Getting Started with MOBILE ATTRIBUTION & MARKETING ANALYTICS
2016 Edition Chapter What’s in the Guide Page
1 INTRODUCTION...... 3
2 THE BASICS OF APP MARKETING...... 6 2.1 Mapping the Mobile Marketing Ecosystem...... 6 2.2 common Pricing Models...... 10 2.3 the Importance of Non-Organic Installs...... 11 2.4 breaking Down Mobile Analytics...... 12
3 MOBILE ANALYTICS UNDER THE HOOD...... 16
3.1 windows of Opportunity...... 17 3.2 attribution Methods — How Does it Work?...... 19 3.3 the Deeplinking Dive...... 26 3.4 integrated Partner Ecosystem...... 27 3.5 unbiased vs. Biased Attribution Providers...... 31 3.6 Mobile Retargeting Attribution...... 35 3.7 tv Attribution...... 37
HOW TO SQUEEZE THE DATA LEMON WITH 4 ATTRIBUTION & MARKETING ANALYTICS...... 38
4.1 in-App Events to Get Started With...... 40 4.2 retention - Keep Users Coming Back for More...... 44 4.3 cohort Analysis...... 46 4.4 one Formula to Rule Them All...... 48 4.5 reporting - Dashboard, Pull & Push APIs...... 49 4.6 Mobile Ad Fraud - No Longer the Elephant in the Room...... 50
5 THE 10 COMMANDMENTS OF MOBILE ADVERTISING ANALYTICS...... 53
6 CONCLUSION...... 54 Getting Started with Mobile Attribution & Marketing Analytics
CHAPTER 1 INTRODUCTION
Before the dawn of the mobile era, digital (web) marketers were able to accurately measure the impact of their campaigns and make smart decisions about their ad spend. Thanks to the cookie, advertisers were able to anonymously identify users in the browser, and that completely redefined traditional spray-and-pray and contextual online advertising.
But then the mobile revolution turned things upside down and the cookie began to crumble because it wasn’t supported on applications and was deactivated by default on Apple’s Safari browser. That left the mobile cookie with extremely limited reach.
IDFA
IDFA IDFA Advertising ID IDFA
Advertising ID IDFA
www.appsflyer.com 3 Getting Started with Mobile Attribution & Marketing Analytics
The mobile world is deeply fragmented.
The problem with mobile is that it is deeply fragmented: There are different operating systems (iOS, Android, Windows Phone), different environments (in-app, mobile web) and different identifiers (Apple’s IDFA, Google Advertising ID, Google Play Referrer, etc).
Can targeted, personalized advertising work and thrive in a post-cookie and mobile-first world? The answer is a resounding yes. This guide will show you how.
www.appsflyer.com 4 “ THE TREND HAS BEEN MOBILE WAS WINNING. IT' S NOW WON.”
ERIC SCHMIDT Getting Started with Mobile Attribution & Marketing Analytics
CHAPTER 2 THE BASICS OF APP MARKETING
2.1 Mapping the Mobile Marketing Ecosystem
The mobile ecosystem can be divided into the following main groups:
BUY SIDE SELL SIDE
Brands App Agencies Ad Publishers Developers Networks
DSPs Ad SSPs Exchanges
ANALYTICS
Independent Marketing App In-App Attribution Analytics Store
OTHER
Mobile Marketing Anti-Fraud Automation Specialists
www.appsflyer.com 6 Getting Started with Mobile Attribution & Marketing Analytics
Buy Side
Brands. Companies that usually have an established presence on the web and/or offline. They are often classified by verticals such as retail (like Macy's or Amazon), travel (American Airlines, Hotels.com), entertainment (HBO, Netflix) and CPG (Coca Cola, Kraft).
App Developers. App owners whose only business is mobile apps (like Supercell, Tinder, Clean Master).
Agencies. A company that creates and manages ad campaigns for multiple advertisers (usually brands).
DSPs. A Demand Side Platform is an advanced software that connects multiple ad exchanges together, automatically making complex media buying decisions to determine how much to pay for a single ad impression — mostly through a process called real time bidding (RTB).
www.appsflyer.com 7 Getting Started with Mobile Attribution & Marketing Analytics
Sell Side
Publishers. A publisher is a mobile website or app that sells ad inventory on its property.
SSPs. A Supply Side Platform helps publishers automatically and efficiently manage the selling of their inventory to multiple buyers in order to maximize their inventory yield.
Connecting the buy side to the sell side
Ad Networks. A company that matches supply from mobile websites and apps to the demand of advertisers. A key industry player.
Ad Exchanges. An automated marketplace that connects advertisers and publishers to buy & sell mobile ads, mostly in real time.
Analytics
Independent Attribution Analytics. Unbiased providers serving as de-facto ecosystem regulators that are able to accurately attribute credit for an install or a re-engagement to a single source (or multiple sources in the case of multi-touch attribution). By definition, truly Independent Attribution Providers are impartial and do not engage in the buying or selling of mobile ads. This is why they are trusted both by players on the buy and sell side to measure and report on campaign performance, and settle reporting discrepancies.
Marketing Analytics. By connecting attribution to in-app activity, marketers can understand which source (network, campaign, publisher and even creative variation) delivered real value (engagement, revenue, lifetime value, return on investment). Attribution firms commonly include marketing analytics in their offering.
www.appsflyer.com 8 Getting Started with Mobile Attribution & Marketing Analytics
App Store Analytics. A category divided into (1) the app stores themselves that offer basic analytics on installs, and (2) specialized companies that provide more advanced analytical tools & data slicing.
