As digital media buying via programmatic channels proliferates globally, data has become centerstage. Whatever the end objective, media buyers are increasingly leveraging data to reach more relevant audiences,  enhance engagement with their brands, drive prospects to specific actions or even understand the consumer journey better.

Whilst the global usage of data has increased due to several reasons, the single most contributing factor has been the exponential increase in high quality data that is being generated due to better tools, internet usage, sensors and other devices. The decrease in computing and storage costs also means that it is cheaper to process this data and make it more valuable. Muehlhauser and Rieber are researchers who have demonstrated that the processing power (measured with a metric called MIPs/$) has been growing by a factor of 10 every 5 years.

Globally, marketers including programmatic advertisers, are experimenting with and adopting various data sources such as digital/search data, ad-serving data, customer data from CRM systems, social media monitoring data and geo-location data. A study from Centro regarding advertiser perceptions about data revealed that advertisers do rely on a number of partners like data providers as well as data management platforms to implement their data strategies.

Closer to home in APAC, a recent study by eConsultancy in April 2019, called “The Omnichannel Imperative”, highlighted that real-time marketing along with omni-channel delivery and 360 View of customers are the key priorities for South East Asian marketers. All of these key business priorities require a deep understanding of the data landscape as well as key frameworks to effectively leverage data resources to effectively pursue these priorities.

Anecdotally, whilst data adoption in South and South-East Asia remains in the early stages of maturity, the dialogue around the usage of data has become more and more nuanced. Against this backdrop, the objective of this piece is to demystify key data definitions and scenarios in which common data types could be used. 



One of the fundamental frameworks of classifying data is created by understanding what aspects of user attributes/behaviour the data represents and where it has been sourced from.

Deterministic data typically represents a ‘fact’. The data that is centered around a ‘fact’ represents attributes from the past that can be conclusively established or verified. Typically, this data is sourced directly from users with their consent through a user data acquisition process. For example, a gaming application on a phone could request users to provide their demographic or other information. This data could eventually be made available to a programmatic media buyer for targeted advertising on this application.

Probabilistic data is a ‘mathematical  guess’ about the key attributes of a user and is derived from modeling the existing data points on user device configuration, user behavior, geo-location and several other factors. An example of probabilistic data treatment when a data provider infers the demographic gender and age of a user through the unique pattern of applications usage of that user on her mobile device.

Deterministic data is valuable because it is considered highly accurate. The key challenge is that it becomes difficult to scale deterministic data outside of the large technology platforms like Google or Facebook. In recent times, publishers are looking to launch common user log-in initiatives to have better access to deterministic data and to also be compliant to recent privacy regulations like GDPR.

With greater access to data and more sophisticated means to probabilistically fingerprint users, the scale of probabilistic data is definitely larger. Whilst the accuracy of probabilistic data has steadily increased, the more recent changes in privacy regulation landscape have created some uncertainty for such data collection at scale.

Either way, marketers stand best placed when they use a combination of both deterministic and probabilistic data to power their digital marketing objectives via programmatic channels.



A more conventional framework for data is based on the sources from which data is procured. This can be briefly defined as:

  • First party data: This is data that is derived from proprietary data sources of an advertiser or a publisher. This data could be from an advertisers CRM (customer relationship management) software or POS (point of sale) systems or even offline data collection tools like forms/surveys. First party data is the most relevant and granular information to an advertisers’ business but often is limited in scale.
  • Second party data: Second party data is the first party data of another organization – usually a partner. An example for an agency could be second party telco data that comes through a strategic partnership with a telco. Second party data is also quite granular, detailed and potentially has a bigger scale than first party data through a strategic partnership approach.For example, a tourism board which might be looking for confirmed travellers can strike a partnership with a large ‘online travel booking portal’ and get access to such data to drive specific advertising campaigns to this target group at scale.
  • Third party data: This is data that is sourced from data exchanges or data providers who buy, aggregate and manipulate raw data to convert into ‘segments’ that are easier to describe and activate. The biggest advantage of third party data is that it can scale a lot better than first or second party data.

