1、Apples privacy changes one year on:The good,the bad and the solutionsAs we are a few weeks from the first anniversary of Apples AppTrackingTransparency(ATT)launching,and with Google now beginning to share Androids mobile privacy plans,it seems like a good time to take stock of ATTs impact on perform

2、ance advertising.02ContentsThe Good 3The Bad 4Measurement 6Optimal Conversion Value and Conversion Window Setup 6Privacy Thresholds 7Deduplication 8Targeting 9Behavioral Targeting and Frequency 9 Capping Became More Difficult Testing on iOS is More Limited 10The Solutions 10Choosing the Optimal Conv

3、ersion Value and Conversion Window Setup 12Modeling and a Mindset Shift 16Privacy Threshold Analysis 17Using 1 Bit of your Conversion Value for Deduplication 17Test on Android and Flip to iOS 18Evolve Targeting 18Conclusion 20About Appsumer 2 1About InMobi 2203The GoodWhile there was a lot of panic

4、and concern about the future of mobile advertising prior to the launch of ATT,it seems safe to say that the sky hasnt fallen and mobile advertising continues to exist.Weve even seen a few silver linings with ATT:The majority of the ecosystem was readyDespite concerns that the ecosystem wasnt prepare

5、d,the numbers show that the vast majority of the industry from publishers to advertisers to demand and supply-side ad tech partners has stepped up and updated their systems to be compatible,particularly with Apples privacy-compliant measurement option SKAdNetwork(SKAN).Opt-in rates werent irrelevant

6、 Consumers are actually opting in to share their IDFA(Identifier for Advertisers,which is Apples device-level identifier)at higher rates than many expected.Opt-in rates vary depending on source but generally theyre over 20%not as low as the sub-10%some predicted;as a result modeling is possible.This

7、 bodes well for a more privacy-centric advertising world because it shows that some consumers are willing to share their data if approached the right way,with transparency and a clear value exchange.ATT knocked down some garden wallsATT has also created a more level playing field with the walled gar

8、dens,as they have often historically not been transparent while also grading their own homework.With SKAN in particular,advertisers are getting more transparency with raw data now available to them.Apple has released SKAN improvementsSKAN is slowly becoming a more viable solution not only did thousa

9、nds of developers and partners across the ecosystem rally to make the technical changes required to be SKAN compatible,but Apple also released enhancements to SKAdNetwork.As a result,some advertisers are increasingly successful,and SKAN is slowly becoming the attribution framework of choice for user

10、 acquisition and app install campaigns on iOS.All-in-all the predicted mobile advertising apocalypse hasnt been anywhere near as apocalyptic as many cynics predicted.04The BadWeve started well,however,we should take off our rose tinted glasses and look at what has been more challenging.The overarchi

11、ng challenge weve seen from conversations with performance marketers is a struggle to prove the value of advertising investments on iOS 14.5+devices.This causes them to struggle to spend on those devices.You just have to look at the results of this eMarketer survey from October last year to see this

12、:Have Mobile App Publishers Worldwide Made Changes to Their Android/iOS Strategy Since the Rollout of Apples App Tracking Transparency(ATT)?%of respondents,by audience size,Oct 2021No,everything remained the sameYes,we reduced our iOS spending but our Android spending remained the sameYes,weve shift

13、ed some of our spending from iOS to AndroidYes,weve shifted much of our spending from iOS to AndroidPublishers with less than 100K daily active users(DAU)Publishers with 100K or more daily active users(DAU)Source:AdColony and Fyber,“Mobile App Monetization Survey,”Nov 10,202165%12%7%16%40%20%9%31%05

14、Overall,60%of large mobile app publishers reduced their iOS spending alongside 35%of smaller publishers.Well break this challenge out into two categories:measurement and targeting.06MeasurementWhile there has been progress by Apple on SKAN,there are still some challenges that advertisers are struggl

15、ing to overcome.You may not think that these are challenges,but thats likely because your measurement solutions are falling back on fingerprinting,meaning you can ignore SKAN.This is like sticking a band-aid over a serious wound,because its not a case of“if”Apple will clamp down on fingerprinting,it

16、s“when.”When that happens,if you havent overcome these challenges,you will be left very exposed.Optimal Conversion Value and Conversion Window SetupA big challenge weve seen for performance advertisers at Appsumer has been identifying the optimal Conversion Value setup and Conversion Window length o

