SlideShare a Scribd company logo
Audience segments
Technical aspects of audience targeting in DSP
What is audience segment
- named set of device ids or cookies
- ids are any advertising device ids or its hashes, IDFA, UDID, AAID, Android
Device ID, MD5, SHA1
- can be self-gathered (1st-party), or provided by DMP (3rd-party)
“In-Market --> Autos --> Makes & Models --> Audi --> A4”
idfa:6630580C-4347-4FFF-8AEB-19530C143800
idfa:4C34D479-2F53-4673-B119-5F9AF0FE6BB7
aaid:3b93df69-8b57-4b64-a874-44eb37e3312c
aaid_md5:134c1ca7d0abc2f5c0b046b81f7941a3
Taxonomy
Some numbers
Total:
- segment size up to billion device ids (109
)
- up to 200000 segments per DMP (105
)
- total IDs set contains more than 7 billion device ids (7*109
)
- gzipped segments content size >75Tb
- several DMPs: BlueKai, Lotame, Mobext, Statiq, etc
Used in active campaigns:
- segments count ~200
- total unique ids count ~1*109
- data size ~500Gb
Requirements
- support big sizes and counts
- reply in 20ms (100ms for whole bidding cycle)
- support multiple datacenters. Bidders are spread over multiple DCs, ADB
instances should be local to keep latency, require full replication
- integration with many DMPs
- regular updates
- short update cycle for self-gathered segments, it should be available during
gathering
Solutions
- kvs-based precise solution
- Bloom-filter based probabilistic solution
KVS-based
- fast real-time storage only for active segments
- slow storage for all segments
- controller that upload data from slow storage
Datacenter US
External
DMPs
External
DMPs
Fast
storage
ADB
controller
Campaign
server
External
DMPs
Postgres S3
Bidder
commands/status
upload
segments
data
Datacenter EU
Datacenter APAC
targeting periods bid
request
KV real-time storage
KV storage issues
- costs. Slow s3 storage >$2k monthly, real-time database >$5k per datacenter
- upload bandwidth is limited and shared with lookups. Lookups 100k/s, upload
30k id/s
Alternative solution: Bloom-filters
- real-time storage is replaced by set of Bloom-filters
- slow storage is the same
- controller create Bloom-filter instead of uploading to real-time storage, resolve
issue with uploading bandwidth
- real-time storage is calculation cluster that check requested ids against all
B-filters
- each segment require Bloom-filter size from 10Mb to 200Mb, false-positive
error rate 0.5%
- cost for one DC is ~$1k
Bloom-filters solution
Datacenter US
B-filter
host 1
ADB
controller
S3,
b-filters
Bidder
commands/status
Datacenter EU
Datacenter APAC
bid
request
S3, data
B-filter
host 2
b-filters
sharding
Cons
- there are no ready solutions which host Bloom-filters and support sharding and
replication
- it is probabilistic and can’t be used for strict segments
- no strict support for cross-device
Q&A
Thank you for attention!

More Related Content

PDF
Портфолио юзабилити услуг
PDF
RTB. Experience in product launch by Dmitrii Kasimtsev Tech Hangout #6
PDF
Finding Your Audience Through Market Segmentation
PPTX
Millennials Don't Suck
PPTX
Moodboard
PPTX
Models of audience segmentation
PPTX
Yield management: от простого к сложному by Kostyantyn Derets' TechHangout #6
PDF
Bbc global audience segmentation
Портфолио юзабилити услуг
RTB. Experience in product launch by Dmitrii Kasimtsev Tech Hangout #6
Finding Your Audience Through Market Segmentation
Millennials Don't Suck
Moodboard
Models of audience segmentation
Yield management: от простого к сложному by Kostyantyn Derets' TechHangout #6
Bbc global audience segmentation

Similar to Audience segments. Technical aspects of audience targeting in DSP by Ivan Michailov TechHangout #6 (20)

