Cross-Device Tracking for Better Engagement

Welcome!
 
What is Cross-Device Tracking?
Presentation by Justin Brookman
Policy Director
Office of Technology Research and Investigation
FTC
What I’m Going to Cover
Background
Probabilistic vs Deterministic Matching Models
Snapshot of Top Sites
Open Questions
Traditional Third-Party Behavioral Tracking
 
Advertising Network
cookie=4qasr4sdf1
cookie=4qasr4sdf1
cookie=4qasr4sdf
1
The Limitations of Third-Party Cookies
Fragile
Browser-specific
Not traditionally tied to personally-identifiable
information
Often no direct relationship to get PII
Privacy concerns
DoubleClick-Abacus merger in 2001
PII (name, email)/non-PII (cookies, urls) distinction
still reflected in a lot of practices, privacy policies
More devices, smarter devices
   
Reasons why you might want engage in
cross-device tracking
Targeting
Security
Analytics
Measurement
Attribution
Reasons why you might want to engage in
cross-device tracking
Attribution
Probabilistic vs Deterministic
Probabilistic: Based on inferences about likely
connections between devices or users
Deterministic: Tying multiple devices to
persistent unique identifier
Probabilistic Matching
cookie=4qasr4sdf1
cookie=f52dh64dhq
Android Advertising
Id=0436732361
Probabilistic Matching
IP address:
23.64.176.179
(early mornings,
evenings, weekends)
IP address:
164.62.9.0
(9am-6pm weekdays)
IP address:
164.62.9.0
(9am-6pm weekdays)
Cellular network
23.64.176.179
(early mornings,
evenings, weekends)
Probabilistic Matching
Work??
Cell??
Home??
80%
80%
IP address:
164.62.9.0
(9am-6pm weekdays)
IP address:
164.62.9.0
(9am-6pm weekdays)
Cellular network
23.64.176.179
(early mornings,
evenings, weekends)
IP address:
23.64.176.179
(early mornings,
evenings, weekends)
Probabilistic Matching
Work?
Cell?
Home?
Location:
38.883914, -
77.020997
Weekday
location:
38.883914, -
77.020997
Evening location:
38.897634, -
77.036544
Location:
38.897634, -
77.036544
95%
95%
Probabilistic Matching
Work
Cell
Home
Technology news
UVa sports
Capitol Hill
Arsenal football
Technology news
UVa sports
Capitol Hill
Arsenal football
Technology news
UVa sports
Capitol Hill
Arsenal football
98%
98%
cookie=4qasr4sdf1
Android Advertising
Id=0436732361
cookie=f52dh64dhq
Device Graph
id=4qasr4sdf1
Android Advertising
Id=0436732361
id=f52dh64dhq
Deterministic: Log-in PLUS Broad Reach
You log onto a lot of services (e.g., social
networking, email) on different devices
BUT, some of those services also provide
functionality on a lot of other websites too
They have visibility into your behavior on those other
services, and can create a broad cross-device profile
about you
 
 
 
 
 
 
First-Party Deterministic Matching
First-Party Deterministic Matching
Login:
JustinBrookman
Login:
JustinBrookman
Login:
JustinBrookman
First-Party Deterministic Matching
Login:
JustinBrookman
Login:
JustinBrookman
Login:
JustinBrookman
Third-party
sites/apps that
embed first-party
Third-party
sites/apps that
embed first-party
Third-party
sites/apps that
embed first-party
Deterministic: Partnering with Log-in Sites
What if I don’t get log-in data?
Look for partnership with publisher who does
BUT trepidation about sharing PII . . . so
identifiers often shared in hashed form
Use an algorithm to consistently convert an identifier
into a different, pseudo-random identifier
E.g., justin@domain.com 
b16f55bbe0ff554fb40003f8e5f96b99 (md5 hash)
Matching through hashed identifiers
Advertising Network
cookie=4qasr4sdf1
IDFA=qpcm12d5a7
Device Id=038573421
Matching through hashed identifiers
Advertising Network
md5=b16f55bbe0ff554fb4
0003f8e5f96b99
Log onto news site
as:
justin@domain.com
Log onto shopping
site as:
justin@domain.com
Log onto video
service as:
justin@domain.com
md5=b16f55bbe0ff554fb4
0003f8e5f96b99
md5=b16f55bbe0ff554fb4
0003f8e5f96b99
cookie=4qasr4sdf1
IDFA=qpcm12d5a7
Device Id=038573421
Device Graph
Advertising Network
cookie=4qasr4sdf1
IDFA=qpcm12d5a7
Device Id=038573421
Use of Email to Enable Cross-Device
Tracking
Purchase item at a
shopping site as
justin@domain.com
Use of Email to Enable Cross-Device
Tracking
Purchase item at a
shopping site as
justin@domain.com
Click on email
from shopping
site
Open email
from shopping
site
Android Advertising
Id=0436732361
cookie=4qasr4sdf1
cookie=a035fs35fm
Use of Email to Enable Cross-Device
Tracking
Purchase item at a
shopping site as
justin@domain.com
Click on email
from shopping
site
Open email
from shopping
site
Advertising Network
md5=b16f55bbe0ff554fb40003f8e5f96b99
Is Hashed PII Still PII?
Provides a layer of protection, but doesn’t
protect against all threats
Ed Felten’s blog post: “Does Hashing Make Data
Anonymous?”
Would we prefer tracking by unhashed PII?
Variations of Probabilistic and Deterministic
Cross-Device Tracking
Blended models
Deterministic company may extend to include
probabilistic data for clients who prefer reach to
certainty
Probabilistic companies may partner with
deterministic companies to verify accuracy of
algorithm, obtain data sets for modeling
Sharing device graphs
Leasing device graphs through cookie syncing
FTC Snapshot
Looked at top 20 sites in Alexa categories for
News, Sports, Shopping, Games, and Reference
Cross-device tracking companies on a considerable
majority of sites
Over 90% can collect PII from users
Numerous instances of emails, hashed emails sent to third
parties
Few privacy policies discuss
How much transparency?
Self-regulatory codes
Whose responsibility?
Does it depend on the model?
AdChoices icon
How achieve transparency for IoT devices?
How much control?
   