In-App Analytics. Companies that specialize in measuring and analyzing app usage patterns. Such insights can be used to improve user experience, retention rate, and user lifetime value.
Other
Mobile Marketing Automation. Providers that develop software that simplifies communication with customers and prospects. In mobile, this involves an ever-growing set of channels such as push notifications, in-app messaging, wearable device notifications, and SMS (or MMS).
Anti-Fraud Specialists. Firms that develop advanced data-driven mechanisms to combat the threat of ad fraud, as it applies to mobile (particularly install fraud and in-app event fraud).
www.appsflyer.com 9 Getting Started with Mobile Attribution & Marketing Analytics
2.2 Common Pricing Models HIGH PERFORMANCE
Cost Per Cost Per Action Install CPA CPI
Payment Terms: Pre-determined Payment Terms: Pre-determined price price paid for every in-app action paid every time a user installs the defined by advertiser (revenue or application on his or her device. engagement related). Pros: Performance model and reduced Pros: Pure performance model risk; organic uplift following increase in that’s connected to users’ real value CPI-driven campaign installs. actions; adopted by the savviest data-driven advertisers. Cons: Risk of non-transparent networks driving a high volume of low quality or Cons: Can be more costly; may hurt incentivized traffic to drive installs; in reach as networks are more cautious hyper-competitive app economy the buying media as they assume install KPI is losing ground to engagement heightened risk. and retention-driven KPIs.
LOW HIGH ADOPTION Cost Per Cost Per ADOPTION Mille Click CPM CPC
Payment Terms: Pre-determined price Payment Terms: Pre-determined price for every 1,000 impressions (cost per paid every time a user clicks on an ad mille – mille being the Latin term for one thousand). Pros: Easier to analyze user engagement through ad creative A/B testing; primary Pros: Maximal brand awareness, reach, model used by tech giants Facebook, lower cost. Google and Twitter.
Cons: Non-performance model; greater Cons: "Fat fingers" phenomenon means chance of fraud and non-transparent you risk paying for unintended clicks and networks sending low quality damaging your brand name with awful impressions. user experiences; higher cost than CPI if you don’t have the resources to optimize click-to-conversion path; lack of robust analysis tools; vulnerable to fraud.
LOW PERFORMANCE
www.appsflyer.com 10 Getting Started with Mobile Attribution & Marketing Analytics
2.3 The Importance of Non-Organic Installs in Today’s App Economy
With about 1.5 million apps on both iTunes and Google Play, the chances of making it on organic installs from app store discovery alone are about as slim as winning the lottery. That’s why apps need to invest in marketing to drive a volume of installs, also called non-organic installs.
Non-organic installs can be divided between:
• Paid user acquisition: Running campaigns with ad networks, agencies etc.
• Owned channel promotion: Running campaigns on a brand’s owned properties such as email, SMS, and QR codes - all part of an advertiser’s overall mobile marketing mix
Other than increasing your user base, non-organic installs will also increase the volume of organic installs thanks to improved App Store Optimization (ASO). ASO is the app store equivalent of SEO as it determines the order of apps presented to users as they search and explore a store.
App store algorithms use a variety of parameters to determine rankings. These include title, keywords, visuals, ratings, reviews, editorial review, uninstalls and, as mentioned above, the actual number of installs an app has.
Ultimately, successful apps grow by investing in marketing and ASO to drive the highest number of both organic and non-organic installs.
www.appsflyer.com 11 2.4 Breaking Down Mobile Analytics
The big data era in marketing is here.
We’re surrounded by megabytes, gigabytes, terabytes, petabytes and exabytes of data.
But even all the data in the world isn’t worth much, especially to marketers, if it can’t be aggregated into meaningful patterns that lead to actionable insights. Getting Started with Mobile Attribution & Marketing Analytics
Successful data-driven marketing – on web or mobile – lives or dies on the ability to gather insights and act upon them. And it all comes alive in the analytics dashboard – which has become the digital marketer’s best friend.
When we zoom-in on mobile app analytics, we’re actually looking at several categories. Of course, these include attribution and marketing analytics, but this has an entire chapter of its own further ahead so we’ll briefly outline the others first:
• App-Store Analytics • In-App Analytics • Predictive Analytics
App Store Analytics (App Stores – i.e. iTunes, Google Play)
Provides a basic understanding of an app’s success, but In a Nutshell that’s about it.
KPIs Downloads, rankings, device, geography, revenue
A starting point for beginners that provides an What It’s Good For overview of an app’s success. It also delivers the highest accuracy since it’s from the source itself.
www.appsflyer.com 13 Getting Started with Mobile Attribution & Marketing Analytics
App Store Analytics (Providers – i.e. App Annie, Appfigures)
Offers advanced analytics that app stores do not In a Nutshell offer – mainly comparing your app to the competition and supporting multiple slicers and filters.
Revenue, downloads, updates, ranks, reviews, rank per KPIs keyword
This category of analytics provides items such as competitive & vertical analyses, in-depth reports (installs, What It’s Good For rankings, geographic trends, alerts), and aggregated data per store, which is a must if you have several apps or different app versions.
In-App Analytics
An extremely important source of data that measures In a Nutshell what users are actually doing inside the app and how they are engaging (or not engaging) with it.
