MODULE IX – ADVANCED

PROGRAMMATIC ADVERTISING
By
Sujith Caldera
Session 2
Session Outline
MODULE Units AREA OF STUDY

Conceptual understanding of Programmatic

Basics of programmatic

Direct buys vs programmatic buying

Importance of programmatic

Programmatic evolution

Advanced Real-time bidding

9.1 Introduction to
Module IX programmatic
programmatic
advertising Pros and Cons of Programmatic

Importance of data in programmatic

Programmatic audience targets

Demand side platforms (DSP)

Supplier side platforms (SSP)

Data management platforms (DPM)

Data driven advertising with programmatic

How Programmatic works
DMP

Ad
AdServer DSP Exchange SSP AdServer

Campaign Some Campaign Website abc.com Website abc.com

Creative Size 728px x 90px Page Title Some News Page Title Some News
Location Colombo, Sri Lanka Placement Size 728px x 90px Placement Size 728px x 90px
… … IP 123.123.123.123 IP 123.123.123.123
Location Colombo, Sri Lanka Location Colombo, Sri Lanka
Device / OS MacBook / MacOS 13.0 Device / OS MacBook / MacOS 13.0
Browser Safari 15.7 Browser Safari 15.7
… … … …
What is real time bidding
Real-time bidding (RTB) is a method
used in programmatic advertising
where digital ad impressions are
bought and sold through an automated
auction in real time. It involves the
buying and selling of ad inventory on a
per-impression basis, with the auction
occurring instantaneously as a
webpage loads.
How real-time bidding works
Creative 1
$0.25

DSP 1
+$0.15
$0.20
Creative 2
$0.35

Creative 5

Creative 3 DSP 2 Ad SSP AdServer

$0.25 +$0.05
$0.20 Exchange +$0.05

Creative 4
$0.15
$0.15 $0.10

DSP 3
+0.05
$0.25
Creative 5
$0.30
 How real-time bidding works
Advertiser Demand: Advertisers express their interest in displaying ads to a specific target audience. They may use data and audience
insights to define their targeting criteria, such as demographics, interests, location, and browsing behavior.

Publisher Supply: Publishers, who own websites or mobile apps, make their ad inventory available for auction. They provide
information about the available ad slots, including the format, size, and placement on their webpages or apps.

Ad An ad exchange or middleware platform acts as a mediator between advertisers and publishers. It facilitates the
auction process by collecting and analyzing data, matching advertisers' demand with publishers' supply, and
Exchange/Middleware: conducting the real-time auctions.

When a user visits a webpage or opens an app, the publisher's site or app sends a bid request to the ad exchange.
User Visit: This bid request contains information about the user, such as their device type, location, and sometimes their
browsing history.

The ad exchange receives the bid request and sends it to potential advertisers who may be interested in displaying
Auction: an ad to the user. Advertisers assess the user's profile and decide whether to bid on the impression. If they choose
to bid, they submit their bid and relevant creative assets (ad content) to the ad exchange.

Real-Time
The ad exchange receives the bids from advertisers and uses its algorithms to determine the winning bid based on
Deaccessioning: predefined rules and criteria. The highest bidder wins the auction.

