UA Media Buying Model

Consumer Acquisition by Brainlabs Media Buying Model

As a complementary tool to our Ad Concept Model, we are introducing our Media Buying Model which uses a similar matrixed approach to systematize best practices for user acquisition campaign development. Any mobile app can leverage a unique combination of distinct UA strategies and campaign parameters to develop a robust testing harness for ongoing mobile campaigns across social platforms.

In its most generic state, our Media Buying Model looks like the matrix below, prior to building out a mobile app user acquisition strategy. The entire matrix is tailored to a respective ad network and operating system requirements and constraints. Geographies, languages, and bid types are all modified based on the client’s needs. Each campaign is structured to meet specific goals at the intersection of strategies and parameters.

The entire matrix is tailored to a respective ad network and operating system requirements and constraints. Geographies, languages, and bid types are all modified based on the client’s needs. Each campaign is structured to meet specific goals at the intersection of strategies and parameters.

Testing on Android

Even prior to IDFA loss, Consumer Acquisition conducted creative and UA testing on Android because it was less expensive, and results translated well to iOS. We saved iOS testing for the rare clients without an Android app or those who were solely targeting iOS users.

Android Capabilities

• Unlimited ad accounts and campaigns
• Unobstructed deterministic tracking
• Comprehensive A/B testing with full reporting
• Distributed scale and spend across campaigns
• Rapid testing, unlimited test campaigns
• Success drivers easy to identify and scale
• Google Firebase enables tROAS, delivering ~15% lift on CPA campaigns with better user mapping

iOS IDFA Impacts

• Many accounts but only 9 campaigns, 5 ad sets
• SKAN tracking disabled after 24-48 hours (See AppsFlyer Conversion Studio below)
• A/B testing in DCO blobs, no asset reporting
• Fewer campaigns, higher budgets, more significant edits
• Restricted testing, only 9 campaigns
• Aggregate insights drive slower optimization
• Post-IDFA, iOS lookalike audiences have lost -65% of their reach and effectiveness

Testing on iOS with AppsFlyer

If you use AppsFlyer on iOS, we recommend testing through Conversion Studio, released September 9, 2021. Conversion Studio helps maximize LTV measurement for post-install activity by:

• Extending post-install event collection up from 24 hours to 72 hours and enabling D3 optimizations
• Tracking multiple iOS KPIs instead of only one in SKAN
• Measuring sequential events in the conversion funnel
• Marrying SKAN data to AppsFlyer attribution
  

Why use a model like this?

Our industry expertise comes from managing over $3.5 billion in creative and paid social spend for the world’s largest mobile apps and performance advertisers. We run our tests using our software AdRules via Facebook, Google, and TikTok APIs. Our Media Buying Model systematizes the UA campaign process, so we know what works, why it works, how to do more of it, and how to easily communicate status. Our process is designed to save time and money by optimizing quickly to reduce non-converting spend and to scale as pockets of efficiency are uncovered.

Below, see our Media Buying Model built out for sample mobile app campaigns across Facebook Android, Facebook iOS SKAN, Google iOS SKAN, and TikTok iOS and Android.

Facebook Android Media Buying Model

On Facebook, UA campaign goals may include:

• High scale AAA campaign in top-performing countries
• Broad targeting split by geo, using CBO to auto-optimize with additional ROAS boost from the top-performing demo.
• Split broad targeting by time zone, using CBO to auto-optimize.
• High Performing LAL audiences based on Primary Metric Value, using CBO to auto-optimize.
• Exploratory spend split by interest groups, using CBO to auto-optimize.

Benefits of Testing on Android

• Android unimpacted by IDFA
• Optimization maintains deterministic efficiency
• Android Facebook primary for A/B creative and audience testing
• Rollout Android Facebook winners to iOS and other platforms

Campaign Strategies

• Scale: increase installs
• Exploratory: test creative, audiences to feed scale
• Interests: bridge volume of broad and precision of lookalikes

Audience Strategy

• CBO structure allows for maximized testing/learning
• The highest-performing ad sets are relaunched as standalone campaigns to increase scale and performance
• Initial 3% Lookalike likely to drive the best performance
• High-performing LALs are immediately scaled as new campaigns
• High-performing audience seeds are recreated with Frequency parameters
• Use FB activity to create Lookalike Audiences (ex: Page Engagement, IG Engagement, Video Views, App Value-Based Lookalikes)
• Purchase: Last 180 days
• Frequent app users: Last 90 Days

Interest Group Testing

• Build interest clusters using contextual targeting best practices leveraging personas, likes, pages, competitive apps, etc.
• Ongoing identification and testing of iterative Interest Groups as campaigns progress

Creative

• Utilize top creative combinations – image/video + body copy, headline, CTA
• Creative Format testing- DCO, Carousel, PAC
• Ad Copy Testing

