Local ROAS does not make countries comparable.
Local ROAS does not make countries comparable. Price, margin, tax, distribution cost, currency, channel maturity and execution capacity transform revenue differently. Convert results to contribution in a common currency, apply exchange and demand scenarios, then respect strategic minima and local ceilings. Hierarchical models can share information but must not erase structural market differences.
Decision and method.
Allocate tranches on risk-adjusted marginal contribution in a common currency, under country-specific presence minima and capacities. Build net price, margin, variable cost, currency, tax and contribution per country, with decision currency and conversion date. Estimate local response with experiments, local/hierarchical models and appropriate references, reporting a range for short history. Test exchange, demand, competition and capacity; calculate central, prudent and marginal contribution. Rank risk-adjusted tranches then apply strategic minima, operational ceilings and exit costs, revising with local results.
Worked example.
An additional €600k must leave at least €100k in each country. Risk-adjusted contributions are €0.84/€ France, €0.79 Germany, €0.706 Spain; gross ROAS would favour Spain. Contribution allocation France 300 k€, Germany 200 k€, Spain 100 k€ gives 252 + 158 + 70.6 = 480.6 k€. ROAS allocation 150/200/250 k€ gives 126 + 158 + 176.5 = 460.5 k€. Economic matrix preserves 20.1 k€ expected contribution while retaining Spanish minimum; review if exchange or capacity leaves range.
Checks and limits.
Results use a common currency and contribution; margin, tax, service and exchange are country-specific; local estimates use documented information sharing; strategic minima and operational ceilings are explicit. Exchange scenario does not replace financial hedging policy. Aggregated cultural comparisons can hide transnational segments. Small countries remain sensitive to hierarchical priors.
Resources and sources.
Download the multi-country allocation matrix. Related: prepare multi-level data, compare risk scenarios, normalise local margins. Sources: Sun et al. (2017); Chandukala et al. (2011); Solnik (1974).