not necessarily supplied one. Give the Canadian test an explicit allocation and a commercial question: can it acquire customers at economics the business can support? Repeat that discipline for the other markets you can afford to test.
This does not require identical budgets everywhere. It requires knowing what each allocation is meant to establish, and admitting when a market received too little evidence for a decision.
A cheap country is not a bad customer
The temptation is to exclude whichever country dominates the early traffic. That can throw away a useful market without fixing the experiment.
A software business described exactly this dilemma: its new global Google campaign was drawing most of its initial traffic from Pakistan, yet some leads looked promising. It wanted to keep those opportunities while testing North America and Europe. The useful choice was not whether to dismiss Pakistan. It was whether one allocation could honestly serve both purposes.
Keep the inexpensive market under a bounded budget if the leads deserve follow-up. Fund the other intended markets separately. Judge the first group by qualification, paid activation and customer value, not by whether its click price or country looks respectable in a presentation.
But do not call it a winning market too soon. The same software business reported that its tracked subscription activation had not happened at all. That materially changes the interpretation: early geographic concentration is not evidence that the campaign has already found paying customers cheaply. The account does not establish why that delivery pattern occurred or verify the tracking; it establishes why a confident success story would be premature.
Protect the allocation, not just the label
On Meta, the structural question often arrives as a choice between one pooled ad set, country-specific ad sets or separate campaigns. Begin with the control you need rather than a universal naming convention.
Before launch, inspect where the budget is set, what can reallocate it and whether the intended country test actually has protected funding. Separate labels make a dashboard easier to read; they are not enough if the allocation you need can still disappear elsewhere.
Country ad sets inside one campaign may be easier to manage. Separate campaigns may make the intended budget boundary clearer. Do not choose between them on an unsupported promise that sharing a campaign automatically transfers useful learning between ad sets. Google and Meta are different systems; a justified setting on one is not documentation for the other.
Nor should separation become permanent bureaucracy. Once countries show sufficiently similar customer economics and no longer need distinct decisions, combining them can be a reasonable management choice. The boundary exists to answer a question, not to multiply campaigns indefinitely.
Do not repair missing evidence with invented values
When paid conversions have not arrived, changing the goal to account creation can make the dashboard busier without answering whether the market will pay. Keep that event observational until there is a credible relationship to customer quality. Do not invent a dollar value for signup merely to give bidding something to chase.
Likewise, a small budget may require fewer markets, not a more elaborate structure. Choose the most consequential market question you can afford to investigate. There is no useful universal minimum that makes every product, country and purchase cycle comparable.
Audit targeting too. Google’s advanced location options distinguish people likely present in a location from people showing interest in it. If the question concerns customers physically in a market, inspect that setting. If people abroad are legitimate prospects, interest may be appropriate. Location signals are not perfectly accurate.
Finish the test with three honest categories: commercially promising, commercially disappointing and not meaningfully tested. That last category is the one pooled budgets make easiest to overlook. Buying traffic around the world is not the same as learning where the business should grow.