It is a M’AD World!
description
Transcript of It is a M’AD World!
Overview• A Bird’s-eye view on the mobile ad landscape
• Mobile Ads evolution
• Device identifiers and Conversion Tracking
• Preview on Ad optimizations (Big Data)
• Best practices for Advertisers and Publishers
• Advanced Targeting Demo
• Miscellaneous
• Q&A
Bird’s eye view• Mobile ads by revenue ~ $9 billion global:
http://techcrunch.com/2013/07/09/iab-mobile-ad-revenue-will-pass-9b-globally-in-2013-after-83-surge-in-2012-thanks-to-smartphone-boom/
• KPCB Internet trends - http://www.slideshare.net/kleinerperkins/kpcb-internet-trends-2013
• Number of apps in iOS app store - ~ 900000 with 15Billion downloads
• Number of apps in Google playstore - ~850000 with 40Billion downloads
Bird’s eye view – cont’d (source: KPCB Internet Trends)
Bird’s eye view – cont’d
Mobile Ad Evolution
Publisher Server
Pre Smart Phone
http http
Ad N/W
request
impression
click
302 redirect
Landing Page
Conversion
Mobile Ad Evolution – cont’d• Lack of advanced html support like iframes, javascipt, css and cookies.
• Mostly mobile web pages
• Publisher side integrates Ad N/W code (in a particular language – php, ruby, java etc..) which proxies http request from device to the Ad N/W servers – HTTP headers from device very important.
• Ad N/W responds with a html snippet which is embedded on the publisher servers before going to the device.
• Winning Ad based on closed second-price auction.
• Mostly Text ads, but banner and video supported in some makes and models.
• Revenue share between Ad N/W and Publisher – e.g. 60 – 40 split.
• Advertiser usually charged on CPM (brand display) or CPC.
Mobile Ad Evolution
Publisher Server (optional for apps)
Smart Phone
http
Ad N/W
request
impression
click
302 redirect
Landing Page / App Store
Conversion
http
Mobile Ad Evolution• Smart phones usually support advanced html features – iframes, javascript, css and so on.
• On mobile web pages, tag based systems ensure request sent directly from the device to the Ad N/W servers
• Arrival of native apps supporting ‘web containers’ in sandboxed environments along with App Stores.
• Native apps usually embed Ad N/W SDK
• App Stores emphasize ‘conversions’ rather than ‘click’ optimizations for performance based advertising.
• With 3rd party conversion tracking, there can be more than one HTTP 302 redirect hops before reaching the landing page / app store.
• Lack of X-domain cookies still a major hurdle compared to desktop online ads – ‘common cookie jar’ in the latter a crucial driver for various innovations in the online ad industry.
• Richer Ad units possible along with the usual Text and Banners.
• Along with CPM and CPC, now CPI a major component in revenue models for performance based advertisers.
Mobile Ad Evolution - metrics digression
• Basic metrics that are tracked:– Requests– Impressions– Clicks– Conversions
• Derived metrics:– Fill rate % = (impressions/requests) * 100– Click-through-rate (CTR %) = (clicks/impressions)*100– Conversion rate (CVR %) = (conversions/clicks)*100
Mobile Ad Evolution – metrics digression
• Spend metrics for Advertisers:– Cost-per-Milli (CPM) = $ / 1000 impressions– Cost-per-click (CPC) = $ / click– Cost-per-install (CPI) = $ / install– Cost-per-acquisition (CPA) = $ / acquisition (CPI is a particular
case of CPA ).
• Advertisers charged based on the aforesaid spend models by Ad N/Ws and resultant revenue shared with the Publishers – e.g. 60% for publishers, 40% for the n/ws. Can be fixed CPM or CPC deals too.
Mobile Ad Evolution
• Traditional Ad N/W model limitations:– Proprietary integration for each n/w if publisher
need to work with different ones.– Mediation layers somewhat mitigate this problem– Overall less transparency
• One solution: Mobile Open Real-time-bidding (OpenRTB)
Mobile Ad Evolution – OpenRTB
App with SSP SDK Ad N/W 1
DSP 1
DSP 2
Ad N/W 2
DSP N
SSP Exchange
OpenRTB Protocol
request
Bid requests
Bid Responses
Win Notificationimpression
click302 redirectLanding Page / App Storeconversion
Mobile Ad Evolution – OpenRTB• Mobile OpenRTB spec:
http://code.google.com/p/openrtb/downloads/list (version 2.1)
• JSON based protocol
• Increased transparency
• Publisher inventory available in a competitive open market
• Pricing/Operational efficiency
Mobile Ad Evolution – OpenRTB• Leading to specializations:
– Supply-Side-platforms (SSP) - Publishers/App Developer side focus– Demand-Side-platforms (DSP) – Advertiser focus
• An ad request translates to several bid requests to all demand participants, and the highest bidder (CPM) picked as the winner. The bidders are expected to respond within 120ms.