Creative Testing

• Phased creative testing, conducted on Android with winners transferred to iOS campaigns

Facebook iOS Post-IDFA Media Buying Model

IDFA Compliance

• Embrace SKAN limitations: 9 campaigns max, 5 ad sets per campaign max, max 45-60 permutations
• Optimize account to achieve a minimum of 128 conversions (installs) per campaign per day

Campaign Strategies

• Scale: drive volume + performance
• Exploratory: Test VO vs. AEO and Broad vs. Top Demos with single vs. bundled Geos to identify highest-performing Optimization/Geo/Audience (lowest CPT cost per trial)

Creative

• Move winning creative from Android
• AAA most successful with groups of high performing creative and copy

Scale

• VO based on $CPT insights from Exploratory Campaigns
• Initial AAA campaigns segregated by Geo can bundle US/CA/UK to reduce SKAN campaign slots (only if needed)
• Campaigns 1-2-3 budgets must be sufficient to support a minimum of 128 installs/day

Exploratory

• Leverage CBO structure in all campaigns to test individual/bundled Geos within single campaigns
• Test Broad vs. Top Demo in campaigns 5, 6
• Test AEO vs VO in campaigns 4, 5
• Pinpoint opportunity for volume and $CPT efficiency by testing low-CPM images vs. video

Measurement Limitations for iOS Campaigns

• Pre-IDFA Loss, iOS < 14.5
• Lifetime deterministic tracking provided in AppsFlyer
• 28 days of attribution across Facebook, Google, Snap, and Tiktok for bid optimization and creative testing

Measurement Post-IDFA Loss: iOS 14.5+ Limits

• User is anonymized; install and event reporting delayed 24+ hours
• Limited to a single conversion value; only one KPI can be reported back to your ad campaign
• Limited to 24 hours post-install for event attribution timeframe
  • Event reporting is triggered after a minimum of 24 hours of inactivity
  • SKAN provides 6 bits of information to hold your conversion event data
  • This conversion value is set each time a user triggers specific events in your app, like a purchase
  • Each time you set it, a 24-hour timer starts, and at the end of that timer, your conversion will be reported back to the ad network with a random delay buffer added
  • Each time you set an event, the 24-hour timer restarts, potentially delaying your reporting to 64 days post-install!

Google iOS SKAN and Android Media Buying Model

On Google, UA campaign goals are progressive:

• Identify primary event performance and establish a baseline for video creative
• Test messaging with upper-funnel campaign type
• Event testing with a creative baseline established by geo
• Continue message testing refinement
• Full expansion by platform with top CPA event and performance messaging determined
• Ramp up top event campaign from the previous week
• Revenue optimization if FB enabled

Exploratory, Expansion, and Scale Campaigns:

Learning > Scale

• Single Event CPA, graduated to event testing by Week 2
• All Purchasers Last 180 days

Event Testing and Consolidation

• Supplement opening creative tests with Install + campaign
• Week 2 CPA event tests

Creative

• Video forward – new concepts weekly
• Ad Copy Testing
  • Low volume description/headline swaps weekly post-learning phase
  • Messaging testing by ad group splits for top volume CPA campaign.

TikTok iOS and Android Media Buying Model

• Behavior Audiences, Interest Categories, Interest Keywords: Identify audiences/strategies to feed scale campaigns. Bridge the volume of a broad audience and the precision of a LAL.
• Lookalikes, Time Zone Targeting, Broad Top Demo: Increase KPI performance

On TikTok, UA campaign goals may include:

• Targeting split by behavior, using CBO to auto-optimize
• Split targeting by TikTok Category segments, using CBO to auto-optimize
• Targeting split by TikTok Interest keywords, using CBO to auto-optimize
• Primary Metric custom LAL audience based on historical data
• LAL audiences of an app activity metric, installs, trials, etc.
• LAL audiences of an ad metric, views, clicks, etc.
• Broad targeting split by time zone, using CBO to auto-optimize; additional ROAS boost from the top-performing demo

Scale

• Increase user base with Broad low CPM campaigns; identify top Geo
• Testing of US, CA, GB Geos separately with broad campaigns; CA and GB inventory can be limited on TikTok
• New accounts must run with an MAI objective until deeper funnel metrics are unlocked
• Due to limitations of SKAdNetwork API, there is a limit of 11 active campaigns and 1 ad group per campaign for iOS 14+

Behavior Audiences

• Testing of video and creator interaction behaviors

Interest Categories

• TikTok-specific interest-categories with large scale

Interest Keywords

• More specific targeting but with a limited scale

Time Zone Targeting

• Broad targeting with dayparting splits

Creative

• UGC creative key for TikTok