Craft of Personalization
Andrey Tyschenko
Country Director, CEE, Russia, CIS, Dynamic Yield
History of Personalization.
4 Season Series.
Personalization Season 1. Show everything to everybody
If I and Warren Buffet were site visitors, average user wealth is $45 Billion
Personalization Season 2. Manual Personalization
Ok buddy, I’ll change text on that button in 3 weeks
Personalization Season 3. Stack of Separate Tools for Each Task
Banners, Recommendations, Popups, Emails, Push, AB-testing… what
else should we personalize?
Personalization Season 4. Single Platform for Personalization
… exactly what Dynamic Yield is
About Dynamic Yield
Dynamic Yield helps marketers increase revenue by personalizing each customer
interaction across web, mobile web, mobile apps and email.
500 MILLION
unique users per month
10 BILLION
page views per month
600 BILLION
events captured per month
BACKED BY:
Trusted by the World’s Leading Brands
Athletics & Outdoors
eCommerce
Fashion & Apparel
Beauty & Personal Care
Food & Beverage
Home & Furniture
Other eCommerce
Horizontal Platforms
Luxury & Jewelry
Footwear & Accessories
Consumer Electronics
Media & Publishing Business to Business
Betting & Gaming
Listings & Classifieds
Travel & Hospitality
DY is a Single Technology Stack for Personalization
• Front-end personalization
• Product & Content Recommendations
• A/B Testing and ML-Powered Experience Optimization
• Message Triggering
• Customer Data Management
• User Segmentation
Across Any Channel: Web Mobile Email APIs
Let’s see some real-life cases!
Demo Use case #1: Homepage Hero Banner
Point and click the element you want to replace!
Demo Use case #1: Homepage Hero Banner
Use Case #1: FjallRaven Homepage Banner - Default
Use Case #1: FjallRaven Homepage Banner - Women
Winter is coming…
Use Case #1: FjallRaven Homepage Banner – Cyber Monday
Use Case #2: FjallRaven: Category Recommendations at HP
Use Case #2: FjallRaven: Category Recommendations at HP
Looked for Women parkas? Hey, we remember that!
Use Case #2: Tottanham Football Club: Category Recs at HP
Use Case #2: Tottanham Football Club: Category Recs at HP
Use Case #3: Urban Outfitters: Packs of auto AB-tested banners
auto-changed weekly
When 10 banners per week route like trains with a schedule
Use Case #4: Lamoda: Changing the whole Homepages: Default
Use Case #4: Lamoda: Changing the whole Homepages: Women
Use Case #4: Lamoda: Changing the whole Homepages: Men
Use Case #4: Lamoda: Changing the whole Homepages: Kids
Use Case #5: Ozon category banners based on user interests
returned by 3rd Party data provider (Mail.ru)
Warm trends in
the air
Use Case #5: Ozon category banners based on user interests
returned by 3rd Party data provider (Mail.ru)
Oh no, I have
never looked
for women
spring fashion,
honestly!
Demo Use case #6: Page Action to change page layout
Use Case #6: Juniqe: Product Page Layout Optimization
Product description under or to the right
of Product name?
Show all Product attributes or only some?
Show Product details closed or open?
What is the best placement for Add To Cart
button?
Use Case #7: Under Armour Mobile App: Homepage personalization
A whole page re-built
personally for you
Use Case #7: Under Armour Mobile App: Homepage personalization
A whole page re-built
personally for you
Use Case #7: Under Armour Mobile App: Homepage personalization
A whole page re-built
personally for you
Use Case #8: Lamoda: Highlight Menu Items
Struggle to find your favorite brands in a long long menu?
Follow the white rabbit ;)
Use Case #9: Hollar.com: Product Listing Auto Personalization by
User Affinity
Use Case #9: Hollar.com: Product Listing Auto Personalization by
User Affinity
Is it a coincidence?
Somehow I start
seeing more and more
relevant product items
in the listing!
Demo Use case #10: Create Personalized Overlays
Use Case #10: Sephora: Mobile Homepage sequence of 3
Subscription Overlays
Tap here
Use Case #10: Sephora: Mobile Homepage sequence of 3
Subscription Overlays
Enter email
here
Use Case #10: Sephora: Mobile Homepage sequence of 3
Subscription Overlays
Only subscribe?
Hey, hold on, sign up
for the loyalty club!
Use Case #11: Ozon.ru: Social Proof – rocket-science logic behind
a super simple message
Use Case #12: Fjall Raven: Product page Low-stock Messages
I feel pretty much like at Booking.com J
Use Case #13: Lamoda: Countdown timer to create urgency
Order within 20 minutes and get 10% off!
Use Case #13: Lamoda: Count-down timer for an Urgency Offer
Hmm, looks like I should decide quickly and not let 10% off go!
Use Case #14: Barnes & Noble: Category context-based promos
Same static UI element. No popups. Always relevant.
Use Case #15: Barnes & Noble: Product Page Exit Intent
Recommendations with Similar products
Give us another
chance – maybe you
like that one better?
