Introduction
The First 90 Days of the Partnership Campaign: Learning Scenario and Control Points
Below is not a real case of a specific partner LuckyBear and not a promise of results in three months. This is a learning scenario: how to break down the first 90 days of work into stages, so as not to mix testing, optimization and scaling into one process.
The main goal of such a plan is not to “get a certain income by the date”, but to accumulate enough data to understand the economy from sources and make the following decisions without guessing.
Stage 1. The First 30 Days: Make Data Readable
At the start, it is better to limit the number of variables. Choose a clear source, one or two GEOs, and a small number of creative hypotheses. Assign your own SubID to each campaign. If postback is available, check the transmission of events before volume increases.
| What to fix | Why? |
|---|---|
| source/site | separate channels from each other |
| GEO | Distinguish between the markets |
| Creativity/placement | Understand what brought the user |
| clicks and registrations | funnel-top |
| confirmed by FTD | quality |
| expense | consider the actual economy |
At this stage, it is dangerous to draw conclusions only by CTR. Creativity can garner a lot of clicks and still bring in users who don’t go any further.
Stage 2. Days 31-60: Finding a bottleneck
After the basic statistics appear, look at where the audience is lost. If clicks are enough, but registrations are few, check the correspondence of the message and landing. If there are registrations, and there are few confirmed actions, check the quality of the source, localization, technical attribution and offer rules.
Change one major variable per test. For example: a new first screen with the previous source or a new creative with the previous landing. It’s not always possible to be perfect, but this approach makes the conclusions much more useful.
Stage 3. Days 61-90: Scaling only what is understandable
Before increasing your budget, answer four questions:
- Which segment generates most of the confirmed result?
- Does the economy continue after the increase in volume?
- Are there signs of burnout of the creative or the audience?
- Does your report match the affiliate statistics?
If there is no answer, increasing the volume turns the test into a more expensive test.
An example without a promise of income
Imagine two campaigns with the same advertising budget. Campaign A brings more clicks, but below is registration → FTD. Campaign B is more expensive by click, but the proportion of confirmed actions is higher. Choosing a CPC winner would be a mistake: calculate the cost of a validated FTD and, for RevShare, the further value of the cohort.
Such an example shows a principle, but does not set "normal" values. They depend on the source, GEO, offer and period.
What should I do after 90 days?
- SubID structure that can be read without manual decryption
- a table of costs and results by segment;
- a list of tested hypotheses and conclusions;
- understanding of the weak phase of the funnel;
- Rules under which a campaign scales or stops
- agreed with the manager parameters of the offer.
For monthly reconciliation, you can use the structure of the material about partner's report.