Screens viewed/exited, buttons clicked, time spent per KPIs page, levels cleared, purchases made, slicing data by device type & user demographics
These analytics help you optimize the user experience and improve your retention rate by knowing exactly how your users interact with your app. It can help maximize What It’s Good For your user’s lifetime value and isolate the reasons behind success or failure by mapping your most common conversion and exit paths.
www.appsflyer.com 14 Getting Started with Mobile Attribution & Marketing Analytics
Predictive Analytics
The up-and-comer on the mobile big data block, these analytics apply big data and machine learning to predict how a single user will react to a specific In a Nutshell offer, how much to bid for that user, and which pros- pect to target based on his or her resemblance to your top customers.
KPIs Expected LTV, expected revenue, expected CTR
These analytics are good for improving retention and lifetime value, conducting look alike modeling for user What It’s Good For acquisition, increasing ROI, and driving heightened efficiency via increased bidding accuracy.
www.appsflyer.com 15 Getting Started with Mobile Attribution & Marketing Analytics
CHAPTER 3 MOBILE ATTRIBUTION ANALYTICS UNDER THE HOOD
Today’s mobile environment runs on a last-click attribution model, yet each advertiser runs with multiple ad networks (some of the big ones have dozens). In such a reality, there’s an undeniable need to use a trusted source – by both networks and advertisers – to rule which network gets credit for reaching a defined goal.
The only way to accurately make it stick is by having a bird’s eye view of the click-to-install path across all integrated networks and informing them in real time, via what is known as a postback, that they have been credited.
Without this view, each network could bill you for a click that led to an install, regardless of whether it was the last click or not. What this means is that different networks can claim credit for the same install.
Proper attribution saves you attention money as it prevents you from marketers being double or triple charged.
www.appsflyer.com 16 Getting Started with Mobile Attribution & Marketing Analytics
3.1 Windows of Opportunity
The business side of install attribution is based on pre-determined timeframes, aka lookback windows. That means a period of time in which a user's action that precedes an install – whether ad click or view – counts as having an impact on the decision to download an app. There are several types of windows:
CLICK-BASED ATTRIBUTION
1) 7-day standard: If a user clicks on an ad served by network x, and then installs the app within 7 days of that click, it would get the credit for that install - assuming there wasn't another click that followed (under the last click rule).
2) 24-hour fingerprinting attribution:If there is no device ID or referrer, an install can be attributed based on fingerprinting. Its accuracy level is highly accurate in the short term and therefore it’s only open for 24 hours (see page 24 for more).
3) Facebook, Google & Twitter: The three tech giants have their own rules in place: For Facebook and Google, attribution windows are fixed (non-configurable) at 28 and 30 days, respectively. Twitter enables advertisers to choose between 1/7/14/30/90 days. Because they mostly work on a CPC basis, these media sources will charge for any click that occurred within their window regardless if it was the last click or not. Regardless, they are informed by the attribution provider that there was an install and use that data for their own optimization purposes.
4) Configurable window: Gives networks and advertisers greater flexibility: networks want the longest possible windows, while advertisers want to have the ability to normalize their data and compare apples to apples when they run their analyses. For example, since effective CPI is very much dependent on the duration of a window, proper comparison is not really possible when comparing networks with different windows.
A configurable window can also be of use when running campaigns that are limited in time. For example, a taxi app’s 24-hour special campaign promising installers the first ride for free — in such a case it would be of great value to get reports only on installs that occurred within a 1-day window.
www.appsflyer.com 17 Getting Started with Mobile Attribution & Marketing Analytics
VIEW-THROUGH ATTRIBUTION
Some agreements include giving credit to an install based on an ad view. In such a case the window is short – usually up to 24 hours. However, since the click is mightier than the view, it always wins (assuming it occurred within its own window of course). Facebook and Twitter will also charge advertisers based on a 24-hour view-through attribution window.
Let's look at the following example to better illustrate the rules of attribution windows:
Click Click View network network network B A A install
Network A: 7-day click window Network A: 1-day view window
Network B: 15-day click window
WHICH NETWORK WINS? The answer: Network B
?
1) Network B had the last click within a lookback window. 2) Click wins over view even if the latter is within its window.
www.appsflyer.com 18 Getting Started with Mobile Attribution & Marketing Analytics
3.2 Attribution Methods – How Does it Work?
Despite the fragmentation of the mobile space, cookie-less mobile measurement has come of age. It relies on advanced technology that uses a variety of highly accurate identification methods at scale.
When cookies are not available, there are 3 main methods to attention attribute credit using different identifiers: marketers Google Play Referrer ID matching (Google’s Advertising ID or Apple’s IDFA) Fingerprinting
www.appsflyer.com 19 Getting Started with Mobile Attribution & Marketing Analytics
1) Google Play Referrer:
A standard and highly reliable and accurate method to attribute conversions through Google Play (but not Android out of store). It enables the attribution provider to send tracking parameters to the store, which then passes them back to the source when the app is downloaded.
Click Referrer + start (referral unique identifier here params) sent to store
Ad Attribution provider Google Play Generates referrer from referral params + unique identifier Postback
OPEN
First app open App installed Provider SDK reports that the app was Google adds referrer + opened and matches it with the identifier to app referrer + identifier in provider's database
The tracking provider will most likely use the referrer method as it only depends on itself to create this match - it simply uses publicly available data from the referral source.
www.appsflyer.com 20 Getting Started with Mobile Attribution & Marketing Analytics
2a) ID Matching: Google Advertising ID (GAID)
Another rock solid identifier in terms of accuracy. This device ID is used to measure Android installs (for Google Play and Android out of store).
