If you need to take action to improve the campaign performance although you've waited enough time for optimization per campaign goal, you can consider the following items.
Note that it takes time for our machine learning model to adapt to the new conditions. Therefore, monitor and wait to see stabilized performance for at least a week after the setting changes. It’s not recommended to change settings too frequently — which can cause a performance drop.
Budget
Check you are following the best practices below.
-
If you change the budget, you need to wait for the performance to be stabilized for a certain period of time. Please avoid budget changes, especially during the first two weeks of the campaign. And when you have to change the budget, we recommend changing it within 30% of the current budget. Keep in mind that it can take a few days for campaign performance to be stabilized after the budget change.
-
Weekly Budget Optimizer will optimize the campaign to spend on the best days and times of the week. The daily spend may vary up to 50% of the average daily budget, but will not spend more than 7x your average daily budget within a week (Monday to Sunday). You can activate Weekly Budget Optimizer by selecting the Average Daily Budget option (default setting) when you set up the budget. Read more about the Weekly Budget Optimizer
-
If target CPI or target CPC is selected as the Bid control priority when setting up a campaign, it is critical to set the appropriate target cost considering the market price and the app you want to promote. If you've changed the target CPI or CPC, our machine learning model needs a learning period to meet that goal. And if you set a target cost that is too low for the market price, your budget may not be spent. If you don't know the proper CPI and CPC for the market and your app, use the Budget as the Bid control priority first and check the actual CPI and CPC.
Traffic
Fully utilize all traffic options and reach more audiences.
-
Under Advanced Settings of the campaign settings page, you can select traffic targeting options related to LAT and SKAdNetwork(iOS). However, we don't recommend using this feature to target certain traffic segments since campaign performance changes depending on the market share of each segment. We recommend maintaining the default settings except for special and experimental purposes.
-
Probabilistic attribution allows you to measure advertising performance even without ADID, and provide us with more contextual data to help our model to learn. In addition, setting up a view-through tracking link helps maximize campaign optimization with more insight and data about users who viewed ads.
Creative
Utilize all available creatives referring to Moloco creative guide.
-
We strongly recommend using all creative formats (image, video, and native) and all required sizes of banners for the target country. Each ad inventory supports different creative formats, so missing creative formats can mean missed opportunities to show your ads on high-performing inventories. For image banners, refer to the guide to set up all the banners of the required sizes for each country.
-
Auto-Generated Creative feature allows you to automatically generate images, native images, and video end cards with all required sizes. We generate those auto creatives based on the app store information of your app. The more sizes and formats of creatives you use, the more inventories you can reach, which helps campaign optimization. Simply click Generate button to leverage the feature when setting up creatives or a creative group.
-
Kakao (Korea targeting campaigns only) and Samsung Exchanges have different creative guidelines from other exchanges. If you don't utilize those creatives yet, add creatives to your campaigns refer to the Samsung Exchange guide and Kakao exchange guide to leverage those ad inventories.
- Dynamic Product Ads(DPA) is a personalized ad showing the most suitable products or content based on user's behavior through Moloco's machine learning engines. It can promote various categories of apps such as E-commerce, food delivery, travel, online cartoons, and OTT. Check Dynamic Product Ads Guide to start integration.
Postback
-
Full postback means transmitting all event data including both Moloco-attributed and unattributed events. This allows Moloco's machine learning model to obtain as much data as possible about user behavior in the app, increasing the accuracy of predicting user actions.
-
Throughout users' journey from acquisition to the final conversions, various in-app events occur. Understanding and analyzing this user funnel is critical to campaign optimization. Especially for the re-engagement campaign, it is essential to understand the user behavior in the app, so forward all types of MMP events via postbacks.
-
If the targeted optimization goal events do not occur frequently, our model cannot learn properly. Change the optimization goal to more upper funnel events that frequently occur. See more detailed explanations on the marketing funnel and how to select the appropriate campaign goal
Target
Target restricts the range of audience of your ad campaign. You can enhance performance by expanding the target audience or refining target settings.
-
The device language of target audiences can be different from the language of the country where the target user resides. For example, if you target the Korean language for your campaign targeting Korea, you can't reach some users who set up their device language as English.
-
Use location targeting only when advertising should be exposed or restricted to specific locations. Moloco can't receive the location information of users in some cases, causing a significant reduction in the target size.
-
We recommend not blocking or targeting certain publisher apps for better campaign performance. Read target settings FAQs for more detailed explanations.
-
If you set up a limited target for a re-engagement campaign, the campaign might not be optimized well enough due to the limited pool of audiences and data that our model can learn from. For example, if you target users who have added certain categories of items to their shopping cart, it may limit the learning process and make the campaign underperform. In this case, expand your target audience. Your campaign may perform better with users who installed the app or the users who've added any category of items to their shopping cart.
Frequently asked questions
When is the right time to try campaigns with a new optimization goal?
To start with an install optimization campaign and then launch a ROAS or in-app optimization campaign later, start the second campaign after 2 weeks of postback integration or 2 weeks after the launch of your install optimization campaign. Doing so will make sure that you have enough in-app data.
What should be considered when making significant changes in the budget?
When there's a drastic increase to the campaign budget, there might be spikes in CPI, CPA, or other important metrics. Additionally, a drastic budget change might also cause overspending or underspending issues. We recommend changing the budget within 30% of the current budget and keeping in mind that it can take a few days for campaign performance to stabilize after the budget change.
What happens if we modify target settings of a running campaign?
Our machine learning model will restart its optimization if you change the target settings while the campaign is running. Therefore, we don't recommend changing the target settings — it can create drastic changes in inventories or target audiences, and hurt performance. If the change is necessary, wait for the model to optimize with the new targeting.
Can I restart a campaign that has been in "Completed" status? How long will it take to resume the campaign?
Change the end date of the campaign to sometime in the future. This will change the campaign status from "Completed" to "Paused", but note that the change would take some time(approx. 1 - 2 hours). Then, you'll be able to activate the campaign again. It should resume within an hour.