Demo Use case #16: Recommendation Widget
Test Algorithms and Visual Design in Real-Time
Use Case #16: LuisaViaRoma: Product Page Recommendations
Use Case #16: LuisaViaRoma: Cart Overlay Recommendations
Use Case #16: LuisaViaRoma: Check-out Recommendations
Oops… No
recommendations!
But wait… is it really
worth to show Recs
at Check-out?
Let’s do an exercize
Example: A Sports Team Tests a New Recommendations Widget
Strategy A : Similar Items
Strategy B :
Similar & Bought Together
Control No Recommendations
Baseline
Strategy A: +27.1% ARPU
No Recommendations
Strategy B: +34.4% ARPU
A Sports Team Tests a New Recommendations Widget Placement
Dynamic Yield’s Predictive Targeting
Recommended:
Use Strategy A for “Kit Buyers” Segment
Use Strategy B for All Other Visitors
Revenue per User:
+54.1%
Predictive Targeting
Revenue per User:
+34.4%
A/B Test
Use Different Merchendizing Strategies for Different Segments
3 people have purchased this item today. Don’t miss out!
Another Example: Impact of ‘Popularity Messaging’
Does ‘Popularity Messaging’ Work?
Segment Results
Italy +9.40% Purchases / User
+14.98% Revenue / User
Rest of the World
+0.60% Purchases / User
But negative revenue impact
Breaking Down ‘Rest of World’
+7.5%
-10.6%
Color of Buy Now button on PDP
+7.8% 98%0.10755,919523,480Pink
-6.5% 0%0.086838,551444,183Pink
eCommerce Manager, Last Week
Run: Switching to Machine-Learning is Simple
Predictive Targeting Drives Massive Uplift
Create a few
variations
Apply Targeting
with One Click
Profit!Predictive Targeting
identifies statistically significant
targeting opportunities
* Patent Pending
Machine-Learning Engine Alerts you upon Recognizing a
Segmentation Opportunity
Serve Control to all users
Serve Variation A to Audience Direct Traffic, and Control to the rest of the users
Serve Variation A to Audience Paid Search Traffic, and Control to the rest of the users
Learning:
Ignore Personalization and Leave
Millions on the Table
Thank You
Want this deck? Email me: andrey@dynamicyield.com

Andrey Tyschenko: Craft of Personalization

  • 1.
    Craft of Personalization AndreyTyschenko Country Director, CEE, Russia, CIS, Dynamic Yield
  • 2.
  • 3.
    Personalization Season 1.Show everything to everybody If I and Warren Buffet were site visitors, average user wealth is $45 Billion
  • 4.
    Personalization Season 2.Manual Personalization Ok buddy, I’ll change text on that button in 3 weeks
  • 5.
    Personalization Season 3.Stack of Separate Tools for Each Task Banners, Recommendations, Popups, Emails, Push, AB-testing… what else should we personalize?
  • 6.
    Personalization Season 4.Single Platform for Personalization
  • 7.
    … exactly whatDynamic Yield is
  • 8.
    About Dynamic Yield DynamicYield helps marketers increase revenue by personalizing each customer interaction across web, mobile web, mobile apps and email. 500 MILLION unique users per month 10 BILLION page views per month 600 BILLION events captured per month BACKED BY:
  • 9.
    Trusted by theWorld’s Leading Brands Athletics & Outdoors eCommerce Fashion & Apparel Beauty & Personal Care Food & Beverage Home & Furniture Other eCommerce Horizontal Platforms Luxury & Jewelry Footwear & Accessories Consumer Electronics Media & Publishing Business to Business Betting & Gaming Listings & Classifieds Travel & Hospitality
  • 10.
    DY is aSingle Technology Stack for Personalization • Front-end personalization • Product & Content Recommendations • A/B Testing and ML-Powered Experience Optimization • Message Triggering • Customer Data Management • User Segmentation Across Any Channel: Web Mobile Email APIs
  • 11.
    Let’s see somereal-life cases!
  • 12.
    Demo Use case#1: Homepage Hero Banner Point and click the element you want to replace!
  • 13.
    Demo Use case#1: Homepage Hero Banner
  • 14.
    Use Case #1:FjallRaven Homepage Banner - Default
  • 15.
    Use Case #1:FjallRaven Homepage Banner - Women Winter is coming…
  • 16.
    Use Case #1:FjallRaven Homepage Banner – Cyber Monday
  • 17.
    Use Case #2:FjallRaven: Category Recommendations at HP
  • 18.
    Use Case #2:FjallRaven: Category Recommendations at HP Looked for Women parkas? Hey, we remember that!
  • 19.
    Use Case #2:Tottanham Football Club: Category Recs at HP
  • 20.
    Use Case #2:Tottanham Football Club: Category Recs at HP
  • 21.
    Use Case #3:Urban Outfitters: Packs of auto AB-tested banners auto-changed weekly When 10 banners per week route like trains with a schedule
  • 22.
    Use Case #4:Lamoda: Changing the whole Homepages: Default
  • 23.
    Use Case #4:Lamoda: Changing the whole Homepages: Women
  • 24.
    Use Case #4:Lamoda: Changing the whole Homepages: Men
  • 25.