The problem is that advertising IDs do not exist in the mobile web and even in the app space they are not always available (requires a user’s consent and the ad network to support and pass them – which is why the referrer method is vital).
start Click here (ad ID)
Ad Attribution provider Google Play/ Android out of store Records the device's GAID for the specific app being advertised Postback
OPEN
First app open App installed Provider SDK reports that the app was opened and matches it with the GAID that was recorded via the click
www.appsflyer.com 21 Getting Started with Mobile Attribution & Marketing Analytics
2b) ID Matching: Apple's IDFA
Apple’s highly accurate and privacy-sensitive device identifier that replaced UDID. The attribution method works in the same way as with Google Advertising ID.
Click start (IDFA) here
Ad Attribution provider App Store Records IDFA for the specific app being advertised Postback
OPEN
First app open App installed Provider SDK reports that the app was opened and matches it with the IDFA for the specific app it has stored in its database
www.appsflyer.com 22 Getting Started with Mobile Attribution & Marketing Analytics
3) Fingerprinting:
Fingerprinting, aka device recognition, is an identification method that uses publicly available parameters (i.e. device name, device type, OS version, platform, IP address, carrier, to name just a few), to form a digital fingerprint ID that statistically matches specific device attributes.
Click start (device here params) Network) (Integrated
Ad Attribution provider Play/App Store Generates fingerprint ID based on the specific device parameters Postback
OPEN
First app open App installed Provider SDK reports that the app was opened and matches it with the fingerprint ID in the provider's database
www.appsflyer.com 23 Getting Started with Mobile Attribution & Marketing Analytics
4 things you need to know about fingerprinting As it’s less straightforward, some information on this method:
1) Fall-back method
Fingerprinting is a probabilistic model and therefore is not 100% accurate. That’s why it’s only used when deterministic identifiers like Google Play Referrer and device identifiers such as IDFA or Google Advertising ID are not available for attribution (e.g. when the click comes from the mobile web, or when the data is not passed by an ad network).
2) Highly accurate in short term
The longer the time gap between a click and an install, the greater the chance that the user would change a setting in his/her device, thereby creating a new fingerprint (this is especially true with IPs as mobile users constantly change their location). That’s why the attribution window is short, usually 24 hours. In app install campaigns, the vast majority of clicks, installs and first app opens occur within 1-2 hours, in which case the fingerprint is extremely accurate.
3) Used primarily in iOS
As the referrer method is not available in iOS (and almost always available in Android via Google Play), fingerprinting is used more often on Apple devices. With only one deterministic option to work with, there’s a greater likelihood to use the fallback method.
4) Anonymous
A mobile fingerprint is privacy compliant and does not include any personally identifiable information.
www.appsflyer.com 24 Getting Started with Mobile Attribution & Marketing Analytics
Multi-identifier logic
Powerful engines enable attribution companies to choose between available identifiers in real time, and to do so billions of times a day based on the following logic:
iOS Android (Google Play) Android out of store
Primary IDFA Referrer GAID GAID Option OR
Fall-back Fingerprinting
www.appsflyer.com 25 Getting Started with Mobile Attribution & Marketing Analytics
3.3 The Deeplinking Dive
In the beginning, the app world was a world of its own. A walled garden. Links and data did not pass through apps or between the mobile web and apps. And that made ad tracking and optimization very difficult, not to mention the frustrating user experience it generated when a user clicked on one thing only to end up at another (or nowhere at all). And that’s how deeplinking was born. Since then innovation kicked in, offering robust solutions to connect the islands together through the following elements:
1) Any Environment, One Smart Link Different platforms, environments app stores, channels (including email, push message, SMS, etc.) can all be configured in a single link which avoids set up nightmares and many broken links
User clicks on promotion
2) Deeplinking User has app installed, right screen opens
Install Right screen If not... or opens on 1st launch
3) Real Time* Deferred Deeplinking
* Whereas deferred deeplinking has become an industry standard, the ability to open a specific screen with the right content in real time (or near real time) is a key aspect that separates the different providers from one another.
www.appsflyer.com 26 Getting Started with Mobile Attribution & Marketing Analytics
3.4 Integrated Partner Ecosystem
When exploring attribution providers, you’ll probably run into each one flexing its muscles by boasting a large number of integrated partners. What is this and why is it important? There are three main reasons:
Reason #1: Universal SDK. If advertisers want to work with an ad network the attribution provider is integrated with, they don’t have to add its SDK. It all goes through the tracking company’s SDK that sends a postback filled with data back to the network.
This solves a major headache for marketers (and their developers) who have a hard time involving their IT to integrate every single network they want to work with; not to mention the impact on app performance when running with a universal SDK.
The more integrations an attribution provider has, the easier it becomes and the more options you have to choose from.
www.appsflyer.com 27 Reason #2: Optimization. Attribution companies can often work with any media source and get data they need to attribute an install through the click. But reporting the conversion back to the non-integrated network, if indeed it happened, won’t be supported.
There’s no postback. Without it, the network won’t be able to properly optimize the campaign and will miss out on key insights.
Reason #3: Stamp of Approval. Running campaigns with integrated partners has become a standard. Advertisers will often refuse to run with networks that are not integrated with a trusted tracking system simply because they place their faith in the latter as the impartial “judge” who makes the rulings - which are then used as the basis for billing.
www.appsflyer.com 28 Getting Started with Mobile Attribution & Marketing Analytics
Official Partnerships with the Tech Giants As dominant forces in the mobile advertising space, Facebook, Facebook-owned Instagram, Google and Twitter have a market share of over 50%. They've built robust solutions and more importantly a powerful position that enables them to manage the data.
