A new client arrives with a budget, a list of GEOs and a deadline. The question that shapes the first two months of work: do we start with a test campaign on a limited budget or scale immediately on a full media plan. Both options have a cost, and the choice depends not on budget size but on the number of unknowns.
01When a test is needed and when it is not
A test is needed when the product has not launched in this GEO or on this source, when there is no historical data on registration-to-deposit conversion, when the landing page and creatives are new and untested on live traffic. A test campaign takes two to four weeks and uses two or three sources across one or two GEOs. The goal is not payback but data: which source delivers deposits, which delivers only registrations, what CPA per first deposit comes out and whether it matches the forecast.
A test is not needed when data exists from a previous vendor or from your own launches in the same GEO, when the product is already running and the task is to grow volume, when certification and tracking are in place and the only question is how much traffic is available at the target price. In that case a test phase stretched over a month is a lost month: the data already exists, and spending time to re-derive it means deferring results.
02What determines the choice
The number of unknowns in the funnel. We count them literally. Whether Google certification will clear in this country is one unknown. Landing-page conversion is a second. The CPA a given source will deliver is a third. If there are more than two unknowns, a test is mandatory.
If there is only one unknown — say a new source with a functioning funnel — it can be tested inside the main campaign: allocate 15–20% of the budget to the trial without stopping everything else.
03Mistakes, and how we handle it
Scaling without data: the client commits the full budget for three months, the vendor opens every source at once, and a month later it turns out that two of five channels produce registrations with no deposits. A third of the budget is wasted, and a test would have saved that money. The opposite mistake is perpetual testing: the vendor adds a new source "for validation" every month and never moves to scaling because the data is "not yet sufficient." At some point testing becomes a way to avoid accountability for results.
A test campaign has a fixed duration and fixed transition criteria. Before launch we agree: at what CPA per first deposit a source scales, and at what point it is shut off. These thresholds come from market benchmarks for the GEO: for Tier-1 the target FTD CPA is usually $150–300, for Tier-3 $20–50. If a source meets the threshold after the test, it enters the main media plan. If not, it is closed without discussion.
The move from test to scale is not a single switch: we first double the budget on the working source and watch whether CPA holds. If it does, we double again. Every source has a volume ceiling, and it is cheaper to find it gradually than to hit it at full budget.
A geo closes mid-flight: what to do with the budget
A regulator does not change the rules at the start of a quarter but on an arbitrary Tuesday. The plan for that day is written before the launch, not on the day of the news.
ReadPostbacks and reconciliation: what to set up before the first click
Data cannot be reconstructed after the fact. Anything that was not recorded at the moment of the action is gone for good — which is why tracking is set up before the launch, not after the first week.
ReadWhy Google Ads certification sets the campaign timeline, not the budget
A certification request takes one week to a month and is tied to a specific country. Until it clears, the budget sits idle.
Read
Calculated for your GEO and budget
Describe the product, GEO and budget — we come back with a media plan in 48 hours. Free and without obligation.
Get the media plan