Ad Delivery: Once the winning bid is determined, the ad exchange notifies the winning advertiser, and the ad is served to the
user's device in real time. The ad content is typically loaded into the webpage or app, and the user sees the ad.
How real-time bidding works
• "Real-Time Bidding Example
• Let's say Silk is a UK-based beauty brand that just launched a new shampoo and is
running a campaign. They set up their campaign on a Demand-Side Platform (DSP) and
are targeting users who regularly shop for shampoo products, are located in the
Manchester area, and are between 18 to 30 years of age.
• The brand also wants its ads to only show on sites related to beauty and lifestyle. A user
visits a publisher's site. The publisher's Supply-Side Platform (SSP) sends a bid request to
the ad exchange, where Silk's DSP will be evaluating the value of the impression.
• The DSP will then determine if the user meets the parameters outlined in the campaign.
If so, the DSP will submit a bid. If Silk has the winning bid, the user will see the ad once
the page loads.
• This process happens thousands of times on different webpages during the length of
Silk's ad campaign. Silk's paid ads manager will also be monitoring their ad's performance
on the DSP to see if it's reaching the desired audience, or if the parameters should be
adjusted.
Bidding Algorithms (Header Bidding vs Waterfall)
• Header Bidding • Waterfall
• Header bidding is an advanced bidding method • Waterfall is a sequential bidding method where ad
that allows publishers to offer their ad inventory to requests are sent to demand sources in a
multiple demand sources simultaneously, before predetermined order. If the first demand source
making a call to their ad server. It aims to maximize doesn't fill the impression, the request is then
revenue by creating a fair and competitive auction passed to the next one until a bid is received and
environment for advertisers. the impression is filled. It follows a prioritized
"falling down the waterfall" approach.
• Here's how it works:
• When a user visits a webpage, the publisher's site • Here's how it works:
sends a bid request to multiple ad exchanges or • When a user visits a webpage, the publisher's site
demand partners, often through a header bidding sends a bid request to the first demand source in the
wrapper. waterfall.
• Each demand partner receives the bid request and • The first demand source evaluates the impression and
evaluates whether to bid on the impression. decides whether to bid. If they decide not to bid or
• Demand partners submit their bids with the relevant don't respond within a set time, the request is passed
creative assets and pricing information back to the to the next demand source in the waterfall.
publisher. • This process continues down the waterfall until a
• The publisher collects all the bids and selects the demand source submits a bid, and the impression is
highest bid to display the winning ad to the user. filled.
• The winning bid is then sent to the ad server to deliver • The winning bid is then sent to the ad server for
the ad to the user's device. delivering the ad to the user's device.
Bidding Algorithms (First Price Auction vs
Second Price Auction)
• First Price Auction: In a first price auction, the • Second Price Auction: In a second price auction,
winning bidder pays the exact amount they bid. also known as a Vickrey auction, the highest bidder
The highest bidder wins the auction and is charged wins the auction but pays the price of the second-
the full price they offered. highest bid.
• Here's how it works: • Here's how it works:
• Multiple advertisers participate in the auction and • Multiple advertisers participate in the auction and
submit their bids for a specific ad impression. submit their bids for a specific ad impression.
• The auction takes place, and the highest bid • The auction takes place, and the highest bid
determines the winner. determines the winner.
• The winning bidder is charged the full amount they bid, • However, the winning bidder is charged the price of the
regardless of whether it was significantly higher than second-highest bid, not their own bid. This encourages
other bids. participants to bid their true maximum value.
• Example: Suppose there is an ad impression • Example: Let's consider the same scenario as
available for auction, and three advertisers submit before with three advertisers: Advertiser A bids $5,
their bids: Advertiser A bids $5, Advertiser B bids Advertiser B bids $7, and Advertiser C bids $4. In a
$7, and Advertiser C bids $4. In a first price second price auction, Advertiser B would win the
auction, Advertiser B would win the auction and be auction. However, they would only pay $5, which is
charged $7, which is the full amount they bid. the second-highest bid submitted by Advertiser A.
Bidding Algorithms (Hard Price Floor vs Soft Price
Floor)
• Hard Price Floor • Soft Price Floor
• In a hard price floor, the minimum price for an ad impression • In a soft price floor, the minimum price for an ad impression
is set as a guideline rather than an absolute value. Bids below
is set as an absolute, non-negotiable value. Bids below the the soft price floor are not automatically rejected but may
hard price floor are automatically rejected, and the still be considered based on certain factors, such as the
impression is not sold. advertiser's relationship or the demand for the impression.
• Here's how it works:
• Here's how it works:
• The publisher or ad network sets a recommended minimum
• The publisher or ad network sets a fixed minimum price for an price for an ad impression, known as the soft price floor.
ad impression. • Advertisers participate in the auction and submit their bids for
the impression.
• Advertisers participate in the auction and submit their bids for
the impression. • Bids below the soft price floor are not automatically rejected
but may be subject to additional evaluation or adjustments
• If any bid falls below the hard price floor, it is instantly rejected, based on various factors, such as advertiser relationships,
and the impression is not sold to that bidder. demand, or other market conditions.
• Example: Suppose a publisher sets a soft price floor of $3 for
• Example: Let's say a publisher sets a hard price floor of $3 for an ad impression. Advertiser A bids $2, Advertiser B bids $4,
an ad impression. Advertiser A bids $2, Advertiser B bids $4, and Advertiser C bids $3. In this case, Advertiser A's bid,
and Advertiser C bids $3. In this case, Advertiser A's bid although below the soft price floor, might still be considered
would be automatically rejected because it is below the hard based on factors such as their relationship with the publisher
or the current demand for the impression. Advertiser B's bid
price floor. The impression would only be sold to Advertiser would likely be accepted, while Advertiser C's bid would
B, who bid above the minimum price. meet the minimum price requirement.
Campaign Planning in
Programmatic Advertising
Planing a Programmatic advertising campaign