In most cases, an advertiser will not use a network’s SDK and measurement, but instead require the network to work with a trusted mobile attribution company. As such, the network is dependent on the measurement provider to send the data.
But when it comes to the big players that measure their own campaigns and are trusted by advertisers to do so, the data travels the opposite way – from the network to the attribution provider.
www.appsflyer.com 29 Getting Started with Mobile Attribution & Marketing Analytics
In order to control the passage of this data, Facebook, Google (with AdWords and DoubleClick), and Twitter have selected a very limited number of measurement partners. All of these partners have gone through an extensive due diligence process, and adhere to strict rules (For example, Facebook decided to remove Tune, formerly known as Mobile App Tracking, from its Mobile Measurement Partner program because they did not comply with Facebook’s data safety rules).
The bottom line is that there are only two ways to measure campaigns on Facebook, Google and Twitter: to embed their own SDK or work with their official measurement partners.
Some claim that Facebook campaigns can be measured with deep links but that’s actually a false claim since there’s no timestamp on a deeplink, making last click attribution impossible.
www.appsflyer.com 30 Getting Started with Mobile Attribution & Marketing Analytics
3.5 Unbiased vs. Biased Attribution Providers
The mobile attribution analytics ecosystem can be divided among the following players:
• End-to-end (or multi) solution firms. These companies buy media, measure campaigns, and optimize them. They are therefore biased attribution providers. As media companies, their core business from which they earn most of their revenue involves buying and selling media. Attribution is therefore a secondary part of the business.
• Attribution-only companies. Providers whose only business is attribution and are therefore in a position to assume the role of ecosystem regulators, raising the flag of neutrality, transparency and reliability. As such, they are unbiased.
Biased solutions indeed bring value as they reduce overhead and simplify workflows. Busy as they are, this appeals to many digital marketers.
However, what they’ll probably find less appealing are the inherent conflicts of interest such a marriage may bring. Let’s explore some of the major ones you should be aware about.
Media Company
www.appsflyer.com 31 Getting Started with Mobile Attribution & Marketing Analytics
Data-sharing.
Attribution analytics providers are big data companies. They collect petabytes upon petabytes of data about ad campaigns, installs and in-app events. If a media company owns an attribution solution, it has a trove of data at its disposal that it could use to optimize and monetize other campaigns.
An unbiased provider would never ever use your data or sell it to third-party companies.
Preferred Networks.
A biased attribution provider may try to sway the advertiser towards other networks on its list, whether owned or preferred. Integrating with the network requested by the advertiser would either drive traffic to a competitor – if the company actually owns a network, or may lead to a drop in revenues – if it runs with a new partner with whom it has less favorable terms.
www.appsflyer.com 32 Getting Started with Mobile Attribution & Marketing Analytics
Discrepancies.
When you ask a biased provider to investigate a discrepancy in attribution data, it may decide not to invest in such a probe but rather “rule” in a way that would best serve its interests (for example, if it owns a network it would make a hefty sum if attributing most of the credit to its own network).
Little or no support.
Advertisers seeking support from a biased attribution provider may find that if they are not a major source of revenue, support of their account would suffer. It simply wouldn’t be economically viable to invest in such accounts if another part of the business generates more revenue.
www.appsflyer.com 33 Getting Started with Mobile Attribution & Marketing Analytics
But wait… There's a third type of attribution company: 'traditional' desktop providers These companies argue that they can offer cross-channel coverage, but in fact they don’t go full circle. Coming from the cookie-based web, mobile web is familiar territory.
However, since mobile is almost entirely cookie-less, they partner with cross-device measurement companies that offer probabilistic identification to connect the dots. Using log-in data can also work, but its scale is extremely limited. Either way, a complete solution it is not.
Developing mobile attribution requires a massive investment that can take years to come to fruition, even for large firms like some of the desktop companies. Here’s a partial list:
• Thousands of integrations with media partners
• Robust algorithms that can work with multiple identifiers
• Ability to support the tracking of billions of in-app events and connecting them to the attributed sources
• Dealing with the fragmented mobile landscape
• Developing a deep-linking solution to connect different mobile environments (a non-issue on the web)
• Adapting to a mobile funnel that includes an app install stage
www.appsflyer.com 34 Getting Started with Mobile Attribution & Marketing Analytics
3.6 Mobile Retargeting The use of retargeting in mobile has all but become mainstream now. In fact, the percent of marketers retargeting on mobile jumped from 54% to 82% in 2015, according to AdRoll.
Advances in deep-linking technology certainly helped push retargeting forward on mobile because it allows a user to be directed to a specific page that’s personalized to fit his/her interests – a pair of Adidas soccer shoes, or an incentive to pass a specific level of a gaming app that a user abandoned. After all, knowing that your users were interested in something specific and then directing them to the home screen of your app would be a complete waste.
Mobile Retargeting Attribution Mobile retargeting, aka re-engagement attribution, occurs when an existing user that has the app installed engages with the retargeting campaign and opens the app. In this case, matching is done by deeplinking (which has an attribution provider's parameters) or a device ID if a deeplink isn’t available.
www.appsflyer.com 35 Getting Started with Mobile Attribution & Marketing Analytics
Re-engagement Attribution Window The number of days in which an event can be attributed to a retargeting campaign. It starts when the actual retargeting attribution occurs (after click and app open), and finishes at the end of the re-engagement attribution window. The default re-engagement attribution window is 30 days, but this can be configurable.
