#MITXData 2015 - Flash Talk: Innovation for a Post-Cookie World

Flash Talk: Innovation for a Post-Cookie World

Brendan Riordan-ButterworthDirector, Technical StandardsIAB’s Tech Lab@iab

Innovation for a Post-Cookie World• What is “State Management”• State Management Techniques– Including examples

• Questions to Ask

After 5 Years at MicrosoftDuring which I did have a home.

WHAT IS STATE MANAGEMENT?

What is “State Management”?• An engineering term. • The problem cookies were designed to solve. • Any technique for reasonably re-identifying a

client over multiple interactions. • What the Federal Government does.

The Future of the Cookie• Additional reading on State Management and

proposed principles is available in IAB’s The Future of the Cookie whitepaper

• IAB.net/futureofcookie/

STATE MANAGEMENT TECHNIQUES

State Management TechniquesDeclarative• Authentication• Identification

Inferred• Location• Behavior• Device Attributes

Declarative - Authentication• User name and password– Secure tokens associated with these. – Derivative identification based on these.

• Limited set of companies that can issue it.

Declarative - Identification• Widely and successfully used. • Big press from AmEx on this in March– Partners include Macy’s, Rite Aid, Hulu

• Offline is the primary experience.

Inferred - Location• Two things are in the same location, and

maybe interacting. • Remember Bump?• Perhaps use location-

specific QR codes?

Inferred - Behavior• Algorithmic determination that behavior

across contexts is sufficiently similar. • Refined Lookalike Modeling.

Inferred - Device Attributes• The Panopticlick project from EFF showed that

a web browser likely has a rather rare configuration.

• Similar techniques can be used to discover when multiple apps are running on the same device.

Inferred - Other• There are other ways of discovering that the

same user is interacting with you in a different context.

• And other ways of discovering that similar users are available in other contexts.

QUESTIONS TO ASK

Accuracy• Does accuracy matter? If you reach a similar

human with an appropriate message, is that OK?

• How do you account for inaccuracy in metrics? Consider that advertising thrived for years with imperfect, extrapolated metrics.

Technology• What technologies suit your needs and your

relationship with the consumer? • Can consumer-initiated engagements be

tweaked to allow them to self-identify? • Does technology allow the bridging of digital

and non-digital campaign measurement?

Privacy and Choice• When you’re dealing with probabilistic re-

identification, how do you handle opt-out? What other controls need to be made available to consumers?

• How do you know if a technique you’re using will remain available in the future?

CLOSING