5 W’s & 1 H

1. WHY? -your business and marketing goals

2. WHO? - defining your customer persona
3. WHAT? - the message you want to drive
4. WHEN? - engaging users at right time
5. WHERE? - reach with relevance
6. HOW? - delivering with efficiency
Targeting ….
Cookies
A cookie is a piece of data from a website that is stored within a web browser that
the website can retrieve at a later time.

In programmatic advertising, cookies are used for:

● uniquely identify a user
● keep the user identities anonymous
● keep track of the user behaviour and preferences
Cookies

Registered Purchased a
Watched a
using an online watch from
movie in Netflix
form Amazon

Cookie ID: 123 Cookie ID: 123 Cookie ID: 123 Cookie ID: 123
2022-01-21 12;04:31 2022-01-21 12;04:31 2022-01-21 12;04:31
Watched and hour of Netflix Watched and hour of Netflix Watched and hour of Netflix

Age - 25 Age - 35

Gender - Female Gender - Male

2022-01-25 21;45:11
Purchased a high-end watch
Data Management Platform (DMP)

Collect Enrich Profile Segment Activate

Collect data from Convert the raw Building a single Group the user Activate the created
1st, 2nd, 3rd-party data into meaningful user profile and profiles based on audience profiles for
data sources insights customer journey rules and ML campaigns through
across devices models DSP
How it works….
Female, age between 20 to 30 with high
household income DMP
abc, xyz, 123, 456, 678, ...

Ad
AdServer DSP Exchange SSP AdServer

Campaign Some Campaign Website abc.com

Creative Size 728px x 90px Page Title Some News
Location Colombo, Sri Lanka Placement Size 728px x 90px
Audience Female, age between 20 IP 123.123.123.123
to 30 with high Location Colombo, Sri Lanka
household income Device / OS MacBook / MacOS 13.0
… …
Browser Safari 15.7
Cookie ID 123
… …
Cookie Based Targeting
Behavioral Targeting • Demographic Targeting
Targeting user based on predetermined set • Targeting user based on profile attributes,
of behaviours. usually registration data.
Ex: • Example :
Males with High Household Income. ● Age
● Gender
● Visited a Sports Website
● Visited a Luxury Car Website ● Education
● Purchased a high end watch ● Occupation
● etc. ● Income
Cookie Based Targeting
Re-Targeting Interest Based Targeting

Targeting a user with an ad after they have Targeting user based on the types they
visited a website, took an action or did not like, such as movies, music, food, sports
take an action. and more.
Ex:
Ex:
Visited an e-commerce website, looked at ● Marvels runs ads for their
shoes, but didn’t purchase. movies targeting fans of
superheroes and comics.
How to Measure the success…
Metrics
SSP win rate is 70%. It indicates
Programmatic that out of the 100 ad
Unit impressions available for auction,
the SSP was successful in winning
Requests SSP Number of Bid requests sent from the Publisher to and monetizing 70 impressions,
SSP while the remaining 30
impressions were won by other
Bid Requests SSP Number of Bid requests sent from the SSP to DSPs
demand sources or buyers.
Bids DSP Number of Bids responses sent from the DSP
DSP win rate is 25%. It indicates
Wins SSP / DSP Number of Bids won SSP Auction that out of the 200 auctions in
Win Rate % which the DSP participated,
SSP Wins / Requests * 100 they emerged as the winning
Win Rate % bidder in 50 auctions, while in
DSP Wins / Bids * 100
the remaining 150 auctions,
other DSPs or advertisers' bids
were more successful.
Metrics
Programmatic Unit