PDF
STOR2RRD for storage edition - white paper
PDF
Ids sdd-jlr manual 02 02-12 (1)
PPTX
Industry’s performance leading ultra low-power dsp solution
PPT
BWC Supercomputing 2008 Presentation
PDF
Kauli SSPにおけるVyOSの導入事例
PDF
Internet of Things: Vehicular Tracking System
PPT
sector-sphere
PPT
Sector Sphere 2009
PPTX
Cloud Storage Introduction ( CEPH )
PDF
“Programming Vision Pipelines on AMD’s AI Engines,” a Presentation from AMD
PDF
Anatomy of ROCgdb presentation at gcc cauldron 2022
PDF
AWS re:Invent re:Cap - 데이터 분석: Amazon EC2 C4 Instance + Amazon EBS - 김일호
PPT
MongoDB Sharding Webinar 2014
PDF
RAD Industrial Automation, Labs, and Instrumentation
PPTX
Mirabilis_Presentation_DAC_June_2024.pptx
PPTX
NoSQL meetup July 2011
PDF
What's new in MariaDB AX webinar
PDF
Android Development Tools
PDF
5.2 QRadar_Architecture_-_General123.pdf
PPTX
Webinar: Deploying MongoDB to Production in Data Centers and the Cloud
STOR2RRD for storage edition - white paper
Ids sdd-jlr manual 02 02-12 (1)
Industry’s performance leading ultra low-power dsp solution
BWC Supercomputing 2008 Presentation
Kauli SSPにおけるVyOSの導入事例
Internet of Things: Vehicular Tracking System
sector-sphere
Sector Sphere 2009
Cloud Storage Introduction ( CEPH )
“Programming Vision Pipelines on AMD’s AI Engines,” a Presentation from AMD
Anatomy of ROCgdb presentation at gcc cauldron 2022
AWS re:Invent re:Cap - 데이터 분석: Amazon EC2 C4 Instance + Amazon EBS - 김일호
MongoDB Sharding Webinar 2014
RAD Industrial Automation, Labs, and Instrumentation
Mirabilis_Presentation_DAC_June_2024.pptx
NoSQL meetup July 2011
What's new in MariaDB AX webinar
Android Development Tools
5.2 QRadar_Architecture_-_General123.pdf
Webinar: Deploying MongoDB to Production in Data Centers and the Cloud
Ad

More from Innovecs (20)

PDF
Building Efficient and High Performing iLottery Solutions
PDF
Innovecs Meetup Lifestory
PPTX
Подходы и технологии в React Redux
PPTX
Redux vs RxJS vs Mobx в связке с React
PDF
React & Redux (Lazarev)
PDF
Web Platform for Fashion Shop
PDF
Programmatic Advertising Platform
PDF
Multimedia Newsroom
PDF
Media Buying Platform (DSP+DPM)
PDF
Web-based Shipment Application
PDF
Digital Trading Platform
PDF
Mobile Insurance Agent
PDF
Online Learning Platform
PDF
Client Bank
PDF
Fertility Tracking App
PDF
Warranty Wallet App
PDF
Online Bingo Game
PDF
Secure Messenger
PDF
Search Data Platform
PDF
Website Builder for Insurance Agents
Building Efficient and High Performing iLottery Solutions
Innovecs Meetup Lifestory
Подходы и технологии в React Redux
Redux vs RxJS vs Mobx в связке с React
React & Redux (Lazarev)
Web Platform for Fashion Shop
Programmatic Advertising Platform
Multimedia Newsroom
Media Buying Platform (DSP+DPM)
Web-based Shipment Application
Digital Trading Platform
Mobile Insurance Agent
Online Learning Platform
Client Bank
Fertility Tracking App
Warranty Wallet App
Online Bingo Game
Secure Messenger
Search Data Platform
Website Builder for Insurance Agents
Ad

Recently uploaded (20)

