I have covered details around Mobile Marketing, and structured my presentation 1) Landscape 2) Dollar Flow in the AD Ecosystem 3) Traditional VAS Marketing Methods for Feature Phones 4) Mobile Marketing 5) Automation of Campaigns based on Data - Machine Learning 6) Investor Interests in the Space 7) Traditional Cookie Data from Traditional Browsers

1. Mobile Marketing
ANSHU SHARMA – IIFT
17TH JAN 2015

2. Scope
 Traditional Telco Mobile Marketing – Feature Phone Marketing
 Channels
 Gratification Business Case
 Mobile App Marketing
 Channels
 Suggested Example Video
 Mobile Market Automation
 Segmentation, A/B Testing,
 In App & PNs Campaigns
 Analytics & Data
 Ecommerce Market Tracking Sheet & KPIs
 Deep Linking
 Cohort Analysis

3. Mobile Ad Landscape

4. * Includes: Ad Networks, SSPs, Private Exchanges, and Ad Rep Firms
** Includes: Media Agencies, Creative Agencies, Trading Desks
*** Includes: Ad Exchanges and DSPs
This slide represents Display, Video, Mobile – based on Rare Crowds Analysis of
Advertising Ecosystem Impression / Dollar
Flow
Note: Ad Servers, Yield Management Systems,
Analytics, DMPs, etc… Take out about 8-10%
of spend in total as they pass through the
ecosystem
20%
ADVERTISER AUDIENCE
Agency*
*
Keeps
5-25%
Large
Publisher
(Direct
Reserved
“Premium
”)
Keeps
75-95%
Large
pub
(Remnant)
Keeps
40-80%
Publisher
Aggregat
or*
Keep
10-80%
(May include
multiple
vendors)
Small-
Med
Publishers
Keeps
20-60%
65%
25%
5%
20%
25%
25%
FLOW OF ADVERTISING
IMPRESSIONS (US)
Percent impressions
40%
35%
40%
35%
Ad spend
100%
Exchang
es, DSPs
***
Keeps
10-20%
(may
include
multiple
vendors)
FLOW OF AD SPEND (US)
Percent Dollars Captured (in
yellow)
25%
5%
PerImpressionPriceDrops
4

5. Communication Platforms for VAS Promotions
VAS
PROMOTIO
N
PLATFORMS
SMS
Out Bound
Dialing
EOCN/USS
D
CELL INFO
PLC Cross
Promotion
IVRS
SELF CARE
START-
STOP
RETAIL/ Go
to Market

6.  Most Popular Communication Platform
 Advantages – Cost effective , Fast n Easy & Reliable
 Perception – Considered intrusive, irrelevant
 Can be leveraged by
 Base Segmentation
 Vernacularisation
 Timelines
 Over 65% customers prefer to read promo SMS on Sunday/Holidays; Weekdays are not
proffered especially Monday & Tuesday
 Customers would also prefer Promo SMS after 5 PM
SMS

7.  Tagging Promotions on Activation and Deactivation messages across related / unrelated
product categories
 Behtar Zindagi 55678 IVR be promoted as a tag promo on Deactivation message of “Mitti Ke rang”
 Behtar Zindagi promo can be tagged with on MT Charging SMS of Talk to Me
 Tagged promo on Activation Confirmation of a related / untelated VAS category across other VAS
Content providers.
 Behtar Zindagi Can be tagged on LAPU Conf SMS on IFFCO Base
 Explore possibilities of tagging promo on HT Song Change Confirmation SMS to Rural base
 Tagged Promo on Service messages - Roaming ; Users usually read sms received during travelling
and a far likely to try a service
 AFC from Altuist can be promoted on Drop over message on Raat Baki IVR of Hungama
 Friend Locator is likely to get better conversion if tagged with PLC content of Friends Chat.
PLC Cross
Promotion

8.  POP Material in the rural market. Nearest Grain/ Vegetable Markets.
 Incentivize Retailer on activation of VAS through PEF/SEF
 POP Material at IFFCO Retailers for VAS Targeted at Rural base
 POP Material at Youth Hangouts for Youth centered VAS
 Festive gatherings, Religious gatherings, Agro Fests or and related events attended by rural folk
 Branding on Public transport
 Branding on Public carriers
 POP at Weekly Bazaars
 Celeb Endorsements on radio and RJ mentions
 Radio Contests
 Cinema Advertising / Brand
 Branding on Transport used by FOS and Incentive on conversion
 Buzz on Social Media through Celeb endorsements, Winners of Various Contests
RETAIL/ Go
to Market

