Tore Sundelin - PowerPoint Presentation

Slide1
Slide2

Tore SundelinPrincipal Lead Program ManagerMicrosoft

The Personalized Inbox

USX204Slide3
Slide4

The Digital Deluge

Humanity created 2.8

zettabytes

(10

21

bytes) of digital data in 2012

Expected 2000% increase in global data by 202075% of all digital data created by individuals154 billion E-mails are sent each day

Source: “A Business Report – Big Data Gets Personal,” MIT Technology Review, May 2013.Slide5

The User Dilemma

The average IW spends almost 50% of his week managing email and searching for information

Source: International Data Corporation (IDC); McKinsey Global Institute Analysis

62% of IWs can’t use information effectively for decision-making

Source: “Three Gaps for Employee Productivity and What They Mean for IT,” CEB Survey, April 2013.Slide6

How We Can Help Slide7

New Mailbox Features Codename

Clutter – remove unimportant items from your inbox, help you focus on what matters

People View

– enable you to see when you have mail from key people and help you manage it

Personalized Search

– use your data to help you quickly find exactly what you are looking forSlide8

Powered by The Office Graph

Your data determines who and what is important to you

User is in control of smart features and their behaviors

Enterprise and user privacy is protectedSlide9

Purpose of this SessionSlide10

Jim Edelen Senior Program ManagerMicrosoft

De-Cluttering the InboxSlide11

Our Target Problem

Problem: People have stuff they don’t care about cluttering

up their inbox, distracting them from what’s important.Slide12

Value Proposition

1

st

2

nd

3rd Slide13

What is “Clutter”?

Mail you are highly likely not going to read

Patterns learned based on your personal behavior

Unlike junk, no universal view of a given sender or messageSlide14

User Experience Principles

Clutter items are not visible in default triage view

Clutter items do not “interrupt” user via notifications

Enable the user to train the system

Users can enable and disable the feature as they see fitSlide15

Codename Clutter DemoSlide16

How Clutter WorksSlide17

How it Works

Personalized Machine Learning Model

Uses

many message attributes and a couple user

actions

Model learned independently for each individual

Bayesian Probit Regression developed with MSR using Infer.NetYour actions teach the model

Implicitly from actions like read, replyExplicitly by moving items in/out of clutter groupSpecify important people who are never clutter Slide18

Mail Content

Actions

Analysis

Who

Sent It?

Was it high importance?

Are you on the to line?

Did you read it?

Did you reply to it?

Training

Learning about you

Classification

Applying that learning

Mail Content

Analysis

Who Sent It?

Was it high importance?

Are you

on the to line?

Will you treat the mail like clutter?

Learning Summary

(Weights)

Learning Summary

(Weights)

Clutter

or

Not Clutter

ExampleSlide19

How it Works

Determining Clutter

Transport agent classifies new items on delivery

Rules bypass clutter classification

Existing content is not

changed

LearningTime-based assistant periodically processes new information

Everything learned securely stored in user’s mailboxSlide20

Target Users

Information Workers

Consumers receive large volumes of well known “types”

At work

patterns are often unique to industry, company, and/or job

High Volume

Not for users who read everythingNot for users with comprehensive rulesUltimately – You Decide

System will proactively notify users it believes will benefitBut, all users are able to enable/disableSlide21

Feature Rollout and SupportSlide22

Krish GaliProgram Manager

People ViewSlide23

Email patterns have evolved, but solutions haven’t

Email volume keeps increasing YOY

Organizing and managing mail is a huge challenge

Communications are people-centricSlide24

Quick triage of emails

Always fresh and relevant

Introducing People View

At a glance view of what is in your InboxSlide25

People View DemoSlide26

How it works

Automatic curation of people

No end user effort in creating a list of ‘key’ people that becomes obsolete later

People are added and removed based on email communications

People are ranked based on frequency of outbound messages

Simple user experience

Dynamic ordering based on email activity

Fast – View messages faster than scanning and scrolling

It’s

a view not a folder – mails are not

copied or moved from InboxSlide27

Rollout schedule and FAQ

Q1 2014

Q2 2014

H

2 2014

Limited release in North America

Limited release to WW

Full WW release

Does Outlook support People View?

First release is OWA Only, Outlook support coming later.

When will on-prem customers get this?

- On-prem customers will get this feature as part of the next on-prem server release.Slide28

Kutlay TopatanProgram Manager

Personalized Search Slide29

New search experiences raised user expectations

Suggestions

Instant results

Relevant refinersSlide30

Users struggle

to find what they are looking for, re-searching frequently

Source: OXG Search Focus Groups

Pre-Organizing doesn’t improve search success

Too slow

Recall - Difficult

to remember

context

Refining

is

difficult & inefficient

PAIN POINTSSlide31

Introducing personalized search

Lightning fast results

that utilize the new indexing and query pipeline

Personalized suggestions

that help interpret user intentions

Content based refiners

that help finding best results with minimal effortHit highlighting to easily find relevant sections in long conversations Slide32

Search DemoSlide33

Come and see more

in-depth discussion about Search in Exchange

ARC.203

Wednesday 1:00-2:15 in Ballroom FSlide34

Go

to

the Pre-Release

Programs Booth

Tell

us about your Office 365 environment/or

on premises plans

Get selected to be in a programTry new features first and give us feedback!Start now at:http://prereleaseprograms-public.sharepoint.com/

Pre-Release Programs Team

Be first in

line

!Slide35
Slide36

©

2014

Microsoft Corporation. All rights reserved. Microsoft, Windows and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries.

The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.