The first thing people assume about running ads in seventy markets is that the hard part is targeting. It is not. Targeting is a settings screen.
What breaks is operations. Seventy markets means seventy sets of creative, seventy compliance positions, seventy currencies and timezones to reconcile, and a review cadence that no longer fits in a weekly call. Programs do not fail at market forty because the ads stopped working. They fail because nobody can see what is happening anymore.
Tier your markets before you touch a campaign
Treating every market equally is the most expensive mistake in multi-GEO. Most programs have a handful of markets carrying the revenue and a long tail that is either growing or should be cut. Sort them deliberately.
| Tier | Budget share | Creative treatment | Review cadence |
|---|---|---|---|
| Tier 1: proven, revenue-carrying | ~60% | Native, market-specific, full production | Weekly with a human |
| Tier 2: scaling, proven demand | ~30% | Native generation from tier 1 winners | Biweekly |
| Tier 3: tests and new entries | ~10% | Adapted, minimum viable localization | Monthly, or on threshold |
The tiers are not permanent. A tier 3 market that hits its CPA target for two consecutive months gets promoted and gets budget. A tier 1 market that stalls for a quarter gets demoted before it quietly eats the budget. Write the promotion and demotion rules down, or the tiers will just reflect who argued hardest in the last meeting.
The account structure question
Consolidate markets into fewer accounts, or split them out? Both are defensible and the wrong choice is expensive to reverse.
| Dimension | Consolidated | Split per market |
|---|---|---|
| Algorithm learning | Faster, more conversion data per campaign | Slower, thin data in small markets |
| Budget control | Harder, spend drifts to easy markets | Precise per market |
| Compliance isolation | Weak, one problem can affect everything | Strong, damage stays contained |
| Reporting clarity | Needs disciplined naming | Clean by default |
| Operational overhead | Lower | Higher, needs automation |
Our default: consolidate within a regulatory bloc where the rules are genuinely shared, split across blocs. Licensed verticals get split further, because compliance isolation is worth the overhead the first time a market has a problem.
Creative: generate natively, do not translate
Translation carries the words and loses the idiom, the reference, and the offer framing that made the original work. Generating natively against the market's own brief consistently beats translated creative, and the gap is wide enough to matter at scale.
This is where automated creative pipelines earn their keep. The concept comes from the team, the modules are approved centrally, and generation runs per market against local constraints. You get market-appropriate volume without seventy separate design queues.
Currency, timezone, and the reconciliation trap
At one market, currency handling is a detail. At seventy it is a reporting integrity problem.
- Pin the FX conversion date and disclose it. Converting at today's rate makes last quarter's performance change every time you open the report.
- Report in both. Local currency for the people running the market, one reporting currency for the people running the budget.
- Pick one reporting timezone and never move it. Otherwise "Monday" means different things in different rows of the same table.
- Reconcile per market, not just in aggregate. Aggregate totals can look right while two markets are wrong in opposite directions.
Compliance is per-market, never global
A creative approved in one market is not approved anywhere else, and a pipeline that treats approval as a global property will eventually serve the wrong asset in the wrong country. Every asset carries the market it was cleared for and is only eligible to serve there.
In licensed verticals this is the constraint the whole program is built around, not a step at the end.
What to centralize and what to localize
| Centralize | Localize |
|---|---|
| Brand system and creative modules | Copy, idiom, and offer framing |
| Budget envelope and tier rules | Bid targets and pacing within the envelope |
| Reporting structure and definitions | Nothing. Definitions must be identical. |
| Compliance framework and tooling | The actual rules per market |
| Testing methodology | What gets tested and when |
The rule of thumb: centralize the system, localize the content. Teams that localize the system end up with seventy different definitions of CPA and no way to compare markets.
The operational rhythm
Seventy markets is only manageable if most of the work happens without a meeting.
- Continuous, automated: bid and pacing adjustments inside guardrails, creative rotation on fatigue, anomaly flags, reconciliation checks.
- Weekly, human: tier 1 markets reviewed properly, anomaly flags triaged, promotion and demotion candidates identified.
- Monthly, human: tier changes applied, tier 3 tests killed or graduated, creative refresh planned from what won.
- Quarterly, strategic: market portfolio reviewed, new entries chosen, exits decided.
If your weekly call covers all seventy markets, you are not reviewing them, you are reading a list.
Where multi-GEO programs break
- Duplicating campaigns by copy-paste. The settings come along, including the ones that were wrong for the new market.
- One approver for every market. They become the bottleneck, then start approving without reading.
- A single global budget with no per-market floor. Spend concentrates in whichever market converts most easily today, and your tier 3 pipeline quietly dies.
- Ignoring what tier 3 learns. Small markets are cheap places to test angles that later carry tier 1.
- No market-level kill switch. When one market has a problem you need to stop that market, not the program.
Get the tiering, the structure, and the rhythm right and adding market seventy-one is a routine operation rather than a project. That is the whole objective.