Impressions SSP / DSP Number of ads served to the users

Reach DSP Number of unique users and ad was served to

Fill Rate SSP Impressions / Requests * 100

Clicks SSP / DSP Number of Impressions that got clicked

CTR % SSP / DSP Clicks / Impressions * 100

Click Through Rate

Conversions DSP Number of Conversions tracked

Metrics
Programmatic Unit

VCR %
Video Complete Rate SSP / DSP Number of Video Ads watched till end / Impressions * 100

Spend DSP Budget spend on the traffic

eCPM SSP / DSP Average spend for 1000 Impressions

eCPC DSP Average spend for a Click

eCPA DSP Average spend for a Conversion

ROAS
Return on Ad Spend DSP Revenue made from Ads / Spend
Optimization
Increasing Campaign Reach
Budget under utilized?

Low Bid Rate?

● Remove extensive targeting such as geo, niche audiences, categories
and site list.
● Raise frequency caps
● Add more creative sizes Frequency Caps
A limit to the number of times each user sees
Low Win Rate? an ad, generally expressed as a per-day
allotment.
● Increase CPM
Increasing CTR and Conversions
Optimize the creatives
● Ensure creative is aligned with your goals
● Pause low performing creatives
● Update creatives by changing images, titles, descriptions and CTAs

Remove Budget Eaters

● Audit the publisher list exclude those that are underperforming.
● Choose the right audience for the campaign, but don’t narrow down too much.

Implement bid multipliers to optimize toward engagement.

A/B Tests
Pick one variable to test.
● Same Audience, Different Creatives
● Same Creative, Different Audience Segments

Split the budget and other variables equally Test both variations simultaneously
Measure the campaign goal
Plan the next A/B test with a different variable
80/20 Segmentation Rule
Create the campaign by targeting a broad audience

Audit the campaign every 2 weeks

Exclude the bottom performing 20% (Budget Eaters) from the target

Subgroup the remaining 80% into two categories

● Common 80% Use an average bid
● Best 20% Use a high bid value
The same frameworks can be used for optimizing a campaign through the right
creatives.
Automated Optimizers
These bid optimizers aim to maximize the efficiency and effectiveness of ad campaigns by automatically adjusting bid
prices based on various factors and goals

Data Collection: Goal Definition: Algorithmic Optimization: Real-Time Adjustments:

• Bid optimizers collect and • Advertisers set specific • Bid optimizers use • Bid optimizers
analyze vast amounts of campaign goals, such as sophisticated algorithms continuously monitor
data, including historical maximizing conversions, and machine learning campaign performance
campaign data, user achieving a target cost- models to process the and adjust bids in real
behavior, ad performance per-action (CPA), collected data and time. They assess the
metrics, and contextual maximizing return on ad optimize bids. They competitiveness of each
information. This data spend (ROAS), or consider various factors auction and adjust bids
serves as the foundation maximizing brand such as user higher or lower based on
for making informed awareness. These goals demographics, browsing the likelihood of winning
bidding decisions. provide guidance for bid behavior, time of day, the impression and
optimization. device type, location, and achieving the campaign
ad placement to calculate goals. This dynamic
the optimal bid for each bidding strategy enables
impression. efficient allocation of
budget and maximizes the
chance of winning
valuable impressions.
Automated Optimizers
These bid optimizers aim to maximize the efficiency and effectiveness of ad campaigns by automatically adjusting bid
prices based on various factors and goals

Predictive Analytics: Machine Learning Feedback: Reporting and Insights:

• Bid optimizers leverage • Bid optimizers employ • Bid optimizers provide

predictive analytics to machine learning techniques detailed reporting and
forecast the likelihood of to learn from the outcomes insights on campaign
achieving campaign goals of bid decisions and performance, including
based on historical data and campaign performance. metrics like impressions,
performance trends. These They use this feedback loop clicks, conversions, cost-per-
predictions guide bid to refine their algorithms action, and return on
adjustments to optimize and improve future bidding investment. These reports
performance. strategies. help advertisers understand
the impact of bid
optimizations and make
data-driven decisions for
campaign adjustments.
Automated Optimizers
Bid Price Optimizers