9. Telco/VAS Marketing Plan:
Gratifications
Operator Subscriber Base of UPU
Airtel 18,055,029
Airtel Rural Base in UPU 9,930,266
Avg Price point @ Rs15, Total Revenue/ Month
72,300 1,084,500
Activity Conversion Number/day
USSD Promotions,4 hrs/ 30,000 300
Cell info, 1 week/ 2,40,000 1,000
SMS push of Toll number - 1 lacs sms for 1K calls 100
Ground Activity - 10
OBD from Airtel End: 300,000 => 4000 Subs 1,000
Monthly Subscription 72,300
Total Airtel Revenue 1,084,500
Company Share @ 36% - Earning 0
Gratification Cost 10% of Handygo share 0
Company Share 0
Projected Subscriber Monthly Projected Revenue (INR)
289200 2,892,000
Airtel Share Share @ 66% 1,908,720
Company Share @ 36% - Earning 983,280
Gratification Cost 10% of Handygo share 98,328
Company Share 884,952
Operator Subscriber Base of UPU Rural Base in UP
1st Month
Penetration
2nd Month
Penetration
3rd month
penetration
4th month
penetration
5th month
penetration 6th month penetration
Airtel 18,055,029 9930265.95 29791 49651 69512 99303 129093 158884
Avg Price point @ Rs20 595816 993027 1390237 1986053 2581869 3177685

10. Mobile Marketing Non Feature
Phones
 The Telco managed mobile Marketing –
 SMS - Short Codes
 Outbound Dialer – Long Codes
 USSD – Customer Self Care Portal - *121#
 Cell-id based Broadcasts
 IVRs
 Missed Call Alerts
 In Message Advertisement
An Example – Behtar Zindagi

11. Mobile Marketing
 Mobile Website / Native App strategy
 Location Based Marketing
 Permissions / Preferences

12. Appstore Optimization : ASO
 Keyword optimization (KWO)
 Keyword Optimization (also known as keyword research) is the act of
researching, analyzing and selecting the best keywords to target and drive
qualified users from app stores to your app. App store optimization tool provider
Sensor Tower also breaks keyword optimization down into three parts: relevance,
Difficulty Score and Traffic Score.[16]
 Conversion rate optimization (CRO)
 Conversion rate optimization involves all metadata available and publicly
accessible in the app stores, like icons, screenshots, description and update
texts.[17][18] This part of app store optimization is responsible to convert the traffic
acquired through keyword optimization into app downloads.
 Location labs Appstore Optimization Slides

13. The Keys to Mobile Marketing
 Usability
 Do you have a mobile optimized site?
 Do you have a native mobile app?
 How many steps from search to purchase?
 Trust
 Do you have a user rating?
 Are people talking about your brand online?
 Personalization
 Are you finding you customer at the right time?
 Are you reaching your customer in the right place?
 Are you targeting your customer with the right message?

14. Customer Engagement &
Retention
 Push notifications
 In-app messaging
 Deep-linking
 Gamification

15. Mobile Marketing Campaign Gist

16. WOWOKAY, SO HOW AM I GOING TO DO THAT ?
DO WE HAVE A DEDICATED MAN SITTING BEHIND AND OBSERVING LIKE WE ARE
DOING IN THIS VIDEO?

17. Challenges
 Location – Where am I ? Office or Home or my shop ?
 Personality/ Personas – If I love to camp or play football ? Which Segment?
 How do you know my existing products, my brands and what needs
replacement ?
 How do you know if I look for expensive brand or a less priced ?
 How do you know that I would love to fly as a second alternative ?
 How come an app knows that I could make payment through my mobile?
 What happens to my privacy ?