Retargeting attribution can get tricky when trying to understand how it interacts with Install attribution. Let’s explore the following example:
UA Re-engagement +3 days +10 days +32 days campaign install campaign $10 Purchase $12 Purchase $21 Purchase click click+app open
30-day re-engagement attribution window
Lifetime install attribution window
The re-engagement campaign will be credited for re-engaging the user after he/she clicks and opens the app. When the user goes on to purchase items, the LTV of both the UA and re-engagement campaign increases as follows:
Event A (+3 days) Event B (+10 days) Event C (+32 days) Campaign Attribution Attribution Attribution LTV
UA True True True $43
Re-engagement True True False $22
If you are wondering about duplication, you are right. That’s why any events that occur within the re-engagement window can be marked as ‘Primary’ to enable simple deduplication.
www.appsflyer.com 36 Getting Started with Mobile Attribution & Marketing Analytics
3.7 TV Attribution TV campaign budgets for apps (not to mention standard brand advertising) are on the rise. In fact, both Clash of Clans and Game of War: Fire Age paid a cool $4.5 million to win national attention in the US during Super Bowl 2015.
The newest form of install attribution is used to measure organic installs by examining the impact of TV ads on immediate actions in the app stores. Basically this is done by matching data on the airing time of each ad with an app install that occurs within a time frame (in minutes) determined by the advertiser. Post install analytics such as retention, LTV and Cohort reports are then calculated to measure effect.
Television is the next attribution frontier. A marketer can already use a deferred deeplink to the app store, and then launch the app with a TV-specific promotion in the first screen (i.e. Install the app and get 1000 free coins, or Install the app and get 10% off your first purchase). And we haven’t even mentioned smart TVs.
www.appsflyer.com 37 Getting Started with Mobile Attribution & Marketing Analytics
CHAPTER 4 HOW TO SQUEEZE THE DATA LEMON WITH ATTRIBUTION & MARKETING ANALYTICS
Now that we understand the ins and outs of how attribution works, let’s turn our attention to the business value of attribution, best practices and pitfalls to watch out for. Obviously, this topic has to it many angles but in this introductory guide we’ll focus on the basics.
App marketing is an evolution in progress but the direction is crystal clear: performance. What started with CPM and CPC is currently dominated by CPI, and gradually moving to CPA (see page 10 for more details).
There may be an ad network that’s delivering tons of new users, but when you take a closer look, you realize that they’re low quality. Your user base may have grown, but many of these users may not have had any active app sessions nor completed any in-app actions.
A properly attributed campaign that’s tied to in-app events allows marketers to better optimize and make informed decisions on which ad networks to work with.
www.appsflyer.com 38 Getting Started with Mobile Attribution & Marketing Analytics
Volume of installs is obviously very important as it pushes an app up the ranking in the app stores. But it also serves as the baseline from which you’ll find the best users. If it’s a matter of percentage, the greater the baseline, the greater the number of loyal users you’ll win who will engage and spend. It’s simple math.
The key lies in your ability to significantly shoot up the percentage of loyal users. How so? For one, by effectively managing your ad spend – investing more in media sources and channels that have shown that they can deliver not only an install but also a loyal user who meets your goals – whether engagement or revenue-related.
So you want to use an attribution provider that not only tells you where a installs came from, but also more importantl, tells you where the best install came from. This is done by continuing to follow a user’s post-install activity and measuring his in-app events, and then tying these events back to the acquiring channel/network/campaign/ publisher/creative - depending on how deep you want to go.
Multiple Channels App Stores
Value-Driven In-App Activity
Clicks Install
Connect value back to acquiring channel & optimize
Not sure which KPIs to focus on? Take a look at some examples on the next page.
www.appsflyer.com 39 Getting Started with Mobile Attribution & Marketing Analytics
4.1 In-App Events to Get Started With:
BASIC
• Install/loyal user, conversion rate • App opens (for retention) • Revenue • Average Revenue Per User (ARPU) • Number of purchases
VERTICAL SPECIFIC
Travel • Hotel/flight/package searched/viewed • Add to wishlist • Registration • Initiated checkout • Hotel/flight/package booked • Logged-In
Gaming • Tutorial completion • Facebook registration • Achievement unlocked • Level passed
e-Commerce • Product viewed • Product added to cart • Registration • Product Purchased • Logged-In
www.appsflyer.com 40 Getting Started with Mobile Attribution & Marketing Analytics
Let’s take a look at some examples of in-app KPIs from AppsFlyer’s own dashboard.
Table 1: Sorted by Number of Installs
If volume is what you’re after, Network 1 attention clearly delivers. But if it’s post-install value, it delivers a loyal-user-to-install ratio that’s marketers well below average. When focusing on this KPI, Networks 5 and 2 reign supreme.
www.appsflyer.com 41 Getting Started with Mobile Attribution & Marketing Analytics
Table 2: Sorted by ARPU (Average Revenue Per User)
This app has a very solid client base with attention owned channels Email and SMS leading the pack. The table also shows that, while user marketers value delivered by organic traffic is below average, it delivers the majority of the apps’ revenue.
www.appsflyer.com 42 Getting Started with Mobile Attribution & Marketing Analytics
Table 3: Sorted by Purchased Items (Unique Users)
Campaign B of a travel app is obviously attention delivering massive scale but when it comes to the average order value (AOV), it marketers actually comes in last. Although Campaign A delivered only 43 purchases, it had the highest AOV.
www.appsflyer.com 43 Getting Started with Mobile Attribution & Marketing Analytics
4.2 Retention – Keep Users Coming Back for More
Your users have downloaded your app. That’s an important step. But don’t rest on your laurels yet – most of the work is still ahead. Now you have to make sure they open the app, use it regularly, and actually drive real value for your business.