    Use Case #4:Lamoda: Changing the whole Homepages: Kids
  • 26.
    Use Case #5:Ozon category banners based on user interests returned by 3rd Party data provider (Mail.ru) Warm trends in the air
  • 27.
    Use Case #5:Ozon category banners based on user interests returned by 3rd Party data provider (Mail.ru) Oh no, I have never looked for women spring fashion, honestly!
  • 28.
    Demo Use case#6: Page Action to change page layout
  • 29.
    Use Case #6:Juniqe: Product Page Layout Optimization Product description under or to the right of Product name? Show all Product attributes or only some? Show Product details closed or open? What is the best placement for Add To Cart button?
  • 30.
    Use Case #7:Under Armour Mobile App: Homepage personalization A whole page re-built personally for you
  • 31.
    Use Case #7:Under Armour Mobile App: Homepage personalization A whole page re-built personally for you
  • 32.
    Use Case #7:Under Armour Mobile App: Homepage personalization A whole page re-built personally for you
  • 33.
    Use Case #8:Lamoda: Highlight Menu Items Struggle to find your favorite brands in a long long menu? Follow the white rabbit ;)
  • 34.
    Use Case #9:Hollar.com: Product Listing Auto Personalization by User Affinity
  • 35.
    Use Case #9:Hollar.com: Product Listing Auto Personalization by User Affinity Is it a coincidence? Somehow I start seeing more and more relevant product items in the listing!
  • 36.
    Demo Use case#10: Create Personalized Overlays
  • 37.
    Use Case #10:Sephora: Mobile Homepage sequence of 3 Subscription Overlays Tap here
  • 38.
    Use Case #10:Sephora: Mobile Homepage sequence of 3 Subscription Overlays Enter email here
  • 39.
    Use Case #10:Sephora: Mobile Homepage sequence of 3 Subscription Overlays Only subscribe? Hey, hold on, sign up for the loyalty club!
  • 40.
    Use Case #11:Ozon.ru: Social Proof – rocket-science logic behind a super simple message
  • 41.
    Use Case #12:Fjall Raven: Product page Low-stock Messages I feel pretty much like at Booking.com J
  • 42.
    Use Case #13:Lamoda: Countdown timer to create urgency Order within 20 minutes and get 10% off!
  • 43.
    Use Case #13:Lamoda: Count-down timer for an Urgency Offer Hmm, looks like I should decide quickly and not let 10% off go!
  • 44.
    Use Case #14:Barnes & Noble: Category context-based promos Same static UI element. No popups. Always relevant.
  • 45.
    Use Case #15:Barnes & Noble: Product Page Exit Intent Recommendations with Similar products Give us another chance – maybe you like that one better?
  • 46.
    Demo Use case#16: Recommendation Widget
  • 47.
    Test Algorithms andVisual Design in Real-Time
  • 48.
    Use Case #16:LuisaViaRoma: Product Page Recommendations
  • 49.
    Use Case #16:LuisaViaRoma: Cart Overlay Recommendations
  • 50.
    Use Case #16:LuisaViaRoma: Check-out Recommendations Oops… No recommendations! But wait… is it really worth to show Recs at Check-out?
  • 51.
    Let’s do anexercize
  • 52.
    Example: A SportsTeam Tests a New Recommendations Widget Strategy A : Similar Items Strategy B : Similar & Bought Together Control No Recommendations
  • 53.
    Baseline Strategy A: +27.1%ARPU No Recommendations Strategy B: +34.4% ARPU A Sports Team Tests a New Recommendations Widget Placement
  • 54.
    Dynamic Yield’s PredictiveTargeting Recommended: Use Strategy A for “Kit Buyers” Segment Use Strategy B for All Other Visitors
  • 55.
    Revenue per User: +54.1% PredictiveTargeting Revenue per User: +34.4% A/B Test
  • 56.
    Use Different MerchendizingStrategies for Different Segments
  • 57.
    3 people havepurchased this item today. Don’t miss out! Another Example: Impact of ‘Popularity Messaging’
  • 58.
    Does ‘Popularity Messaging’Work? Segment Results Italy +9.40% Purchases / User +14.98% Revenue / User Rest of the World +0.60% Purchases / User But negative revenue impact
  • 59.
    Breaking Down ‘Restof World’ +7.5% -10.6%
  • 60.
    Color of BuyNow button on PDP +7.8% 98%0.10755,919523,480Pink -6.5% 0%0.086838,551444,183Pink
  • 61.
  • 62.
    Run: Switching toMachine-Learning is Simple
  • 63.
    Predictive Targeting DrivesMassive Uplift Create a few variations Apply Targeting with One Click Profit!Predictive Targeting identifies statistically significant targeting opportunities * Patent Pending
  • 64.
    Machine-Learning Engine Alertsyou upon Recognizing a Segmentation Opportunity Serve Control to all users Serve Variation A to Audience Direct Traffic, and Control to the rest of the users Serve Variation A to Audience Paid Search Traffic, and Control to the rest of the users
  • 65.
    Learning: Ignore Personalization andLeave Millions on the Table
  • 66.
    Thank You Want thisdeck? Email me: andrey@dynamicyield.com