18. How is a mobile marketer going to
make all these rules?
 Should I map the Customer Journey ?
 Event based Rule engine should do ?
 How many such rules would be required to monetize the moments of
truth ?
 Is it feasible ? YES
 Lets Make some Rules - Exercise

19. Mobile Market Automation Aspects
 Segmentation
 PUSH
 In App Campaigns
 A/B Testing
 Analytics & Data

20. Automated A/B Testing
 Experimenting Options Available w.r.t every touch points with the
customer in terms of
 Message
 Content
 UI elements
 Run the experiments on consumer segments to target based on
Location, device type, app version, traffic source, and in-app
behaviors.
 Add your own Attributes and events specific to the particular brand
app
 Running concurrent experiments on the Fly

21. Event-Based Triggers
• Behvior Triggers
• Add to Card
• Card Abandonment
• Checked a Product /discount
items
• Has Shared socially with friends
• Life cycle Based
• New User
• Repeat Customer
• About to Churn
• Calendar Events
• Real Life Event
• Instant Gratification
• Entering a Mall/Shop

22. Geofence and iBeacon Triggers
 Wifi Finger Printing
 iBeacon Indexing
 GPS/Cell-Id

23. custom in-app message templates to drive higher
user engagement.

24. Dynamic Targeting
Send highly personalized messages to
• specific customer segments
• geography,
• device type,
app version,
• traffic source,
• custom user attributes, and in-app
behaviors

25. Native Ads – How and ROIs
 TBD
 For Reference – Snap Deal Marketer
 http://appiterate.com/usecases.html

26. App Widgets – Mobile Screen Real
Estate
 TBD

27. Ecommerce Marketing Tracker
 Marketing Spend:
 Visitors to the website/ Mobile /channel
 Orders / channel
 Orders / Customer: How many orders does a customer submit, who was first
attracted through this channel. This KPI is influenced by other factors as well, but
gives you an initial feeling for the customer quality. It is up to you how you define
“lifetime” (1m/6m/1y). In a later version I will go into CLV management more
deeply.
 Revenue/ channel
 Discounts Needed/ Channel

28. Are these KPI Useful ? What do I do now?
 Basket Size (Basket): How much revenue did this channel generate per order. It helps you to
understand the economic value of the customers that you attract through the different marketing
channels.
 Conversion Rate (CR): How many customers per 100 visitors, that came to your site, finally ordered a
product? The Conversion Rate helps you to understand if people that came through this channel only
browsed around, or actually purchased something.
 Cost Per Order (CPO): How efficient is this marketing channel? The cost per order tells you how much
you spent to generate one order.
 Real Cost Per Order (real CPO): To be able to compare channels on a CPO basis (which channels
generates the cheapest customers), the CPO should be adjusted by certain factors. One big
influencer is the discounts you needed to give to generate an order. E.g. flash sales are usually
relatively cheap to initiate, however you need to give huge discounts. This increases the adjusted
(real) CPO accordingly.
 Customer Lifetime Value (CLV): How much revenue does a customer that is generated through this
channel generate for you. This factors in the average basket size, as customers that come through
different channels reorder differently, and spend different amounts.
 Return On Marketing Investment Factor (ROMI Factor): How much revenue did you generate per dollar
invest? It helps you to understand the return per marketing channel. It is already adjusted with
discounts to compare the factor cross channel. If you want you can adjust the factor by other
influencers.

29. Exercise – Abandon Cart?
 How do I retarget ?
 Rules
 Time Duration after someone traversed and left
 Message Method
 Discount
 Segment
 Timing

30. Attribution
Modeling
A NECESSITY FOR CONVERSION STRATEGY

31. Last Touch Model
 Measure - What was the last touch point the customer interacted ?
 What is the conversion % for each last touch point ?
 Generally, Email is considered the greater impact for sales conversion

32. First touch Model
 Measure – What % of the consumer use which channel as First
touch or the first interaction or the first keyword (in direct search)
 What is the impact % table for each touch point in FTM ?
 FTM works as a strategy tool for Social Media marketer

33. Equal weight Distribution
 Natural search %
 Comparison Shopping % - A-B/Split Testing & Strategy
 Banner %
 Email %
 Interstitial
 In App
 Push Notifcaion
Measure – Resultant Comparison

34. Customer Credit Sharing
 Engagement Factors
 Click on Ad
 Simply Viewed
 In Video Banner Ad
 Media Factor
 Std Ad
 Animated Ad
 Size of the Banner Ad
 Time Factor
 Time span between ads viewing & conversion
 Position Factor
 Position the offer/ad conversion %

35. Attributions Model -Campaign
Measurements & Results
 Which Ad channels drive the best results ?
 How are these channels influence each other ?
 Which mix of ad/market spend on these channels works best ?
 What kind of creative ?
 What size, placement & frequency drives the consumer behavior ?
Basically what tactics works best for your business !