17、n SKAN.To recap,Conversion Values are a 6-bit code(six 0s or 1s)included in the SKAN postback that creates up to 64 combinations of post-install conversion events you can track.They are designed to highlight the value of an install by identifying conversion events that happen such as a purchase,ad i

18、mpression,trial signup,etc.There are many approaches to Conversion Values based on the monetization model of an app and what your mobile measurement partner(MMP)offers.You can see a summary of different approaches that we put together prior to ATT launching here.In addition,theres the Conversion Win

19、dow.This is the time period you as an advertiser decide to leave between the install and the SKAN postback being sent to you.You only get one postback sent for each install and the minimum length is 24 hours.However,if there is no change in the Conversion Value for 24 hours it is automatically retur

20、ned.Initially,Meta(the company formerly known as Facebook)dictated that advertisers on their platform needed to set this window to 24 hours.However,since then Meta has stepped back from influencing Conversion Value and Window setup,leaving advertisers free to test longer Conversion Windows.07The new

21、ness of SKAN,along with its greater freedom for advertisers to test their setup,has created a challenge.Longer Conversion Windows mean installs from the same day are firing back at random times,making it hard to cohort back to a specific install day of spend.Similarly,playing around with Conversion

22、Value setup creates an inconsistent dataset,making it hard to compare performance apples to apples over time.Additionally,if your app doesnt monetize quickly after an install or lacks the stickiness to keep users coming back every day after the initial install,then the Conversion Value on its own do

23、esnt give you much indication on the value of an install.Constantly changing Conversion Windows and Values creates an inconsistent dataset that makes it hard to compare performance over time.If you dont monetize on day one or have a sticky product that keeps users coming back daily in the days post-

24、install,its hard to predict the overall value of an install using only 24 hours of conversion data.This all makes it difficult to create a business case for iOS 14.5+ad investments using reliable data.Challenge SummaryAnother big challenge we see with SKAN is Apples mysterious privacy threshold.This

25、 is a mechanism that returns a“NULL”Conversion Value and source app details when a specific volume of installs isnt met for a campaign.Its mysterious,because Apple doesnt reveal what that volume of installs is per a campaign.Facebook revealed that on their platform you require 128 installs per campa

26、ign per day to avoid falling foul of the privacy threshold.However,we suspect that the privacy threshold is between 10-20 installs per day per SKAN Campaign ID.The reason its higher for Facebook is that they dont have a one-on-one relationship between their campaigns and SKAN campaigns as they use e

27、xtra Campaign IDs for their own learning.Privacy Thresholds08In your reporting,this means you now have duplicate installs in your MMP and your SKAN postback with no way of identifying them.Another challenge is around duplicate installs across different attribution sources.On iOS 14.5+after an instal

28、l happens,SKAdNetwork immediately starts tracking and attributing them.Simultaneously,an ATT prompt will be shown during the onboarding flow and the user may opt in to share their IDFA,meaning the install will also be attributed separately by the MMP.If a user opts in to share their IDFA via an ATT

29、prompt your MMP and SKAN will simultaneously and separately attribute that install.This duplicate install makes it hard to trust and unify your data across your MMP and SKAN.Challenge SummaryThe Deduplication ChallengeATTPromptSKAN TracksPost-Install ConversionsUserInstalls AppSKAN Registers Install

30、UserOpts-InMMP TracksPost-Install ConversionsSKAN ReportsInstallSKAN ReportsInstall09TargetingThe fact that iOS budgets did slump initially and switch to Android does also show that there have been issues with targeting that are now being overcome.Behavioral Targeting and Frequency Capping Became Mo

31、re DifficultBehavioral advertising has relied so heavily on tracking users across multiple mobile properties using the IDFA.Building a behavioral profile of a user using the IDFA including intent signals was a powerful tool.However,this got smashed apart,meaning algorithms(and the human brains that

32、build them)have had to relearn targeting in this new world.Similarly,frequency capping would historically be done at the user level using the IDFA.Controlling the frequency of ad impressions for an individual user was largely taken away with the IDFA.Behavioral targeting algorithms have had to adjus

33、t to life without the IDFA,meaning short-term performance took a hit for many behavioral targeting options across channels.Frequency capping became challenging as it could no longer be done directly at the user level using the IDFA.Challenge SummaryThe way to test and scale new targeting,creatives o

