Maximise the value of app users with predictive analytics

Cubitic’s suite of behavioural prediction algorithms tell you what your users will

do, before they do it

…allowing you to proactively drive engagement and monetisation.

COST OF ACQUISITION IS RISING

CPI up by 84% since January 2012

Harder to rely on acquisition to drive game profit

Rising importance of maximizing value of existing users

Source: Venture Beat & Fiksu.com, 31 December 2014.

$-

$0.50

$1.00

$1.50

$2.00

$2.50

Jan-

12

Mar

-12

May

-12

Jul-1

2 S

ep-1

2 N

ov-1

2 Ja

n-13

M

ar-1

3 M

ay-1

3 Ju

l-13

Sep

-13

Nov

-13

Jan-

14

Mar

-14

May

-14

Jul-1

4 S

ep-1

4 N

ov-1

4

Cost Per Loyal User Index iOS

THE TRADITIONAL APPROACH

Segment users by historic behaviour

Define desired outcome

Targeted intervention

Strategy used for CRM, paid re-targeting, tailored game experience

This is a reactive approach

Cubitic’s suite of predictive analytics include…

PROACTIVE APPROACH

Use Predictive Analytics to anticipate user behaviour

Act ahead-of-time to influence behaviour

Increase effectiveness & ROI of marketing

spend

Which users will stop playing?

Churn prediction

Which users will convert?

Conversion prediction

How much will users spend?

Spend prediction

EXAMPLE – BRIAN’S GAME!

Brian has a small iOS game with around 1,000 daily installs and 50,000 DAU

The game is doing well but he thinks retention could be higher

0%

20%

40%

60%

80%

100%

0 5 10 15 20 25 30 Day After Install

RETENTION

£0.00

£0.10

£0.20

£0.30

£0.40

0 5 10 15 20 25 30

Day After Install

REVENUE PER INSTALL How can we use predictive analytics to

improve retention?

BRIAN’S CURRENT APPROACH

With no intervention, on a given day 20% of DAU will not return to the game

Brian is sending a push message to users inactive for 3 days (40% of DAU)

Reduced churn to 15% of DAU

Cost Per Impression Users targeted Total Cost Churn

Prevented

Traditional Approach $0.001 20,000

(users inactive 3 days) $20 25%

BRIAN’S CURRENT APPROACH

20,000 Inactive for 3 days

30,000 Active

7,500 Churned

42,500 Active

now + 3 days future

All Users

50,000

push msg

15% of DAU

BRIAN’S CURRENT APPROACH Problems with this approach?

1.  Not the right time

2.  Not the right users

3.  Not the right message

PROACTIVE PREDICTIVE APPROACH

SEGMENTATION BASED ON PREDICTED BEHAVIOUR

Transient Abandoning

Recruit Veteran

Early Player

Established Player

Predicted Churners

Predicted Returners

This can be incorporated into existing segmentations or

be used as basis for new segmentations

TARGETED INTERVENTION For example: CHURN PREVENTION

push message

in-game offer

Transient Abandoning Predicted Churners

Target each segment with relevant message!

Early Player

Established Player

PROACTIVE PREDICTIVE APPROACH

5,000 Churned

45,000 Active

now future

All Users

50,000

push message

in-game offer

transient

abandoning

10% of DAU

AB TESTING

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 5 10 15 20 25 30

Days After Install

Old Reactive Approach (A) New Predictive Approach (B)

SUMMARY 1.  Churning users are targeted early (right time!) 2.  Only users likely to leave are targeted (right users!) 3.  Users receive a relevant message (right message!)

Cost Per Impression Users targeted Total Cost Churn

Prevented

Traditional Approach $0.001 20,000

(users inactive 3 days)

$20 25%

Predictive Approach $0.001 10,000

(churning users)

$10 50%

QUESTIONS

Thank you for your time!

Get in touch

www.cubitic.io

[email protected]

Follow us! @ryan_does_data

@cubitic