How much control?
   
How much control?
   
How much control?
What should you be able to control?
Targeting?
Collection/sharing?
Rise of ad blocking?
Panel 1: A Technological Perspective
Ashkan Soltani
 
Chief Technologist, FTC
Joseph Lorenzo Hall
Chief Technologist and Director of the Internet Architecture Project,
Center for Democracy & Technology
Jonathan Mayer
PhD Candidate, Computer Science, Stanford University
Andrew Sudbury
Co-founder and CTO, Abine, Inc.
Jurgen J. Van Staden
Director of Policy, Network Advertising Initiative
Panel 2: Policy Perspectives
Megan Cox
 
Attorney, FTC
Genie Barton
 
Vice President and Director, Online Interest-Based Advertising Accountability Program,
 
Council of Better Business Bureaus
Leigh Freund
 
President and CEO, Network Advertising Initiative
Jason Kint
 
CEO, Digital Content Next
Laura Moy
 
Senior Policy Counsel, Open Technology Institute, New America
Joseph Turow
 
Professor, Annenberg School for Communication, University of Pennsylvania
Slide Note
Embed
Share

Delve into the world of cross-device tracking with insights on probabilistic vs. deterministic matching models, limitations of third-party cookies, reasons to engage in cross-device tracking, and the distinctions between probabilistic and deterministic matching methods. Explore how tracking across multiple devices can enhance targeting, security, analytics, measurement, and attribution in the digital landscape.

  • Cross-Device Tracking
  • Probabilistic Matching
  • Deterministic Matching
  • Digital Advertising
  • User Engagement

Uploaded on Oct 03, 2024 | 0 Views


Download Presentation

Please find below an Image/Link to download the presentation.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.

E N D

Presentation Transcript


  1. Welcome!

  2. What is Cross-Device Tracking? Presentation by Justin Brookman Policy Director Office of Technology Research and Investigation FTC

  3. What Im Going to Cover Background Probabilistic vs Deterministic Matching Models Snapshot of Top Sites Open Questions

  4. Traditional Third-Party Behavioral Tracking Advertising Network cookie=4qasr4sdf1 cookie=4qasr4sdf1 cookie=4qasr4sdf 1

  5. The Limitations of Third-Party Cookies Fragile Browser-specific Not traditionally tied to personally-identifiable information Often no direct relationship to get PII Privacy concerns DoubleClick-Abacus merger in 2001 PII (name, email)/non-PII (cookies, urls) distinction still reflected in a lot of practices, privacy policies

  6. More devices, smarter devices

  7. Reasons why you might want engage in cross-device tracking Targeting Security Analytics Measurement Attribution

  8. Reasons why you might want to engage in cross-device tracking Attribution Image result for wonder bread

  9. Probabilistic vs Deterministic Probabilistic: Based on inferences about likely connections between devices or users Deterministic: Tying multiple devices to persistent unique identifier

  10. Probabilistic Matching cookie=4qasr4sdf1 Android Advertising Id=0436732361 cookie=f52dh64dhq

  11. Probabilistic Matching IP address: 164.62.9.0 (9am-6pm weekdays) IP address: 23.64.176.179 (early mornings, evenings, weekends) IP address: 164.62.9.0 (9am-6pm weekdays) Cellular network 23.64.176.179 (early mornings, evenings, weekends)

  12. Probabilistic Matching 80% 80% Work?? Cell?? Home?? IP address: 164.62.9.0 (9am-6pm weekdays) IP address: 23.64.176.179 (early mornings, evenings, weekends) IP address: 164.62.9.0 (9am-6pm weekdays) Cellular network 23.64.176.179 (early mornings, evenings, weekends)

  13. Probabilistic Matching 95% 95% Work? Cell? Home? Weekday location: 38.883914, - 77.020997 Location: 38.883914, - 77.020997 Location: 38.897634, - 77.036544 Evening location: 38.897634, - 77.036544