34、r channels previously has been to do small budget tests and scale(or not)depending on the results.The challenge now is that Apples privacy threshold means you wont understand results on iOS with small-scale testing at the campaign level due to the privacy threshold.For example,if you want to test a

35、new Facebook campaign youll need to invest enough budget to drive 128 installs per day and sustain this over multiple days.For many advertisers,this is no longer a small-scale test.This makes it hard to test campaigns for many channels on iOS devices.With the privacy threshold,small-scale testing is

36、 difficult when testing new targeting options and channels on iOS.Challenge SummaryTesting on iOS is More LimitedThe SolutionsOne interesting thing that we see in our data at Appsumer is that despite the initial challenges advertisers faced with measurement and targeting,larger advertisers have been

37、 able to overcome them and actually find opportunity on iOS.When we look at the overall spend that we track,the percentage of spend that was going to iOS pre-ATT in Q1 2021 for larger advertisers was 44%and in Q1 2022 it was 46%.1011However,smaller advertisers have clearly struggled more as the impa

38、ct of ATT set in.They saw their iOS share of spend drop from 41%pre-ATT in Q1 2021 to 35%post-ATT in Q1 2022.Clearly,larger more sophisticated advertisers have been able to overcome the challenges of a post-ATT world,whilst smaller advertisers have struggled to come to terms with the changes and mai

39、ntain iOS investment levels.So what we want to do is outline solutions weve worked on with these larger advertisers to overcome some of the challenges that weve highlighted.Q1 2021(Pre-ATT)Q1 2022(Post-ATT)iOS Share of Wallet by Monthly Advertising Spend SizeMonthly Advertising Spend SizeiOS Share o

40、f Wallet$1m+$250k+$250k0%10%20%30%40%50%12Speed of monetization:Ultimately,ask yourself how many days does it take to monetize the majority of users?This will dictate the type of Conversion Value setup that will be optimal and give you a sense of how long your Conversion Window setup should be.Mostl

41、y you want to identify,does a good majority of monetization happen in the first seven days?If so,by what day?Product stickiness:What percentage of users use the app every day in the first seven days after install?Your Conversion Value will postback if it doesnt change in 24 hours.If users are not co

42、ming back on day two or every day in the first seven days,you likely want to set the Conversion Window to one day or as many consecutive days as the majority of users return.Linearity of onboarding funnel:How predictable or linear is your onboarding funnel?In the first few days after install,do user

43、s go through a consistent set of steps that are predictable and give a good indication of their monetization likelihood?Once you have analyzed your analytics data to understand the answers to these questions,you can essentially follow this decision tree to define the optimum starting Conversion Valu

44、e and Window setup for your app.Choosing the Optimal Conversion Value and Conversion Window SetupAn important part of the Conversion Value and Conversion Window setup to emphasize is not to mess with it too often.That ensures you have consistent datasets to compare over time.When it then comes to ch

45、oosing the right Conversion Value and Conversion Window there are three key factors to consider using product analytics:13The Conversion Value and Window Decision TreeHow quickly do the majority of users monetize?Do you have a linear onboarding funnel?Do the majority of users return every day in tho

46、se 7 days?Conversion Value Model=RevenueConversion Window=7 daysIs the number of days it takes the majority of users to monetize equal to the number of consecutive days they return post install?7 DaysConversion Value Model=RevenueConversion Window=the number of consecutive days the majority of users

47、 return post-installDo you have a linear onboarding funnel?Conversion Value Model=Highest Event Conversion Window=the number of consecutive days the majority of users return post-installConversion Value Model=Conversion/User JourneyConversion Window=the number of consecutive days the majority of use

48、rs return post-installYesNoYesNoYesNoYesNoConversion Value Model=Highest Event Conversion Window=the number of consecutive days the majority of users return post-installConversion Value Model=Conversion/User JourneyConversion Window=the number of consecutive days the majority of users return post-in

49、stall14As a refresher,here are some examples of what those different Conversion Value models/schemas look like:Revenue Model ExampleEach increase in the Conversion Value is a revenue increment.In the below example each increment=$1A Conversion Value that returns a$1 purchase looks like this:A Conver