36. Custom Credit Rule Examples
 What are the customer credit Rules that works best and leads to conversion.
 It could be a series of interactions before the moment of truth
 Mapping the Offer/Ad Interaction Maps and conversions provides the best
media/campaigns budgets.

37. GA – Just for Reference

38. Measuring Mobile Marketing
Success
 Campaign Measurement / Channel
 Cohort Analysis

39. Sample Cohort Analysis

40. Analyzing the Cohorts
On Boarding Trend, the orange left arrow, indicates the
product’s effectiveness in its first month of use and its trend
over time, which is nothing less than a metric for user on
boarding effectiveness.
The first cell in each column indicates the monthly active rate
for the cohort’s first month as users. In our hypothetical data
set, that number’s growth varies from 35% to 41% over time.
The product team has done a reasonable job of improving
user on boarding and engaging users when they sign up
Longitudinal Trend, the top red arrow, indicates how the
activity rate changes as users continue to use the product.
The first row is the oldest cohort of users with the most recent
data, the ones who signed up most recently. The bottom row
is the newest cohort. Time flows right in this chart.

41. Cohort Analysis
 Average Revenue Per Customer Over Time - Chart monthly revenue
over time to contrast with cohort data
 Individual Channel Growth Over Time - Chart all accounts to visualize
trends.
 Number of Customers in Each Cohort - Chart number of customers in
each cohort to see how sensitive cohort data is to sample size and also
see the size of the new customer pipeline over time.
 Average Monthly Revenue By Cohort - Chart the revenue by cohort to
see if newer customers generate more or less revenue than older
customers. Really good for marketing spend evaluation.
 Cohort Comparison - Chart the different cohorts over time to see how
their revenue characteristics compare.

42. Why Deep Linking Matters ?
 Links in email/sms not prompting the user to open the native app and
straightway opening a browser
Quite prominently seen in most of the apps
 Ecommerce Alerts providing the Access to Purchase, however not
taking directly to Purchase page or CTA landing page
 Asking to login on a portal – Ebay and not taking to PDP
 ETSY has fixed it and fixed the revenue leakage as well
 Marketing links automatically detects the presence of app and auto-
matically takes to either browser or app
 Gaming Apps can save the session today, however they can include
the saved sessions in the PN and straightway take it to the that level
 In App / PNs directly translate into revenue
 TW Cards / Google App Indexing / FB App Link Tags

43. Deep Linking is Context Critical for Mobile Marketing
 ADs 2 App
 SMS 2 App
 QR 2 App
 EMAIL 2 App
 WEB 2 App
 SOCIAL 2 App
 App 2 App
Landing Page Optimization Strategy per traffic source.
Cohorts / Traffic Source would enable to fix the leakages and marketing campaigns
improvement

44. Good Read on Deep Linking

45. Investment Landscape

51. Merging the Old & the New World
 Browser Cookie Data Management - DMP

52. Data Management Platform ( DMP
)
 A very smart, very fast cookie warehouse with analytical firepower
to crunch, de-duplicate, and integrate your data with any
technology platform you desire
 Demdex (now owned by Adobe), RedAril, and Krux are what I
would consider pure-plays, while Lotame, Collective, and Turn
provide services

53. DMP Information flow

54. Multi Channel Data
Aggregations@DMP

55. Cookie Match Companies
 LiveRamp,
 DataLogix,
 Datran,
 TargusInfo,
 Acxiom

56. Ad Cookies – Step by Step
 A request is sent from the browser to retrieve an ad from the publisher
network.
 The publisher network retrieve a number of variables from the publisher -
in some exchanges the publisher chooses which data it wants to share.
Publisher variables include:
Anonymity (if anonymous URL above is not shown)
URL e.g. youtube.com
CookieId (buyers can use this to match the user to a pervious seen user,
e.g. for a remarketing campaign)
Vertical - e.g. Videos > Sport
Blocks - e.g. no Google Chrome ads please
Location - e.g. user is in UK, London


57. Ad Cookie – Step by Step …
 The network then sends out a request to the buy-side to find ads
e.g. a request is sent out to the DSPs and Buyer Networks.
This includes the publisher variables that are set in the request
sent to the exchange.
 The DSPs and BuyerNetworks then run a query.
SELECT snippet, bid FROM all_advertisers WHERE targeting_url = request_url
 Bids are then returned by the buyers within the 120ms threshold.
 bidder_A:
Advertiser: amazon.com
CPM: $2.50
bidder_B:
Advertiser: ebay.com
CPM: $0.50
 The winning ad is then selected. The ad is sent back to the users
browser. The buyer often pays the second highest price to the buyer.

