The data driven evolution of a consumer product

Yuval Ariav, Chief Architect

@yuvalariav

THE DATA DRIVEN

EVOLUTION OF A CONSUMER

PRODUCT

Yuval Ariav, Chief Architect

@yuvalariav

Consumer mobile product

❯Millions of users (hopefully)

❯Taste becomes a quantifiable metric

❯Ecosystem provides implicit & explicit feedback

❯Cloud architecture (frequently)

❯Services easily become data intensive

❯Data-driven product decision making

❯State-of-mind during planning & execution

Yuval Ariav, Chief Architect

@yuvalariav

Example: A/B testing (1)

A B

18.6% 31.8%

+70.1%

Yuval Ariav, Chief Architect

@yuvalariav

Example: A/B testing (2)

A B

2.00% 1.00%

+102%

Yuval Ariav, Chief Architect

@yuvalariav

Onavo – overview

❯Putting smartphone users in control of

their mobile data usage

❯Make mobile data more transparent

❯Shrink mobile data volumes

❯Helping users save money on monthly bills

❯Serving users in 150 countries, 290 mobile

operators

Yuval Ariav, Chief Architect

@yuvalariav

2 MONTHS TO LAUNCH

Phase 1

Yuval Ariav, Chief Architect

@yuvalariav

Phase 1 – Overview

❯Beta program before launch

❯Test architecture

❯Evaluate compression rates

❯Build up a friendly user base

❯Bare minimum

❯Registration (e-mail)

❯Device configuration

❯Saving statistics

Yuval Ariav, Chief Architect

@yuvalariav

Registration and configuration

1. 2. 3.

Yuval Ariav, Chief Architect

@yuvalariav

Lessons learned – phase 1

❯It’s never too early to start A/B testing

❯Use Beta / testing phases to collect “product

intelligence”

❯Run an A/B-test-friendly operation

❯Plan information oriented architecture

❯Design a “test modular” product

❯Have a flexible BE to complement

Yuval Ariav, Chief Architect

@yuvalariav

LAUNCH +2WEEKS

Phase 2

Yuval Ariav, Chief Architect

@yuvalariav

Phase 2 – overview

❯Launch – 29/04/2011

❯Good reception, coverage

❯We’re out in the wild

❯Troubleshooting the world

❯Conversion pretty uniform, except…

❯Some countries… or operators?

❯Problems with installation

Yuval Ariav, Chief Architect

@yuvalariav

Testing installation screens (1)

84.2% 80.3% 83% 84.5%

A B C D

Yuval Ariav, Chief Architect

@yuvalariav

Testing installation screens (2)

49.5% 54% 47.7% 49.8%

A B C D

Yuval Ariav, Chief Architect

@yuvalariav

Lessons learned – phase 2

❯Money talks, but not always

❯DoV isn’t a substitute for clarity

❯Experiment with different conversion

approaches (walkthrough, goal, etc.)

❯Measure through

❯Tag users into cohorts ASAP

❯Plan your data for troubleshooting /

optimizing

Yuval Ariav, Chief Architect

@yuvalariav

LAUNCH +2MONTHS

Phase 3

Yuval Ariav, Chief Architect

@yuvalariav

Phase 3 – overview

❯A couple of months in

❯Growth based on mainly on PR

❯Popular blogs

❯Weekly / monthly infographics

❯User reviews

❯In-app mechanisms for sharing on social networks

❯Kvirality not high enough…

❯Looking to improve virality

Yuval Ariav, Chief Architect

@yuvalariav

Testing sharing mechanisms

Yuval Ariav, Chief Architect

@yuvalariav

Behind the scenes

❯Everything is metered

❯Who’s* sharing?

❯Who* with?

❯How effective?

❯When?

❯Couple numbers with service events

❯“User saved >50MB / $10 overall”

❯“User saw 90% savings on a single app”

Yuval Ariav, Chief Architect

@yuvalariav

Lessons learned – phase 3

❯2 types of shareable content

❯Content you “shout” about (share)

❯Content you “whisper” about (invite)

❯Sense of accomplishment / progress goes

a long way

❯Measure everything

❯Quantify sentiment / identify advocates

❯Use as an internal meter

Yuval Ariav, Chief Architect

@yuvalariav

MOVING TO A PAID MODEL

Phase 4

Yuval Ariav, Chief Architect

@yuvalariav

Phase 4 – overview

❯Evaluating business models

❯Key questions

❯How to frame the offering?

❯What payment model to use?

❯What charging model to use?

❯What to charge for the offering?

Yuval Ariav, Chief Architect

@yuvalariav

Framing the offering

❯“Raw” technological offering

❯50% optimization of mobile apps’ data

❯2 distinct product messages

❯“Pay less”❯“Onavo helps you save 50% on your data”

❯“Do more”❯“Onavo lets you double your data plan”

❯Reaching out to potential users

Yuval Ariav, Chief Architect

@yuvalariav

Testing product offering

“Pay less”

“Do more”

Control

Yuval Ariav, Chief Architect

@yuvalariav

Picking the model

❯On top of product messaging

❯Payment model❯Paid app?

❯Freemium?

❯Trial?

❯Charging model❯Price tiers

❯Pay per X

❯Renewable subscription

❯Again – reaching out

Yuval Ariav, Chief Architect

@yuvalariav

Testing potential models

How much would you be willing to pay per month to get more out of you data plan?1. $0.992. $1.993. $2.994. $3.995. $4.99

Yuval Ariav, Chief Architect

@yuvalariav

Testing potential models

Would you be willing to pay a $2.99 monthly subscription to get more out of your data plan?1. Yes2. No

Would you be willing to pay at all?1. Yes2. No

How much would you be willing to pay?…

no

yesyes

no

Yuval Ariav, Chief Architect

@yuvalariav

Lessons learned – phase 4

❯The answers you get are only as good as

the questions you ask

❯Use neutral wording

❯Be accurate

❯Less is more

❯Remember attention span is limited

❯Consider cohort sizes

❯Be creative about reaching out

Yuval Ariav, Chief Architect

@yuvalariav

Lessons learned - overall

❯ Get in a data-oriented-decision state of mind

❯ Design your operation around making informed decision

❯ Plan ahead, data-wise

❯ Learn from others’ product decisions, but don’t

generalize

❯ Product decisions are often not optimal when taken out of

context

❯ A/B testing will make your business excellent, but it

won’t make it a business

❯ Doesn’t substitute business plan, marketing, etc.

Yuval Ariav, Chief Architect

@yuvalariav

THE DATA DRIVEN

EVOLUTION OF A CONSUMER

PRODUCT