To start working with a smart model, first, you need to evaluate/measure the effectiveness of the campaign. In order to do this, the campaign must pass a “test”.
Conversion cost starts at $0.01, still, we advise you to set 60-80% of your network’s payout. Thus, the SmartCPA algorithm will be able to pick up more traffic from high-quality zones.
Here you set the amount of funds that will be deducted from your campaign budget if the cost of produced clicks or conversion won’t cover expenses for the tested traffic.
There is no universal answer. We recommend setting your test budget equal to the amount that you usually spend to estimate campaign performance.
However, you should take into account the volume of traffic on the GEO. If the selected GEO has a high volume of traffic (for example, IN, PK, BR, etc.), then it is better to set a higher test budget to be able to test more zones and further, get more traffic.
If the selected GEO has a low volume of traffic, then it is better to use the smaller test budget otherwise, the test will last for several days.
Set impressions limit per zone
This option excludes zones from the test that don’t provide a single conversion for a fixed volume of impressions. This will prevent you from buying unnecessary traffic that does not show good results with your offer.
Early Efficiency Checker
This option checks the efficiency of your campaign in the early stages. So, in case your campaign has a low performance this option can stop the test before it is completed to prevent the loss of your test budget.
Very helpful in case you want to test several offers or landing pages and you don’t want to pay for the complete test for all of them. Campaigns with low performance will be stopped in the beginning.
What is the test?
To evaluate the effectiveness of the campaign, it is necessary to send a certain amount of traffic to the offer and measure the results. You have to set the test budget, which will be used to buy traffic for evaluation. A special algorithm is used to calculate the optimal bid for each specific zone
Tests could be free and compensated.
If the cost of produced conversions covers the expenses for the tested traffic, the test is considered free since you pay only for conversions.
If the cost of produced conversions won’t cover expenses for the tested traffic, the difference is compensated from the test budget. Such tests are called compensated.
For instance, we have “Campaign A” and “Campaign B”.
Campaign A – free test
The testing budget is set to $ 10. This means the campaign has received the amount of traffic worth $10 within the test.
Conversion Cost is 0.5
Number of conversions is 25
Thus, the cost of produced conversions is 0.5 * 25 = 12.5
cost of produced conversions 12.5> Traffic Cost 10
As we can see, the cost of produced conversions fully covered the cost of traffic, that was spent on the test. This way you pay only for conversions.
Campaign B - Compensated Test
The testing budget is $ 10. So accordingly, the campaign has received the amount of traffic worth $10 within the test.
Conversion Cost is 0.25
Number of conversions is 20
Thus, the cost of produced conversions 0.25 * 20 = 5
cost of produced conversions 5 < Traffic Cost 10
As we can see, the cost of produced conversions is not enough to cover the traffic cost. Therefore, the system compensates for traffic costs from the test budget.
So we get
10 TB = 5 Cost of produced conversions + 5 Traffic Cost
In this case, you pay both for conversions and for the traffic provided for the test.
! The larger the test budget, the more traffic the campaign will receive during this test, and more zones will be tested.
Testing zones/traffic sources.
Now let’s take a look at what happens to the zones during and after the test.
Smart campaigns have a self-optimization function. "Set impression limit by zone" section allows you to appoint an allowable amount of impressions from a zone in case it does not produce conversions. This option excludes zones from the test that don’t provide a single conversion for a fixed volume of impressions. This will prevent you from buying unnecessary traffic that does not show good results with your offer.
The main purpose of the test is to find traffic that produces conversions. The zones with conversions are the most lucrative and we want to get more traffic from them. Let’s take a close look at what happens when the test finds a good zone. After a zone produces conversions, the system forms an eCPM at this zone and further buys traffic according to this eCPM. From now on, the needed traffic is purchased both within the test and out of it.
Thus, in the dashboard, you can see what the current test is costing you as well as separate costs of the zones that work on their own eCPM. For this reason, you can see the campaign in the test with a cost exceeding the established test budget.
ECPM calculation example
eCPM = (Conversion price * Total number of Conversions / traffic volume) * 1000
conversion cost = $ 0.5
Total number of Conversions = 10
Traffic spent = 10 000
eCPM = (0.5 * 10/10 000) * 1000 = $ 0.5
End of the test period, summarizing
The test is considered completed when the campaign has received the volume of traffic worth the defined test budget. Now the system should estimate the results according to the given metrics and assign the status to the test. The status of the test will determine how the campaign will work further.
Failed - Either there are absolutely no conversions, or their cost forms insufficient CPM for further work of the campaign. The campaign automatically stops after the test.
Possible reasons for the status:
- Might be that your offer is no longer active
- The postback URL is set incorrectly (for instance, you can track conversions in your external tracking solution, but they are not registered in our system)
- The selected creative is ineffective
Not so good - the campaign showed quite good results during the test. But this campaign is still not good enough to work on all zones (RON). In this case, the campaign’s CPM will quickly decrease and then it will stop receiving traffic. With this status, the campaign will only work in the zones where the conversions were produced.
Possible reasons for the status:
- Low conversion cost (try increasing your conversion cost)
- Low CTR (try using another creative)
- Small test budget (a small amount of zones have been tested/ try increasing your test budget)
Succeeded - the campaign showed excellent results, will work on all zones (RON) and receive more traffic. This status is given to campaigns where the test was considered free. Also, this status can be given to campaigns with a compensated test, because the system relies on various metrics and can predict the further success of the campaign.
The duration of the test depends on the test budget and the volume of traffic on the selected GEO. The test can last from several minutes to several hours for GEOs with a high volume of traffic and from several hours to several days for GEOs with a low volume of traffic. Also, please note that the more specified the campaign targeting, the longer the test will last.
Why the test is needed?
1) Campaign optimization during the test
Zones with low performance are blocked from the test and from the campaign. Also, it helps to form blacklists for future tests.
2) To search and purchase suitable traffic
The campaign purchases more traffic from successful zones.
3) The main purpose of the test period is to form eCPM on the campaign, which will help the campaign to purchase traffic in the system.
For all these reasons the test cannot be passed.
What cannot be done during the test and why?
1) The campaign settings cannot be changed
The algorithm sets up the test based on targeting at the stage of creating a smart campaign. If you change the targeting settings, the test will not work correctly. Therefore, you cannot make changes to the campaign once it was started.
2) The conversion/click cost cannot be changed
The whole test is based on the conversion/click cost. If change these parameters, then it will lead to incorrect test results estimations and wrong predictions made by the algorithm.
3) The test cannot be stopped
The smart algorithm is based on a continuity of working. If you stop the campaign during the test period, then you will not be able to resume it. However, when a company goes out of the test period, it can be stopped and restarted.