Breaking the Magician’s Code

“Any sufficiently advanced technology is indistinguishable from magic.” – Arthur C Clarke.

For people who aren’t aware of the inner workings of an ad network, user targeting can seem like magic. User targeting refers to advertisements being placed in such a manner that they reach appropriate consumers based on various traits such as demographics, location, interests etc.

Understanding user behaviour is instrumental in delivering an enhanced consumer ad experience. Consumer attention to ads, especially on the mobile platform, rises if it relatesdirectly to their interests. User targeting is an essential tool to help advertisers reach their desired segment of audience. This tailored approach yields greater user-engagement and interaction, providing advertisers with better ad campaign results.

For example, a travel app might be told by a network that their ads will be shown to users who are travelers or potential travelers. But, how does an ad network identify travelers and how does the brand ensure its targeting is effective?

This blog addresses the top five questions you should ask your ad network so that you know exactly what you are signing up for.

Question 1: How do you uniquely identify users?

For most of the targeting to work correctly, you will need to identify unique users first. If this user identification is not accurate, then this could potentially derail all your campaign setup and targeting.

Without user de-duplication,(a technique for eliminating duplicate copies of repeating data) you could end up with a situation shown in Figure 1. A single user might access different websites/apps that send different device IDs to the ad network. If the ad network is not able to map the different device IDs to the same user, then the multiple ad requests will be registered to different users instead of the same user.

Imagine that you were doing frequency capping, where you wanted to show your ad to each user just five times. Without de-duplication, and since you cannot map all the device IDs to the single user, that user might see your ad 15 times instead of the intended five times.

With de-duplication, a global unique identifier is created for each user and all the different device IDs associated with that user are mapped and stored for reference. This lookup table is referenced each time so that, irrespective of the device ID, we are always able to identify the user as shown in Figure 2.

Question 2: What sort of user data do you collect?

In this day and age, it is extremely important for you to be careful not to collect any information that is classified as Personally Identifiable Information (PII), or be associated with an ad network that collects such information. Hence, it would be prudent to ask the ad network about the information collected by them so that you are in the clear. For example, the company AdTruth, a digital fingerprinting company, is very strict about not using PII information. So, technologies that leverage AdTruth to track users might be worth considering.

Another point to note is the ad network’s compliance with COPPA (Children’s Online Privacy Protection Act). COPPA is an act that prevents the online collection of personal information by persons or entities under U.S. jurisdiction from children under 13 years of age.

You would want to work with an ad network that is aware of these restrictions that exist to protect people’s privacy, and can adhere to it strictly.

Question 3: What is the potential scale of my campaign?

It is well known that if you try to fine-tune the performance of your campaign by having extremely granular targeting, the scale of your campaign will take a hit. Performance and scale are like a seesaw. Usually an increase in one leads to a decrease in the other. The key to a successful campaign is in trying to find the right balance. Since most developers have strict budget constraints, ad networks usually ensure that costs are met, but sometimes they fail to tell you that the scale at those costs might not really be up to expectations.

If you are not getting sufficient scale, it might also make sense to revisit the campaign. The ROI of the campaign might be generated even at slightly higher costs as long as you are getting sufficient scale. Targeting is another aspect that you can consider reducing to get better results. But then again, you need to keep a strict eye on performance while you are trying to balance scale so that your campaign costs do not go through the roof. A good ad network would try various options such as dynamic pricing based on conversions to ensure you get both scale and performance.

Question 4: What third party data sources do you use?

Enriching data with third party data sources can be hugely beneficial to advertisers. For example, BlueKai provides vital user classifications such as travelers, sports enthusiasts, and so on. They do this by effectively collecting user information across different sites.

If your ad network tells you that they enrich their data through third party vendors, you would want to ask them which vendors they work with, which vendors are used for the different regions, and what sort of data coming from the vendors is leveraged and how. An example of how the data can be leveraged is to strengthen the capabilities and results of look-alike targeting as shown in the image below.

Question 5: What are the limitations or constraints for the targeting to work?
The final question is on the constraints or limitations, especially relevant if you are at the stage of evaluating the different ad networks available. For example, you might decide that you want to use geo-targeting and target tier one cities and an ad network might claim that they have geo-targeting. But, when you run your campaign, you might realize that they have geo-targeting only in North America whereas your primary market is Asia. The tracking technology that needs to be used to ensure that a targeting feature works might also be a constraint and you would need to know this from the start, to decide which tracking technology you should integrate with, so that it supports all the targeting functionalities that you want to incorporate.

While targeting might seem mysterious or even magical, it really boils down to one simple thing – data, and how this data is used. So, if you are thorough and persistent during your conversations with an ad network, you will not only know exactly what you are getting into, but also what the best targeting is to obtain optimal results from your ad campaigns. Good luck!

This article first appeared in TechCircle.in