Nowadays, in the online world, hardly any user is surprised to see ads on apps and websites (sometimes even too many ads). So far, everything seems “normal.” However, some users might find it curious that just minutes after seeing a certain banner or clicking on a particular creative, that same ad—or another with a similar or more striking message—appears again, perhaps offering a bigger discount or a new deal on the product or service they previously interacted with. It wouldn’t be unusual either for that same user to turn off their computer and switch to watching TV to relax, and—surprise!—that same “annoying” ad they saw earlier on their computer now shows up again, this time as a pre-roll video ad on a streaming platform.
No, it’s not magic—this technique is known as retargeting or remarketing. It’s about reaching users again after they’ve shown some level of “interest” in an ad, either by clicking on it or simply by browsing the brand’s website. After that, they’re “followed” as they continue browsing, with the goal of showing them ads that are more relevant to them. This “magic” happens thanks to cookies, device IDs, tracking pixels, or user IDs (if they’re logged in), in some cases. It’s a broad topic, and we might dive deeper into it in future articles.
Now, let’s focus on what happens after a user has interacted with an ad and a goal (conversion) has been achieved—whether it's a sale, an app download, filling out a form, or something similar.
Let’s start by defining the term attribution model. It refers to the process of determining which campaign, channel, or user touchpoint gets credit for a conversion—in other words, which action is responsible for a purchase, an app download, or even a visit to a physical store, for example.
1. Last Click
The channel that generated the last click before the conversion gets 100% of the conversion. This is the most commonly used method, mainly because of its technical simplicity. The vast majority of online marketing tools offer it.
It’s important to note that all previous user actions leading up to the conversion—such as views, other clicks, etc.—are ignored.
2. First Click
This model is similar to the previous one, but in this case, 100% of the conversion credit is assigned to the first interaction the user had with the ad. It shares the same limitation as the last-click model.
3. Linear Attribution Model
In this model, an equal percentage of credit is assigned to each interaction throughout the conversion journey. It’s very useful for analyzing in detail the steps a customer takes before completing a conversion.
4. Decreasing over time
This model gives a higher value to the last interactions produced with the ad, than to the oldest ones.
5. Position-based model
In this model, a percentage is assigned to each point of the user's journey until the conversion takes place, giving more value to the first and last interactions.
It is a good model for advertisers as it provides an overview of the user's interaction, allowing to strengthen those steps where most users drop off in the conversion funnel.
6. Based on data
This model uses Machine Learning algorithms to determine the contribution of each interaction in the conversion process, applying analysis to a large number of previous conversions, in order to detail which steps are the most important.
7. Custom Models
There are many tools on the market that allow you to customize the attribution model to analyze user intent throughout the conversion process, assigning different weights to each step of this process.
It is important to highlight the significance of attribution when multiple devices are used. For example, someone may perform a search on a desktop, then view information on a tablet, and finally make a purchase on a mobile phone. The attribution is ultimately assigned to the mobile device, and thanks to cross-device technology, that conversion can be tracked—and even attributed to the final device used along with the steps followed.
Traditionally, online conversions were tracked using third-party cookies. However, since cookies are tied to a specific device and browser they are useless for cross-device attribution.
Therefore, to attribute conversions across different browsers and devices, several solutions are used by AdTech companies specialized in this area:
This technique links devices through a common identifier, such as an email address or phone number, allowing users to be identified and matched across different devices. It is highly accurate, but it requires an active login and cannot be applied without it.
This technology uses IP addresses, geolocation, device type, and even timing and behavioral patterns to infer that different devices belong to the same user. It is logically less accurate than the previous method, but on the other hand, it does not require a declarative user input such as an email address.
The attribution process is similar in both approaches, differing mainly in the type of data used to identify the user. For Walled Gardens such as Meta and Google, it is easier to identify users through deterministic methods, since many users are logged into their accounts across multiple devices.
On the other hand, independent AdTech companies need to rely on technologies like DMPs and measurement tools to carry out attribution. These companies collect data from various sources, build user profiles, and generate what is known as an identity graph. With this identifier—or identity graph—AdTech platforms can use it to target users and attribute conversions.
At this point, thinking about attribution models or conversions in CTV can seem complicated—but it is indeed possible.
CTV is not a click- or interaction-based environment like desktop, mobile, or tablet. However, by adapting the attribution approach, it is possible to measure which actions on CTV ultimately led to a conversion for the advertiser.
A connected TV is a device typically used by multiple members of a household. There are no cookies, and tracking is often limited—third-party cookies cannot be used, or they have restrictions based on the app or operating system.
Because of all the factors mentioned above, the way to attribute a conversion is usually by determining whether a user saw a video ad and then converted on another device. However, there are several techniques that can be used:
View-Through (Post View):
A user sees an ad on CTV and later converts on another device. By using the same IP address, for example, it is possible to attribute this type of conversion.
Households Graphs
As we previously discussed in our article about HouseHolds IDs, these are identifiers that remain associated with a household over time. This identifier is unique and represents a household, grouping together all devices and users that share certain characteristics—such as IP address, router MAC address, Wi-Fi network, geolocation, etc.
But the Household Graph goes a step further. It is a more complex structure—a graph—that connects multiple user and device identifiers to represent the digital composition of a household. In short, a Household ID is a node within the broader structure known as the Household Graph, which provides a comprehensive view of how users and devices within a household are connected. This is used for attribution modeling in CTV.
Incrementality-Based Model
This model uses an A/B test approach with one sample of users who are shown ads on CTV and another similar sample who are not. Conversion results between both groups are compared, and the lift in conversions among the exposed group is measured. Tools like Google’s Conversion Lift are based on this model.
It’s also possible to use QR codes in CTV ads to later track activity through tagged URLs opened on mobile devices, for example, or to use promo codes. These are forms of direct attribution, although in some cases they may be slightly less precise than the previously mentioned models.
Retargeting is one of the most powerful tactics to maximize conversions, as it targets users who are already familiar with your brand. However, its effectiveness in a modern environment increasingly depends on identity technologies, first-party data, and a privacy-compliant approach. To make it work cross-device, it is essential to have a robust data infrastructure—whether through proprietary logins, integration with identity platforms, or the use of advanced programmatic solutions.
On the other hand, attribution in CTV is possible but requires a different approach than the traditional web and online methods. It’s not about tracking clicks, but about measuring exposure, estimating impact, and connecting devices at the household or identity level. Technologies such as the aforementioned ID graphs, household graphs, or incrementality models make this kind of conversion measurement possible.
At tvads we has a professional team able to advise you on this field and and guide you in any area of your streaming advertising business, advising you or even operating it on your behalf if necessary
All author posts