First and second party data is typically collected using various methods ranging like code snippets on digital properties (called tags or SDKs), server-to-server data transfers and log-in based methods. Data that is collected by these means could be used by advertisers or could be further procured by downstream payers like Data Providers (e.g. Oracle Data Cloud, Eyeota) who could use this as the raw material to generate more specific third party data. This third party data is typically propagated into the programmatic media ecosystem via direct integration of Data Providers with programmatic platforms or via integration of Data Exchanges to programmatic platforms.                                 

Once the data has been procured from first, second or third party data sources or a combination of them, this data is typically activated in terms of audience segments.The following section describes four common types of audience segment targeting:

Demographic Targeting
  • Data and parameters used for demographic targeting are socio-economic in nature such as age, gender, income, household, race, education etc. A bank, which is looking to target young urban millennials who are professionally employed, for a relevant lifestyle credit card can apply suitable demographic filters to arrive at an optimal audience segment for its marketing campaigns.
  • The advantages of using demographic segments in targeting is that this data tends to be accurate and widely available ensuring at least a broad compliance to the brand or product target audience. Hence, these data segments are more suitable for early stage brand building or awareness campaigns that go hand-in-hand with product refreshes or new market entries.
  • Key data providers and management – Oracle Data Cloud, Eyeota, Facebook
Behavioural Targeting
  • Behavioral targeting is best described as targeting users based on their behavior or intent on the internet. For example, a company looking to promote fitness watches will be effective with audiences who have a clear health and fitness affinity based on their internet behaviours such as health/fitness gear purchase behaviours or consumption of health/fitness content on websites.
  • The advantages of using behavioural segments is that they tend to be closer to the client ‘personas’ that a brand marketer typically creates. This makes such data targeting more amenable to prospecting style campaigns where marketers might have specific objectives like winning more market share. A key limitation to keep in mind is that behavioural data at times tends to be limited in scale and at some other times might just not be accurate enough to drive the desired outcomes.
  • Key data providers and management – Oracle Data Cloud, Eyeota, Google
Contextual Targeting
  • In relation to digital advertising, contextual targeting of ads refers to the adjacency of viewed ads to the content, flow and environment of the web-page or app that the user is visiting. A user viewing sports on a news website can be targeted to view sports-related ads.
  • Context is an old concept that goes way back to the days before digital marketing became prevalent. The advantage of using contextual data is that marketers can reach out to users when they are in an appropriate context and hence exploit their “in-the-moment mindset”. Context also provides more control to marketers to shape the environment in which advertisements are displayed. 
  • Key data providers and management – Oracle Data Cloud
Location Targeting
  • Location-based targeting is employed to target the segment of users who are in a set physical location. Cosmetic brands looking to drive potential consumers to their latest ‘organic serum collection’ can set geographical locations to prime shopping district zone near their showrooms which have good footfall of the young women demographic. Targeting can be done down to the zip code or geo fenced areas, if the data is available.
  • Location targeting enables brands and advertisers to reach out to their prospects and customer at places where they are likely to make a purchase or demonstrate intent. These locations could be based on historical behaviour or even real time user presence. Whilst effective location based advertising is the holy grail of programmatic advertising, this data at times can be limited in scale.
  • Key data providers and management – Factual, Facebook, Google

Depending on the audience that advertisers are looking for, multi-layering of the above segments and targeting are typically deployed to reach the most relevant and qualified audience.



The recent growth in programmatic adoption along with the availability of data from various ecosystem actors offers very exciting opportunities to marketers in APAC. At the same time, the industry faces some challenges due to regulatory headwinds and browser policy changes. Another challenge outside of the pure data realm is the shortage of relevant talent in the region. The players who constantly invest in developing the right talent and collaborate with the appropriate ecosystem partners will be best placed to ride this explosive wave of data-powered programmatic advertising. 


IAB SEA+India Programmatic Committee