50、sion Value that returns a$63 purchase looks like this:0 0 0 0 0 1111111Highest Event Model ExampleEach increase in the Conversion Value is an event deeper into the conversion funnel.A Conversion Value that returns a registration might look like this:0 0 0 0 0 1A Conversion Value that returns an annu

51、al subscription might look like this:111111Note:61 conversion events can be tracked in between these two e.g.levels completed,invited friend,purchased starter pack15You could also create a hybrid custom model where you use a couple of bits/digits to track key conversion events and a couple of bits/d

52、igits to track revenue generated.Most advertisers use these base models/schemas as a starting point though.The key to getting more valuable data,from what weve seen,is aligning product and user acquisition(UA)teams.The more your product and UA teams can do to accelerate monetization,increase product

53、 stickiness and make your onboarding flow more linear,the better value data you will get.The challenge here is balancing this against user experience and negative impacts on overall monetization.Also,in most scenarios the further right you go on the decision tree youre capturing data to be used for

54、modeling revenue,which brings us to our next section.Conversion/User Journey Model ExampleAnnual Subscription0=Didnt Happen1=Did Happen0 0 0 0 0 0Purchased Starter PackLevel 10CompletedInvited FriendLevel 1CompletedRegistration16Linear redistribution:In this simplistic model you you start with two d

55、ata sources:1)actual revenue cohorted to install day via internal databases and product analytics,and 2)SKAdNetwork data with channel attribution and post-install data cohorted to the day when you received it or assumed install day.You can then assign users into clusters based on conversion events p

56、ost-install and then linearly assign revenue to each channel based on the number of installs in a cluster.Probabilistic redistribution:This works similarly to linear redistribution,which takes SKAdNetwork data,network-reported metrics and internal user level data.The key difference is adding determi

57、nistic attribution from MMPs to assign opt-in installs via the ATT framework and advanced clustering using algorithmic modeling,which enables the creation of many more clusters.Top-down incrementality:This is a more advanced approach that takes three major inputs:1)Aggregated cost data by channel fr

58、om tools like Appsumer,2)App event data and 3)Real revenue data from product analytics and internal revenue sources.Econometric models are then applied to attribute revenue incrementally by channel and at more granular levels.Modeling and a Mindset ShiftUnless you have a rapid speed of monetization

59、and a very sticky product in the days post-install,most of your revenue/lifetime value(LTV)data will need to be modeled.Essentially,this involves taking early monetization signals/conversion events and revenue data from internal systems to predict revenue at the campaign level.There are really three

60、 approaches to modeling,which we cover in more detail in this post:17As mentioned,to get a detailed rundown on these approaches and vendor options check out this post.However,youll need to adjust to the fact that you cant deterministically measure everything to the nth degree.As user acquisition exp

61、erts,we have a mindset of looking for the most accurate degree of measurement with user identifiers providing deterministic data and not trusting anything else.This world is gone.We should stop trying to cling onto it for iOS.In this new world we need to get comfortable with modeling and extrapolati

62、ng data to understand campaign performance.The alternative is not having any data to justify iOS spend and ultimately the existence of team members.Your focus now needs to be on gathering the richest data possible through SKAN,and building the models and infrastructure to get data as accurate as pos

63、sible.It wont be perfect,but its better than nothing when building business cases and optimizing campaigns.The SKAN data being returned on certain campaigns is limited by the privacy threshold.As such,measurement infrastructure and analytics needs to be expanded.This data loss makes it near impossib

64、le to measure performance as accurately as before.Weve set up reporting for a number of customers now to help them understand the campaigns where the highest percentage of SKAN postbacks are being returned with“NULL”values.This view enables you to identify where you need to consolidate or expand cam

65、paigns to get richer data returned.This can then actually inform optimization and justify the business case for investments.Privacy Threshold AnalysisTo overcome the duplicate install problem,an increasingly popular approach,championed by Appsflyer,is to use one of the 6-bits in the Conversion Value

66、 to identify whether or not an install(i.e.the person that just installed the app from an ad)has opted in via the ATT prompt.Then on the back end you can remove those who opted-in via ATT from your SKAN data to overcome the issue of duplicate installs.Using 1 bit of your Conversion Value for dedupli

67、cation18With privacy thresholds limiting the ability to test new channels and targeting on iOS at a small scale,this testing is now having to switch to Android.The focus now is on running small-scale tests on Android and then running winning approaches on iOS,evolving and optimizing from there when