  14. Probabilistic Matching 98% 98% Work Cell Home Android Advertising Id=0436732361 cookie=f52dh64dhq cookie=4qasr4sdf1 Technology news UVa sports Capitol Hill Arsenal football Technology news UVa sports Capitol Hill Arsenal football Technology news UVa sports Capitol Hill Arsenal football

  15. Device Graph id=4qasr4sdf1 Android Advertising Id=0436732361 id=f52dh64dhq

  16. Deterministic: Log-in PLUS Broad Reach You log onto a lot of services (e.g., social networking, email) on different devices BUT, some of those services also provide functionality on a lot of other websites too They have visibility into your behavior on those other services, and can create a broad cross-device profile about you

  17. First-Party Deterministic Matching

  18. First-Party Deterministic Matching Login: Login: Login: JustinBrookman JustinBrookman JustinBrookman

  19. First-Party Deterministic Matching Login: Login: Login: JustinBrookman JustinBrookman JustinBrookman Third-party sites/apps that embed first-party Third-party sites/apps that embed first-party Third-party sites/apps that embed first-party

  20. Deterministic: Partnering with Log-in Sites What if I don t get log-in data? Look for partnership with publisher who does BUT trepidation about sharing PII . . . so identifiers often shared in hashed form Use an algorithm to consistently convert an identifier into a different, pseudo-random identifier E.g., justin@domain.com b16f55bbe0ff554fb40003f8e5f96b99 (md5 hash)

  21. Matching through hashed identifiers Device Id=038573421 IDFA=qpcm12d5a7 cookie=4qasr4sdf1 Advertising Network

  22. Matching through hashed identifiers Log onto video service as: justin@domain.com Log onto shopping site as: justin@domain.com Log onto news site as: justin@domain.com IDFA=qpcm12d5a7 Device Id=038573421 cookie=4qasr4sdf1 md5=b16f55bbe0ff554fb4 0003f8e5f96b99 md5=b16f55bbe0ff554fb4 0003f8e5f96b99 md5=b16f55bbe0ff554fb4 0003f8e5f96b99 Advertising Network

  23. Device Graph Device Id=038573421 IDFA=qpcm12d5a7 cookie=4qasr4sdf1 Advertising Network

  24. Use of Email to Enable Cross-Device Tracking Purchase item at a shopping site as justin@domain.com

  25. Use of Email to Enable Cross-Device Tracking Android Advertising Id=0436732361 cookie=a035fs35fm cookie=4qasr4sdf1 Purchase item at a shopping site as justin@domain.com Click on email from shopping site Open email from shopping site

  26. Use of Email to Enable Cross-Device Tracking Purchase item at a shopping site as justin@domain.com Click on email from shopping site Open email from shopping site md5=b16f55bbe0ff554fb40003f8e5f96b99 Advertising Network

  27. Is Hashed PII Still PII? Provides a layer of protection, but doesn t protect against all threats Ed Felten sblog post: Does Hashing Make Data Anonymous? Would we prefer tracking by unhashed PII?

  28. Variations of Probabilistic and Deterministic Cross-Device Tracking Blended models Deterministic company may extend to include probabilistic data for clients who prefer reach to certainty Probabilistic companies may partner with deterministic companies to verify accuracy of algorithm, obtain data sets for modeling Sharing device graphs Leasing device graphs through cookie syncing

  29. FTC Snapshot Looked at top 20 sites in Alexa categories for News, Sports, Shopping, Games, and Reference Cross-device tracking companies on a considerable majority of sites Over 90% can collect PII from users Numerous instances of emails, hashed emails sent to third parties Few privacy policies discuss

  30. How much transparency? Self-regulatory codes Whose responsibility? Does it depend on the model? AdChoices icon How achieve transparency for IoT devices?

  31. How much control?

  32. How much control?

  33. How much control?

  34. How much control? What should you be able to control? Targeting? Collection/sharing? Rise of ad blocking?

  35. Panel 1: A Technological Perspective Ashkan Soltani Chief Technologist, FTC Joseph Lorenzo Hall Chief Technologist and Director of the Internet Architecture Project, Center for Democracy & Technology Jonathan Mayer PhD Candidate, Computer Science, Stanford University Andrew Sudbury Co-founder and CTO, Abine, Inc. Jurgen J. Van Staden Director of Policy, Network Advertising Initiative

  36. Panel 2: Policy Perspectives Megan Cox Attorney, FTC Genie Barton Vice President and Director, Online Interest-Based Advertising Accountability Program, Council of Better Business Bureaus Leigh Freund President and CEO, Network Advertising Initiative Jason Kint CEO, Digital Content Next Laura Moy Senior Policy Counsel, Open Technology Institute, New America Joseph Turow Professor, Annenberg School for Communication, University of Pennsylvania

Related


More Related Content

giItT1WQy@!-/#giItT1WQy@!-/#giItT1WQy@!-/#