HeaderBidding From Hack to Holistic Approach Presented by

Header-Bidding: From Hack to Holistic Approach Presented by Emry Downing. Hall General Manager

§ Why this tech has grown so quickly § Header-Bidding auction mechanics § Latency and how to approach it § Price flooring and why it’s critical § Wrappers § Best practices: level up § What’s next for header-bidding?

ü Header-Bidding Tagless Ad-Tech

100, 000 STUDENTS WORLDWIDE USE OUR FAMILY OF WEBSITES AS READING, WRITING AND CRITICAL THINKING SOLUTIONS

Header-Bidding % of total revenue 3% From one enabled partner in 2014 to seven and counting From manual to automated DFP line item setup 2014 24% From placed on the page to home brewed framework 2015 43% 2016 From no-timeout to controlled timeout Largest revenue driver in 2016

Why is header-bidding growing so quickly? Dynamic Allocation Vs.

Why is header-bidding growing so quickly? Increases the value of inventory Increases dynamic competition Growth Drivers Encourages unique demand Easier to monitor discrepancy Eliminates passbacks

Auction Dynamics: Bid Entry § Enabled partners pass bids into a publishers mediation layer INDEX Open. X § Each enabled partner runs a 2 nd price auction within their exchange AOL App. Nexus SOVRN § Bids must be returned before timeout occurs Timeout 1200 MS

Auction Dynamics: Publisher Mediation Layer $5. 50 bid INDEX $4. 00 bid Open. X AOL $3. 00 bid App. Nexus SOVRN $5. 75 bid Publisher Mediation Layer $2. 00 bid 1 st price auction

Auction Dynamics: DFP Mediation Layer § This process sets a more effective floor price for Ad. X. Minimizing auction spread and increasing performance. $5. 75 bid DFP Mediation Layer 2 nd price auction § Cuts the bid disparity of a nonheader setup due to pricing granularity. This is why you need so many line items. § Dynamic pricing enables appropriate bid levels for high and low valued inventory.

Latency: Fighting the Good Fight Page latency negatively impacts: Billable inventory User experience Viewability Fight latency with: ü Async integrations ü Testing different timeouts ü Limiting partners

Pricing: You’re Being Floored The vast majority of header partners use publisher specific price floors to optimize and apply bid pressure on their advertising partners. So what? Why does that matter? § Floors too high: Partner may not bid deep into page sessions § Floors too low: Partner wont be competitive for ‘first look’ inventory

Pricing: You’re Being Floored Ask your partners how your inventory is being floored Anonymize and share bid data Flooring Best Practices Diversify floor pricing where available Push for dynamic pricing

Let’s talk about rappers

rappers wrappers § No market leader: Why? • Choice paralysis • Difficult to test multiple solutions • Partners don’t play nicely with one another • Misinformation amongst partners and publishers § Do you need a wrapper solution to be successful?

The Levels of Header-Bidding Level 1: Getting Started Level II: Power Up Level III: Flight Async integrations Tracking impression/revenue discrepancy vs. DFP Dynamic price floors/timeouts Price priority line items More granular line item setup A/B testing All bid values are net Static timeout enforced Throttling partners Examine what revenue impact new partners provide A sense of partner(s) flooring strategy Bid multipliers A sense of average CPM per pageview Unified reporting Throttling partners: The ability to cut or add partners based on performance per user session. Bid multipliers: Applying a multiple to a partner’s bid to increase its chances of winning the impression. Unified reporting: Automated insights into partner performance, including latency and percentage of revenue. This is used for determining what dynamic bid floors to pass.

What’s next for header? § Alleviating the publisher complexities required to play § Server to server integrations § Automated data collection § Defined strategies for handling many partners § Google’s response? It wont be ‘First Look’

THANK YOU! QUESTIONS? @EDH 5 emry@studybreakmedia. com

DATA SLIDES WERE NOT INCLUDED IN PRESENTATION

Key metrics for ad performance Measurement What does it measure? Formula Real CPM based on total impressions, not paid Revenue / (total impressions / 1, 000) Revenue per 1, 000 pageviews Revenue earned for every 1, 000 pageviews Revenue / (total pageviews / 1, 000) Revenue per 1, 000 sessions Revenue earned for every 1, 000 sessions Revenue / (total sessions / 1, 000) Pageviews per session Average pageviews seen per user session Pageviews / Sessions Impressions per pageview Average impressions seen per pageview shown Impressions total / Pageviews Ad type SOV Percentage of total revenue per ad type Ex. Header, Ad. X, Rich-Media Ad type rev total / Total rev

Header-Bidding: metrics to monitor Measurement What does it measure Formula DFP impression total vs. partner impression total Impression discrepancy Partner imp / DFP imp DFP revenue vs. partner revenue Revenue discrepancy Partner rev / DFP rev Impression weight % of total impressions HB partner accounts for Partner imp total / total header impressions Revenue weight % of total revenue HB partner accounts for Partner rev total / Total header rev Header-Bidding revenue vs. total revenue % of total revenue headerbidding accounts for HB rev total / Site rev total Pageview vs. Impression growth Difference between pageview and impression growth abs(Imp growth % - Pageview growth %)