So how do you get a user to open your app and not the dozens installed (on average) on his device? Not to mention the thousands of potentially relevant apps for that user in the app stores just waiting for their chance to take your place.
User retention is a major topic in app marketing. You can read more about it here and here. In this guide, we focus on attribution analytics, and how they can improve your retention rate and maximize lifetime value.
www.appsflyer.com 44 Getting Started with Mobile Attribution & Marketing Analytics
The retention rate is calculated as the unique number of users acquired by a specific network who were active on a specific day/week out of the total number of unique users who first launched the app in the selected timeframe.
The following chart from the AppsFlyer dashboard clearly shows that Network 1 has the highest retention rate from the get go, while Network 2 starts strong and remains strong although not as good as Network 1. Network 4 has the lowest retention on Day 1 but over time, it is Network 5 that loses the most users as only 4.8% remain after 10 days.
Chart 1: Retention Table
39.13% 34,860
37.78% Retention Report 12,901
Media Source Install Day Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 8 Day 9 Day 10
100.00% 39.13% 31.44% 27.64% 25.19% 23.65% 22.45% 21.51% 20.54% 19.76% 19.40% Network 1 89,095 34,860 28,012 24,624 22,444 21,072 20.003 19,168 18.303 17,607 17,283
Network 2 100.00% 37.78% 28.14% 24.15% 21.34% 19.53% 18.60% 17.12% 16.16% 15.30% 15.05% 34,509 12,901 9,712 8,335 7,364 6,741 6,420 5,909 5,578 5,279 5,193
100.00% 23.62% 17.92% 14.00% 11.85% 11.24% 9.73% 8.92% 7.86% 7.55% 7.64% Network 3 13,684 3,232 2,452 1,916 1,621 1,538 1,332 1,221 1,076 1,033 1,045
100.00% 13.15% 11.24% 10.51% 8.81% 8.65% 7.72% 6.87% 6.48% 6.12% 6.07% Network 4 10,595 1,393 1,191 1,114 933 916 818 728 687 648 643
100.00% 18.82% 13.48% 10.55% 9.05% 7.64% 7.31% 6.37% 5.54% 4.80% 4.84% Network 5 9,187 1,729 1,238 969 831 702 672 585 509 441 445
13.15% 6.07% 1.393 643
18.82% 4.84% 1.729 445
www.appsflyer.com 45 Getting Started with Mobile Attribution & Marketing Analytics
4.3 Cohort Analysis
A cohort report enables you to group users with common characteristics and measure specific KPIs over different timeframes.
For example, one cohort can be users who first launched an app any time during the month of January while another cohort are those who launched it during February and live in the US. This form of grouping enables an “apples to apples” comparison and therefore a really good indication of change over time. It tells us about the quality of the average customer and whether it’s increasing or decreasing over time.
Let’s explore the following example on the next page from AppsFlyer’s cohort report. This cohort includes users from Great Britain who installed the app between January 1 and January 31.
They are then grouped by the media source that acquired them, which allows us to analyze which network delivered users with the highest average sessions per user over time.
www.appsflyer.com 46 Getting Started with Mobile Attribution & Marketing Analytics
Chart 2: Cohort Report
Average Sessions per User
12.00
11.00 Network 1 3.79 10.00 Network 2 1.45 9.00 Network 3 11.49 8.00 Network 4 10.32 7.00 Network 5 12.32 6.00 Network 6 10.57
5.00
Sesions Per User Network 7 10.79 4.00
3.00
2.00
1.00 1 3 5 7 14 30 Days
Unlike retention, the metric is calculated per different timeframes, which represent the first X activity days per user, and then accumulated among all users (that’s why the graph points up).
What can we learn from the graph?
• Networks 1 and 2 underperform – consider removing these
• Network 5 growth (purple) is most impressive and constant over time – budget increase can make a lot of sense here
• Network 7 (pink) line loses its curve from day 14 – meaning engagement is dropping. Perhaps a retargeting campaign before day 14 can help maintain the curve in the long run
www.appsflyer.com 47 Getting Started with Mobile Attribution & Marketing Analytics
4.4 One Formula to Rule Them All
If we had to sum up a single formula of success in mobile advertising, this would be it:
attention LTV > CPI marketers
The bottom line is that, if your users generate greater value over time (spend or engagement), than what you invested to acquire them, you’re doing something right!
This formula factors in a lot of what we’ve discussed: acquiring the best users in the first place and then maximizing the value of the existing ones.
www.appsflyer.com 48 Getting Started with Mobile Attribution & Marketing Analytics
4.5 Access to Data
With so much good data to use, where and how can you actually view the data and analyze its patterns? Overall, there are 3 formats:
• Analytics Dashboard. View aggregated data or export performance, aggregated, and user level data to Excel. • Pull API. The advertiser pulls a bulk of data (hourly or daily). • Push API. The provider pushes the data to the advertiser in real time with information on a specific event.