58. Real Time Bidding
 User visits a website: say abcd.com.
 Within abcd.com there is a HTTP request to SSP, to fill an ad slot.
 On receiving the request for showing the ads, the SSP conducts a
real time auction. To each of the DSPs whose have expressed interest
in this user (some SSPs may be willing to show ads to users from a
specific geo, some may be willing to show ads on a certain website,
a retargeting DSP may be willing to show ads to a predefined set of
users etc.)
 The SSP sends a bid request. The bid request looks like this:
[ "auction_id": 1234abcd,
"geo": "Bangalore, India",
"ad_width": 728,
"ad_height": 90,
"website": "abcd.com",
"id": ssp1234
]

59. Real Time Bidding – Demand Side
..  On receiving the bid request, each DSP needs to send a bid
response. DSPs typically calculate bid response based on the
parameters in bid request (geo, banner size, etc) and the user profile
that DSP has stored for user id dsp1234 (Remeber that ssp1234 was
mapped to dsp1234 in cookie mapping stage, and data provided by
DMPs is stored against the key dsp1234)
Bid response looks like this:
["auction_id": 1234abcd,
"bid_value": 12.34
"adTag": "<script type='text/javascript'>
document.write ("<script type='text/javascript'
src='dsp.com/showad'");
document.write ("</script>");
</script>"
]

60. Real Time Bidding – Awarding the
Inventory
 The SSP compares the bid response of each of the DSPs, and
awards the impression to the highest bidder. These auctions are
usually second price auctions: the highest bidder wins and cost to
highest bidder is second highest bid in the auction.
 The SSP redirects user browser to the ad tag provided in the bid
response, which renders the ad to the user's browser.

61. APPENDIX

63. Reference Digital Marketing Periodic Table for Lingo
& KPIs

64. Players
•Adobe Marketing Cloud Appboy
•Appiterate
•Artisan
•DeltaDNA
•Kahuna
•LeanPlum
•Localytics
•Nudge
•Playfab
•Playnomics
•PushSpring
•Scientific Revenue
•Silverpop
•Swrve
•Tapjoy (via 5Rocks)
•Tapcrowd
•Upsight (formerly Kontagent)
•Urban Airship
 Which mobile automation features do you use? *This
question is required.
 Targeting
 Cohort analysis/user profiles
 Segment targeting
 Retargeting app users outside of your app
 Targeted in-app rewards or incentives
 Analysis
 A/B and/or multivariate testing
 LTV tracking
 Real-time analysis
 Surveys, ratings, feedback

• Messaging
• Push notifications
• Email campaigns
• SMS campaigns
• In-app messaging
• Location-specific
behavior/messaging
• Optimal time, delayed, or
flexible messaging
• Engagement
• Personalizing app content
• Personalizing app
functionality and/or
gameplay
• Promotions and/or sales
•

65. Acronyms
 Eng. Rate Engagement rate. For promoted tweets on Twitter, engagement rate is calculated by dividing the number of
engagements a promoted tweet receives by the number of impressions.
 Follow Rate For promoted tweets on Twitter, follow rate is calculated by dividing the number of follows by the number of
impressions within a campaign.
 Form Submits Form submissions on a website. These metrics indicate what percent of form submissions come from each
digital marketing source (e.g. paid search and referral traffic, email campaigns, social media).
 Gross Open Rate The number of times an email message is opened, either by the original recipients or by those to
whom the recipient forwarded the message, divided by the total number of delivered messages. Also known as total open
rate.
 Like Rate Facebook page like rate. The number of page likes divided by number of impressions per ad.
 MQL Marketing-qualified lead. This is a lead that Marketing has vetted and passes on to Sales.
 RL Raw lead. This is a lead that has not yet been vetted and accepted by Marketing.
 SQL Sales-qualified lead. This is a lead that has been passed on to Sales from Marketing, and accepted by Sales.
 Unique Open Rate The number of unique recipients that opened an email message divided by the total number of
delivered email messages. This measure does not count multiple email opens by a single recepient

66. Thanks
Anshu.bakshi@gmail.com
9910695016