Personalization Palooza 2016

28
Personalizing Television. #ppalooza - Personalization Palooza 2016 - NYC MEDIA LAB

Transcript of Personalization Palooza 2016

Page 1: Personalization Palooza 2016

Personalizing Television.

#ppalooza - Personalization Palooza 2016 - NYC MEDIA LAB

Page 2: Personalization Palooza 2016

Screens User

Marketing & Editorial Team

ONE SIZEFITS ALL UX

Brand

Page 3: Personalization Palooza 2016

Dangers

Most of the audience does not find relevant content

with low effort.

When it’s time to optimize bills, a portion of subscribers doesn’t see

enough value.

Users don’t see the content offers suitable for them.

The screen is small and people don’t scroll or dig enough to

find what’s good for themLower ARPU

Lower retention

Users tend to ignore or even “mute” sources

that are not relevant.E-mail and mobile notifications

contain generic suggestionsLow ROI on marketing to

subscribers

At the end of the free trial most of the prospects didn’t see the full

value of the offeringThe free trial period provides a

generic user experienceLow conversion ratio of

free trials

One-size-fits-all Impact

Page 4: Personalization Palooza 2016

Screens User

Marketing & Editorial Team

ONE SIZEFITS ALL UX

Brand

Page 5: Personalization Palooza 2016

Screens User

Marketing & Editorial Team

UXUX Autopilot

ONE-TO-ONE

Actionable Analytics

Brand

Page 6: Personalization Palooza 2016

Marketing & Editorial Team

UX Autopilot

Actionable Analytics

EDITORIAL CURATION BUSINESS RULES

A/B TESTING

SELF-TUNING

Hints KPIs

UX

USAGE TRACKING

Page 7: Personalization Palooza 2016

ContentWise

Automate the Digital Storefront

Deliver a Personalized, One to One User Experience

Assist the Content Curation Process

Page 8: Personalization Palooza 2016

Personalization of TV & Video Services

Search, Discovery, Prediction ➜ UX Autopilot

Multi-catalog, Multi-language, Multi-screen

Analytics, A/B Testing, Metadata Management

Page 9: Personalization Palooza 2016

EFFECT: Widening the Catalog Coverage

Catalog portion watched by users

OTT Service

80%

42%

No Personalization

With ContentWise

Page 10: Personalization Palooza 2016

EFFECT: The Long Tail That Really Works

Playback Distribution

Content Assets

ContentWise uplift

Popular content

No Personalization

Playbacks

Page 11: Personalization Palooza 2016

? ? ? ? ? ?

HOW TO: Targeted Promotions

You have 20 movies to promotebut space on screen for 3 elements only.

You’d like to display the relevant ones to each user.

Page 12: Personalization Palooza 2016

Let’s say that we have 20 movies to promote but space on screen for only 3 elements.

We want to display the relevant ones for each user.

HOW TO: Targeted Promotions

Page 13: Personalization Palooza 2016

HOW TO: Next-to-play & Binge-viewing

Episode 5

Page 14: Personalization Palooza 2016

Episode 5Ep.6 Alternative

content

News 1 News 2 News 3 …

Prediction Discovery

HOW TO: Next-to-play & Binge-viewing

Page 15: Personalization Palooza 2016

Personalize Through NavigationContent in a flat list. No visual help to

process what’s on the screen.Collections and micro-genres:

easily scannable

Page 16: Personalization Palooza 2016

Surface a personalized set of collections

including content with one or more relevant

“features” (micro-genres)

Page 17: Personalization Palooza 2016

LIVE EVENT LIVE EVENT

EPG

AppsSports Highlights

LIVE EVENTLIVE EVENT

Cross-domain

Page 18: Personalization Palooza 2016

LIVE EVENT VIDEO CLIP APP LIVE EVENT

EPG

AppsSports Highlights

LIVE EVENTLIVE EVENT

Surfacing elements from other catalogs

Cross-domain

Page 19: Personalization Palooza 2016

Classic Recommender System

Collaborative Filtering Suggests content that has been relevant for other users with a viewing history similar to mine.

WHO LIKED THIS ALSO LIKED…

Tends to surface popular content Cannot suggest new additions (“cold start problem”)

Content-based Suggests content similar to what I watched in the past. Tends to stay confined in the user’s comfort zone Limits true catalog exploration

Page 20: Personalization Palooza 2016

Hybrid Algorithm Blends collaborative and content-based models to balancetaste-matching, popularity and serendipity.

New content items are assigned an initial score based on each user’s taste and content metadata. This puts them “in motion” and, if they are watched and become popular in certain audience segments, the collaborative component starts prevailing.

PROBLEM: new content is “cold”

SOLUTION

Page 21: Personalization Palooza 2016

12am 4am 8am 12pm 4pm 8pm

TV ScreenMobile

Contextual User Habits Learns user’s habits in the context of time, location and device. Predicts user’s intentions by surfacing content typically watched in the specific context.

For example: - around 3pm of a Sunday, on the living room TV - at 6pm of a Wednesday, on the smartphone, out of home Very effective for shared devices with no user login.

PROBLEM: Shared Devices Without Login

SOLUTION

Page 22: Personalization Palooza 2016

Semantic Enrichment and Knowledge Graph

Movie Episode Gossip Video

Talk Show Clip

spouse 2015..

spouse 2000..2005

Gossip Video

appearsIn appearsInactorOf appearsIn

Season

Series

Special

spinOff

appearsIn Channel BrandTalk Show

Brand

Movie

sequelOffranchise

James Bond

franchise

Schedule

interviewedIn

Page 23: Personalization Palooza 2016

UX Engine

Consumer UI

Admin Console

You are in control

One API for all client platforms

Page 24: Personalization Palooza 2016

Don’t fly blind: Analytics on UX PerformanceEXAMPLE: effect of a rule change on the relevant KPIs

Page 25: Personalization Palooza 2016

On-boarding Trial Training Retention

Convert to paying userSign-up 2nd payment Monthly

renewal

Cold start

WIP: Shift Gears as User Relationship Matures

Page 26: Personalization Palooza 2016

WIP: Think in Two Dimensions Vertical Layouts Personalized Selection

TOP PICKS FOR YOU

TRENDING SERIES

ACTION MOVIES

COMEDIES SET IN NEW YORK

SPY MOVIES BASED ON BOOKS

Alice

TOP PICKS FOR YOU

MOST VIEWED

NEW ARRIVALS

WHAT’S TRENDING

TRENDING SERIES

FAMILY MOVIE NIGHT

DS: <GENRE> MOVIES

DS: COMEDIES SET IN <CITY>

DS: <EDITORIAL COLLECTION>

WATCH IT AGAIN

BECAUSE YOU LIKED…

OSCAR WINNERS

ENABLED STREAMS VERTICAL LAYOUT

MANUALLY PINNED

ALGORITHMIC SELECTION

Page 27: Personalization Palooza 2016

THE FUTURE?Own your knowledge:

audience behavior & content performance

Know your users through the stories they like:

emotional traits lifestyle traits social traits

Own the ability to expose YOUR users to the

relevant brand messages

Page 28: Personalization Palooza 2016

Thank you!

www.contentwise.tv

Visit our website or contact us

[email protected]