• SSP exchanges select a winner by closed auction (usually second price), and the winner notified with the CPM price in real-time
• DSPs’ buy impressions on a CPM basis, but most likely getting paid only on CPM, CPC or CPI basis – effectively a trading platform!
• Publishers / App Developers can set a price floor for various ad units dynamically.
• For more details refer to: http://www.mikeonads.com/2007/05/01/the-ad-exchange-model-part-i/
• Sample RTB requests and responses - http://doc.drawbrid.ge/#65,67,88-
Device Identifiers• Crucial for conversion tracking (see best practices)!
• Only available in native app traffic.
• iOS 4 and older – usually SHA1 of UDID (now deprecated).
• iOS6 and above – Identifier-For-Advertisers (IDFA):– Usually passed in raw.– Users can turn off tracking, or change the identifier (transient).
• During transition between UDID deprecation and IDFA, various solutions were in place:– SHA1 of MAC Address (all uppercase colon separated)– ODIN1: SHA1 of MAC Address in byte form– OpenUDID (uses iOS Pasteboard)
• iOS7 onwards: IDFA the only approved device identifier:– Pasteboard changes deprecate OpenUDID– MAC Address api always returns 02:00:00:00:00:00
• Android ID (permanent identifier) still acceptable in Android platforms along with IMEI.
Device Identifiers – cont’d• Real-time conversion tracking essential for CPI based campaigns.
• Conversion events can be latent
• Standard click to conversion window is 14 days but can be configurable in some DSPs’
• Various third-party conversion tracking solutions now available which solves the multiple attribution problem – usually ‘last click’ wins.
• Increasingly advanced advertisers demand post-conversion optimization – analyzing life-time-value (LTV) of users.
• Conversion tracking still possible without device identifiers via ‘device finger-printing’ approaches at the expense of accuracy.
Ad Optimizations – Big Data
• In a nutshell, serve the most optimal ad that maximizes advertiser ROI and other goals, and maximize revenue to the Ad N/W / DSP and publishers.
• Optimal Ad selection based on various features/dimensions:– Attributes of the source of the request (native app or mobile web page):
• Keywords• categories
– Ad position– Ad dimension– Geo Location (Country, State, City)– DMA zone– Past performance of Ads under various dimensions: CTR, CVR– User attributes (behavioral targeting):
• Demographic (Age, Gender, Profession, Income Range)• Audience segments applicable to the user (technology professional, high-income)
– Much more!
• With OpenRTB, effective bidding strategies!
• Details not in the scope of this discussion but happy to answer them in the Q&A portion.
Best Practices - Advertisers• For CPI based campaigns, collect device identifiers for real-time conversion tracking either via 3rd party
conversion tracking solutions, or in-house approaches. This needs to be integrated to Ad N/Ws and/or DSPs’
• Have all the various standard IAB creatives for banner ads ready.
• Well designed creatives that grabs user attention.
• Periodically refresh the creatives (for at least top performing units every month)
• Choosing the right campaign strategy:– ‘Run of Data’– Audience Targeting– Re-targeting (Desktop to mobile or vice-versa)– Post-conversion optimization
• Support ‘App urls’ for launching and deep-linking (if applicable) – can enable ‘re-engagement’ and other post-conversion optimizations: http://iosdevelopertips.com/cocoa/launching-your-own-application-via-a-custom-url-scheme.html
Best Practices - Advertisers
• CTR By Ad unit size
CVR by Ad unit size
Conversions/1000 impressions
Best Practices - Advertisers
•
Best Practices - Publishers• Ensure advertising fits into the larger monetization strategy
– Free version of app with Ads leading to fuller paid versions – In-App purchases– Freemium model
• Design up-front on where to place the ads - native app or mobile web page
• Use device ids (for native app publishers) to maximize revenue from performance-based advertisers.
• Provide as much hints about the user as possible – see sample ‘user’ fields in OpenRTB spec.
• Maximize on larger ad unit sizes – e.g. interstitials, using natural pauses/breaks in the app to present ads (e.g. - leveling up in a game), not too distracting to the users and so on.
Advanced Targeting
• Demo
Misc.• Lot of improvement and innovation possible in richer ad units
highlighting mobile strengths:– Geo Location– Click to Calls– More interactivity (esp. in tablets)– In-App purchases (Ads leading to purchases without exiting the apps)
• ‘Re-engagement’ campaigns
• Dynamic Ad creatives
• Native Ads - increasingly a popular ad unit’that fits well into the app or web page content – e.g. Facebook mobile feeds.
Misc.• Future trends seem to indicate moving towards ‘cross screen advertising’.
• Re-targeting from desktop to mobile or vice-versa – e.g. user doing some preliminary research in a mobile device, but completing the transaction on the desktop.
• Devices will include next-generation DVRs’ – convergence of desktop/laptops, tablets, smart phones and TV.
• Ability to link users across their various respective devices.
• Taking to the next level, ability to link online campaigns leading to a offline purchase.
Q&A