Guidelines for working with Unity Ads || iKon

CPE Campaigns

When launching CPE campaigns, it is recommended to use a sufficiently competitive bid to help the campaign enter auctions more quickly, gather the required volume of data, and complete the learning phase faster.

The bid can, and often should, be set 50–80+% higher than the target CPA.

For example, in Argentina, a starting bid in the range of $20–25 has often proven effective in achieving stable delivery and generating enough data for the algorithm to learn. Once performance stabilizes, the bid can be gradually reduced while monitoring traffic volume, registration cost, and CPA.

Avoid making abrupt bid changes, as they may negatively affect the stability of the algorithm's learning process.

ROAS Campaigns

For ROAS campaigns to optimize correctly, it is essential to send deposit revenue to Unity Ads. Without revenue data, the algorithm cannot effectively optimize toward users with high lifetime value.

When launching a campaign, it is recommended to start with moderate ROAS targets.

For example, in Argentina, an initial D7 ROAS target of 6–8% typically allows the algorithm to complete the learning phase more quickly and achieve stable delivery. Once sufficient data has been accumulated, the target can be gradually increased depending on campaign objectives.

When optimizing ROAS campaigns, keep the following in mind:

• Lowering the target ROAS generally increases traffic volume.

• Increasing the target ROAS may improve audience quality and average deposit value but typically reduces volume.

• Setting an excessively high ROAS target can significantly limit scale, as the algorithm begins searching only for the highest-value users, often through more expensive placements.

• Any target adjustments should be made gradually, allowing the algorithm sufficient time to adapt.

It is also important to remember that ROAS campaigns require a longer learning period than CPE campaigns. Performance should only be evaluated after a sufficient amount of data has been collected.

Daily Budget

The daily budget directly affects a campaign's ability to participate in auctions and successfully complete the learning phase.

An insufficient budget may limit impressions, slow down algorithm learning, and lead to unstable delivery. In most cases, it is recommended to allocate a budget that allows the campaign to generate a sufficient number of optimization events each day.

During the launch phase, frequent budget changes should also be avoided.

Campaign Structure

A best practice is to use 2–3 unique creative packs within a single campaign.

This approach allows you to:

• Identify top-performing creative combinations more quickly.

• Analyze Creative Testing results more effectively.

• Allocate budget more efficiently to the best-performing creatives.

Whenever possible, avoid using the same creatives across multiple campaigns for the same app, as this may create internal auction competition.

Playable Creatives

Since Unity Ads is a gaming-focused advertising network, playable creatives often outperform traditional video creatives.

Interactive creatives can help:

• Increase CTR and CVR.

• Improve user engagement.

• Enhance traffic quality.

• Increase campaign competitiveness in auctions.

Even a single high-quality playable creative can significantly improve campaign performance compared to using video creatives alone.

Testing Different Optimization Strategies

In addition to D7 ROAS, it is recommended to test D28 ROAS optimization.

For D28 ROAS campaigns, the initial target can generally be set higher than for D7 because the algorithm optimizes toward a longer payback window and considers users' long-term value.

However, these campaigns require a longer learning period and closer performance monitoring.

App Selection

It is recommended to continuously expand the pool of tested apps and publishers.

Like most machine learning-driven traffic sources, Unity Ads performs significantly better when it receives a sufficient number of relevant optimization signals (postback events). Apps with a larger amount of historical data often enable the algorithm to identify high-quality audiences more quickly and achieve stable performance sooner.

Additional Recommendations

• Avoid making multiple major campaign changes simultaneously (bid, budget, targeting, or creatives).

• Scale budgets gradually to minimize the risk of triggering a new learning phase.

• Refresh creatives regularly, as even high-performing creatives naturally lose effectiveness over time.

• Send the most comprehensive set of postback events possible (registration, deposit, and revenue) to provide the algorithm with more optimization signals.

• Avoid making optimization decisions based on small data samples. Unity Ads' machine learning algorithms require sufficient time and enough signals to optimize effectively.

Useful Contacts:

t.me/Gleb_iKon - Head of Service Delivery

t.me/DanilP_iKon - BDM