68、you can reach the scale to avoid the privacy threshold.Obviously,you will see behavioral differences from Android to iOS,and you will need to adjust campaigns on-the-fly for this.However,this approach avoids burning budgets with large scale testing flops on iOS.At the same time,its worth keeping tra

69、ck of what impact Googles upcoming(in 2 years)privacy changes might have on this approach over the long term.Test on Android and Flip to iOSWere starting to see algorithmic targeting evolve across channels to adapt to this new world,with performance recovering for many advertisers on channels that w

70、ere hit hard.Advertisers are also starting to explore new channels on iOS earlier in their growth,as they learn how they need to adjust targeting in this new world.For many smaller advertisers,the scale challenge to overcome the privacy thresholds impact on Facebook has hit perceived performance on

71、one of their core channels.This means that they need to consider diversification into new channels where the privacy threshold isnt an issue and they can reliably measure performance.Although IDFA is now more limited,there are still other valuable signals that can be used to inform targeting and bid

72、ding,including app metadata like app category and version,content consumed within apps,device information,time spent in app,etc.Contextual targeting has long been popular,and in-app advertising provides more contextual signals than would be available for,say,a linear television buy.Evolve Targeting1

73、9For example,on InMobis DSP they look at things like:If they see a user is in a certain app,they may be interested in a similar app.If they see a users battery level is low,that user may be less likely to install a new app.The local time and keyboard language preferences of a user reveal the rough l

74、ocation (country/region)of the user.When you take available signals together,its very powerful as you can start to paint a picture of what a users interests or preferences may be,and then target them with relevant ads,all in a privacy-safe way.The key point here is that targeting is evolving.Context

75、ual targeting will become more evolved and powerful and scaling advertisers are needing to diversify earlier from core channels than they perhaps have previously.Its also important to work more closely with partners to understand their targeting approaches and how you need to evolve targeting with i

76、ndividual channels to optimize performance in this new privacy era.20ConclusionThe impact of ATT is now clearer.While there will be smaller adjustments in the coming months and years,its now easier to define and plan your new approach for the privacy era in mobile performance advertising.If you deci

77、ded to jump off iOS for a while,its time to jump back on the wagon just be sure to evolve your measurement infrastructure and targeting approaches to fit the new world.Areas to focus on include:Define the optimal Conversion Value and Window setup that you will use for the long term.Consider reservin

78、g 1-bit for deduplication.Build reporting to give you transparency on the impact of the privacy threshold.Define your approach to modeling and get comfortable with measurement being more estimated than definitive.Run all small-scale testing on Android(for now)and flip learnings over to iOS and evolv

79、e with larger scale tests there.Diversify earlier and work closely with partners to understand their targeting adjustments when the IDFA isnt present and evolve targeting approaches at an individual channel-level.About AppsumerAppsumer offers an off-the-shelf BI solution for performance marketers at

80、 consumer mobile apps with:Easy-to-use dashboards and reports which unify cost,attribution and revenue data across advertising channels and operating systems,so you can get a complete view of performance down to a granular level.A unified view of SKAdNetwork data alongside other attribution sources

81、so you can get an apples-to-apples performance comparison across newer versions of iOS and other OSs to continue proving the business case of iOS investments.If you already have an existing BI setup for your performance marketers,we offer a cost data pipeline from 100+channels to save time on mainta

82、ining API connectors and deliver granular data across all channels you can trust every day.App developers like Picsart and Miniclip have already seen significant performance improvements and time savings from using Appsumer.Book a demo today to get the performance marketing insights you deserve.21In

83、Mobi is a world-leading provider of marketing and monetization technologies reaching billions of consumers around the globe.With deep expertise and unique reach in mobile,it is the trusted and transparent technology partner for marketers,content creators and businesses of all kinds.InMobis mission i

84、s to power its customers growth by helping them engage their audiences and build meaningful connections.Its affiliated businesses Glance,the worlds largest lock screen-based content discovery platform and video platform Roposo help InMobi create new content and commerce experiences in a world of connected devices.InMobi maintains dual headquarters in San Francisco and Bangalore with operations in New York,Chicago,Kansas City,Delhi,Mumbai,Beijing,Shanghai,Jakarta,Singapore,Manila,Kuala Lumpur,Sydney,Melbourne,Seoul,Tokyo,London and Dubai.To learn more,visit 22About InMobi