Bid Factor
○ Base CPM = $0.1
○ If gender is male = Bid x 2
○ If city is Colombo = Bid x 5
○ If time between 2pm to 5pm = Bid x 0.5

Dynamic Bid Optimization

○ If spend > $100 and CTR < 0.10% → Add the site to blocklist
Dynamic Creative Optimizations
Automatically perform A/B testing with multiple variants of creatives based on a
template and use the right creative for the right segment.

Book a Test Drive CTR - 0.35%

Book a Test Drive CTR - 0.05%

Book a Test Drive CTR - 0.01%

Book a Test Drive CTR - 0.15%

Book a Test Drive Book a Test Drive CTR - 0.20%

Creative Formats Available..
 ALL AD FORMATS

leaderboard mpu interstitials native ads video

 RICH MEDIA FORMATS

rich media interstitial leaderboard expand larger expand full screen expand dynamic content
+sticky remaining ad https://www.youtube.com/wat refresh
https://www.youtube.co
m/watch?v=t_14Pgn-l_o ch?v=LhreMVpA4uk
 CLICK TO ACTION CAMPAIGNS

click to dial click to SMS click to USSD

 CONTEXT BASED DYNAMIC CREATIVES

Context based creatives - reacting to certain events or situations.

Weather, sports results, time of day (e.g. for food), etc. Rich media
to image solution to increase reach.
INTERACTIVE CREATIVES

Ads with user interactions:

scroll, swipe, or scratch
effects

Touch, swipe, complete

metrics are available as
trackable events
Gamifications Ads
• https://www.youtube.com/watch?v=PQOMTRyCdBY
DV 360
• Interface -
https://www.youtube.com/watch?v=omXYqDrUzgA&list=PLv6KfiLx9P
_254oGognQ5p7gdK7wX4MvF&index=2

• Create a Campaign - https://www.youtube.com/watch?v=bav5LRn5v-

c&list=PLv6KfiLx9P_254oGognQ5p7gdK7wX4MvF&index=4
 Questions
Questionon Correct
Question Text
Number Option_1 Option_2 Option_3 Option_4 Answer
Enable advertisersto Enable publishersand Enable Publishersto
Enable publishersto bid on ad inventory advertisers to bid on ad inventory
Demand-Side Platforms (DSPs) were builtprimarily
2 manage/sell their across multiple manage/sell their across multiple
to:
inventory exchanges inventory exchanges B
programmatically programmatically programmatically programmatically
True or false? OTT/CTV inventory on smart TVs and
8 TRUE FALSE A
other devices can be purchased in real timevia DSPs.
Programmatic advertising is the use of test consumer facilitate media identify marketing
12 All of the above B
automation to: interest transactions goals
Questions
14 Programmatic is the dominant ad buying Banner ads Video ads Native ads All of the above D
approach for which of the following adformats?
72 What type of programmatic channel involvesdistributing dynamic Print Virtual reality Augment Digital D
advertising via digital place-based networks? edreality out-of-
home
76 Programmatic is only appropriate for directresponse marketing TRUE FALSE B
campaigns. Is it True or
false?
77 Which term best describes how programmaticadvertising helps Optimization Personalization Efficiency Effectiveness C
marketers save time and money while achieving positive results?
 Questions
80 What does RTB mean in programmatic Real Task Bidding Real-Time Risk Test Bidding Risk Time Bid B
advertising? Bidding
87 You’re looking to buy a fixed number of ad impressions Auction packages Private auctions Preferred deals and Preferred deals C
for a guaranteed price. Which dealtype(s) would be best Programmatic and Auction
suited for you? Guaranteed deals packages
100 The ability to optimize
Which of the following is a benefit of targeting/delivery Improving marketing Avoiding human
All of the above D
programmatic advertising? using "big data." ROI error
THANK YOU