Let’s explore the following marketer personas to better understand when to use each method:
I'm a small/medium sized I'm a medium/large sized app, with no internal BI app with a high volume system and no need for of data. I don't have an user level data. I want daily internal BI system and I monitoring that provides don't need organic data. a quick overview of media I use the source performance. I use the DASHBOARD PULL API
I'm a small/medium sized I'm a large sized app with app, with no internal massive volumes of data. BI system. I'm an Excel I have an advanced Bl wizard who likes to go system and I need organic deep by slicing & dicing. & non-organic user level data. I use the EXPORT FROM I use the THE DASHBOARD PUSH API
www.appsflyer.com 49 Getting Started with Mobile Attribution & Marketing Analytics
4.6 Mobile Ad Fraud – No longer the Elephant in the Room
Wherever there is money, there are bad guys looking for a piece of the pie. And it’s a growing piece of the pie with an estimated loss of $1.3 billion annually to mobile fraud (compared to $3.2 billion in desktop fraud), according to the IAB.
The good news is that there are plenty of ways to fight back. It won’t eliminate fraud entirely but it can definitely minimize it. Also, the subject is now increasingly raised across the mobile ecosystem, leading to more collaborations, which is an important step forward in the battle against fraudsters.
www.appsflyer.com 50 Getting Started with Mobile Attribution & Marketing Analytics
Common Attack Methods
IMPRESSION FRAUD The shady tactic by which publishers stack multiple display ads on the same piece of real estate.
CLICK FRAUD This black-hat technique uses an automated script or a computer program (aka bots) to imitate a legitimate user that clicks on ads.
INSTALL FRAUD With CPI being the most common pricing model in app marketing, install fraud is also most prevalent. It happens when Install bots mimic human behavior to automate an install.
IN-APP (EVENT) FRAUD An attempt to impersonate in-app activity: using an app, playing a game, or making fake in-app purchases (through a transfer of virtual goods where no real money is being exchanged).
Generally speaking, the deeper the funnel stage, the harder it is for the bad guys to succeed in their attacks (see chart below). But since the financial reward associated with each pricing model is highest at the bottom of the funnel (CPI and CPA), fraudsters also try harder, which leads to an increase in fraudulent activity.
www.appsflyer.com 51 Getting Started with Mobile Attribution & Marketing Analytics
Counter attack Only run with networks you trust: Thankfully, the ecosystem has hundreds of reputable and established sources of ad inventory.
Demand transparency: Make sure the networks you work with are transparent about their sources and sub-sources.
Use direct publishers: When you work with direct publishers or with networks that have relationships with direct publishers, you know where your ads are running.
Prevent fraud from reaching your dashboard: Automatic IP filtering based on machine learning algorithms can detect data anomalies in real time and reject them.
Dive deep into raw data reports: On-going monitoring and in-depth pattern analysis can detect fraud and block fraudulent sources.
Use in-app receipt validation: With receipt validation, you’ll know that real money was spent and that the revenue figure you see in the dashboard is accurate.
www.appsflyer.com 52 Getting Started with Mobile Attribution & Marketing Analytics
CHAPTER 5 THE 10 COMMANDMENTS OF MOBILE ADVERTISING ANALYTICS
1 2 3
Tie user acquisition Run with an attribution to post- Increase ROI by unbiased attribution install events that tracking Facebook provider that you are closely related and Twitter can trust to your business campaigns goals
4 5 6
Use a single Measure the Improve user attribution deeplink success of your loyalty with to measure all your retargeting cohort analysis campaigns across campaigns all platform
7 8 9
Measure a Run campaigns on campaign's impact your attribution Integrate in-app on web, cross- provider's events you care device and offline integrated networks about in the SDK purchases and to achieve the best from day 1 engagements optimization
10
Remember that LTV > CPI is the new formula for app success Getting Started with Mobile Attribution & Marketing Analytics
CHAPTER 6 CONCLUSION
The mobile revolution is in full swing. That means the app stores are packed with numerous competitors of your own app – sometimes even in the hundreds. Making it in this ultra, shark-infested competitive landscape, beyond having a superior app experience, requires investment in paid campaigns – whether acquisition or retargeting.
Relying on app store discoverability to drive organic installs will not do the trick. What is required is a mix of organic and non-organic installs to drive not only a volume of new users but more importantly a volume of new loyal users who will actually engage with your app. Also, it is recommended to run retargeting campaigns to better communicate with existing users and keep them happy and coming back for more.
An advanced attribution analytics platform can help you reach these goals by simplifying the fragmented mobile ecosystem and properly identifying users, while tying attribution to post-install events to achieve the highest possible ROI on your advertising campaigns.
www.appsflyer.com 54 Getting Started with Mobile Attribution & Marketing Analytics
About AppsFlyer
AppsFlyer is the leading mobile attribution and marketing analytics platform, measuring more than $4 billion in mobile ad spend annually. Over 10,000 app marketers, agencies and brands use our proprietary solutions to measure and optimize their performance. With over 1,900 integrated ad networks, and as an official measurement partner of Facebook, Google, and Twitter, AppsFlyer provides marketers with unbiased attribution, smart deeplinking, mobile campaign analytics, in-app tracking, lifetime value analysis, ROI and retargeting attribution for over 800 million installs each month. Clients include Playtika, IHG, Alibaba, Baidu, Trivago, Macy’s, Samsung, DeNA, and HBO.
About the Author
Shani Rosenfelder is a senior marketing manager at AppsFlyer. He has over 10 years of experience in key content and marketing roles across a variety of leading online companies and startups.
www.appsflyer.com 55 VISIT US AT:
Berlin London Beijing New York Seoul San Francisco Tokyo Tel Aviv
Bangalore Bangkok
Buenos Aires
www.appsflyer.com