PDF
Approach and Philosophy of On baking technology
PDF
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
PDF
Unlocking AI with Model Context Protocol (MCP)
PDF
Dropbox Q2 2025 Financial Results & Investor Presentation
PDF
Chapter 3 Spatial Domain Image Processing.pdf
PPTX
Cloud computing and distributed systems.
PDF
NewMind AI Weekly Chronicles - August'25 Week I
PDF
MIND Revenue Release Quarter 2 2025 Press Release
PDF
Encapsulation theory and applications.pdf
PPTX
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
PPTX
Effective Security Operations Center (SOC) A Modern, Strategic, and Threat-In...
DOCX
The AUB Centre for AI in Media Proposal.docx
PPT
“AI and Expert System Decision Support & Business Intelligence Systems”
PDF
Building Integrated photovoltaic BIPV_UPV.pdf
PDF
Machine learning based COVID-19 study performance prediction
PDF
KodekX | Application Modernization Development
PPTX
20250228 LYD VKU AI Blended-Learning.pptx
PDF
cuic standard and advanced reporting.pdf
PDF
Peak of Data & AI Encore- AI for Metadata and Smarter Workflows
PDF
Diabetes mellitus diagnosis method based random forest with bat algorithm
Approach and Philosophy of On baking technology
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
Unlocking AI with Model Context Protocol (MCP)
Dropbox Q2 2025 Financial Results & Investor Presentation
Chapter 3 Spatial Domain Image Processing.pdf
Cloud computing and distributed systems.
NewMind AI Weekly Chronicles - August'25 Week I
MIND Revenue Release Quarter 2 2025 Press Release
Encapsulation theory and applications.pdf
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
Effective Security Operations Center (SOC) A Modern, Strategic, and Threat-In...
The AUB Centre for AI in Media Proposal.docx
“AI and Expert System Decision Support & Business Intelligence Systems”
Building Integrated photovoltaic BIPV_UPV.pdf
Machine learning based COVID-19 study performance prediction
KodekX | Application Modernization Development
20250228 LYD VKU AI Blended-Learning.pptx
cuic standard and advanced reporting.pdf
Peak of Data & AI Encore- AI for Metadata and Smarter Workflows
Diabetes mellitus diagnosis method based random forest with bat algorithm

Audience segments. Technical aspects of audience targeting in DSP by Ivan Michailov TechHangout #6

  • 1. Audience segments Technical aspects of audience targeting in DSP
  • 2. What is audience segment - named set of device ids or cookies - ids are any advertising device ids or its hashes, IDFA, UDID, AAID, Android Device ID, MD5, SHA1 - can be self-gathered (1st-party), or provided by DMP (3rd-party) “In-Market --> Autos --> Makes & Models --> Audi --> A4” idfa:6630580C-4347-4FFF-8AEB-19530C143800 idfa:4C34D479-2F53-4673-B119-5F9AF0FE6BB7 aaid:3b93df69-8b57-4b64-a874-44eb37e3312c aaid_md5:134c1ca7d0abc2f5c0b046b81f7941a3
  • 4. Some numbers Total: - segment size up to billion device ids (109 ) - up to 200000 segments per DMP (105 ) - total IDs set contains more than 7 billion device ids (7*109 ) - gzipped segments content size >75Tb - several DMPs: BlueKai, Lotame, Mobext, Statiq, etc Used in active campaigns: - segments count ~200 - total unique ids count ~1*109 - data size ~500Gb
  • 5. Requirements - support big sizes and counts - reply in 20ms (100ms for whole bidding cycle) - support multiple datacenters. Bidders are spread over multiple DCs, ADB instances should be local to keep latency, require full replication - integration with many DMPs - regular updates - short update cycle for self-gathered segments, it should be available during gathering
  • 6. Solutions - kvs-based precise solution - Bloom-filter based probabilistic solution
  • 7. KVS-based - fast real-time storage only for active segments - slow storage for all segments - controller that upload data from slow storage Datacenter US External DMPs External DMPs Fast storage ADB controller Campaign server External DMPs Postgres S3 Bidder commands/status upload segments data Datacenter EU Datacenter APAC targeting periods bid request
  • 9. KV storage issues - costs. Slow s3 storage >$2k monthly, real-time database >$5k per datacenter - upload bandwidth is limited and shared with lookups. Lookups 100k/s, upload 30k id/s
  • 10. Alternative solution: Bloom-filters - real-time storage is replaced by set of Bloom-filters - slow storage is the same - controller create Bloom-filter instead of uploading to real-time storage, resolve issue with uploading bandwidth - real-time storage is calculation cluster that check requested ids against all B-filters - each segment require Bloom-filter size from 10Mb to 200Mb, false-positive error rate 0.5% - cost for one DC is ~$1k
  • 11. Bloom-filters solution Datacenter US B-filter host 1 ADB controller S3, b-filters Bidder commands/status Datacenter EU Datacenter APAC bid request S3, data B-filter host 2 b-filters sharding
  • 12. Cons - there are no ready solutions which host Bloom-filters and support sharding and replication - it is probabilistic and can’t be used for strict segments - no strict support for cross-device
  • 13. Q&A Thank you for attention!