Help Center

Key terms for implementing the Roqad graph

We support two ways of integration: by tracking codes and server-to-server. 
The former could be done with JavaScript codes which requires either cosmetic changes (campaign or use-case specific) or for NO changes at all.
 We can also provide an IMG tracking code but in this case it needs to be customized.

The second integration type (server-to-server or s2s) is just a RAW data file sharing using (s)FTP or AWS S3. For this scenario some additional criteria must be met, i.e. presence of all required attributes in the data exports. 
Before the project starts, we always engage the Customer Team in a use-case clarification discussion to make sure the most beneficial way of integration is chosen.

Deterministic data (also called “label data”) includes information such as user name, email address, etc. tied to a specific user. It is considered one of the most accurate ways to identify a person in the online world. Typically, social media platforms, gaming sites, and large eCommerce businesses have the largest number of log-in data. However, when the person moves away from these platforms to other digital properties or does not login to a specific app or website,  it is almost impossible to track the user across their multiple devices. Here, a probabilistic approach is needed in order to solve for online identity resolution.

Probabilistic data itself means observation data that does not contain a persistent identifier belonging to a given user (e.g. an e-mail address). A probabilistic identity solution collects a wide range of signals from the observation data and computes connections between the underlying device identifiers. The more matching attributes that are received and mapped for a given device and user, the higher the likelihood that these device identifiers belong to the same person.

In summary: A deterministic Cross-Device solution works well, as it holds high precision level between the device connections; however, it lacks in scale. Using a probabilistic solution in addition to the deterministic one increases scale of the model at relatively high precision rates (+90%).

Some of the common use cases for the Graph include:

  • cross-device retargeting -> reaching more devices of people from the group of your interest – perfect for branding campaigns;
  • cross-device analytics -> including media mix performance measurement, attribution, funnel analysis;
  • BI driven insights -> creating personalized reporting, showing holistic view of the audience / users;
  • performance optimization -> making informed decision about the digital budget allocation;
  • sales recovery for affiliate marketing networks filters the traffic of its data partners on user consent for Here, uses the technical mechanisms of the IAB Europe Transparency and Consent Framework. has vendor ID_4 on which the filter is set, thus we receive only device requests containing user consent for

The Cross-Device matches we produce are of high precision and generally reflect the similarity between devices and similarity between how they are used by particular people. The high precision requires that a device has to be similar to a particular user.

Therefore device with unspecific usage characteristics (because used by many people) will most probably not connect to any user.
Additionally, if a device is highly similar to many potential users, which is rather uncommon, it will typically connect to the user that it’s most similar to it. Moreover, filters users that are too big in size, which is an additional reason why the device most probably will not connected to any other device. USP is privacy compliance and strongest local Cross-Device database in central Europe. works on 100% opt-in users, and thus, is operating under the guidelines of the ePrivacy directive.

Active user consent means that will not process any data without a user agreeing to it. The user’s consent for processing can be withdrawn anytime via several opt-out mechanisms. provides identity products that are covering different regions. Our Private Identity Graph is available globally, the Public Identity Graph is live in the European Unions and North America. Roqad Link, which is our data onboarding product, is also live for EU/UK and the United States. For more information, check the pages for Roqad Graph and Roqad Link data onboarding.

Please send a mail to, or fill in our contact form at the bottom of the page.

We will set up a clarification call or personal meeting to answer all questions.

The Graph computation frequency varies from Customer to Customer – from once per week up to once per month. It’s strongly dependent on the use-case and specific needs.

Tuning the machine learning algorithms is our secret sauce :)‚ However, to understand the process a bit better we strongly suggest reading our 
White papers. It’s definitely worth spending 30 minutes to understand the idea of cross-device matching.

We strongly recommend talking with our Team to discuss your use case / needs in depth. In most cases, our standard products Roqad Identity Graph and Roqad Link data onboarding are sufficient to meet the customer’s expectations.

However, we are always open to discussions concerning improvements – that’s why we invite our clients to talk about their product roadmaps and ideas every quarter.

The entire adtech ecosystem has been impacted by Intelligent Tracking Prevention (ITP) regulations from 
the various browsers. In general, large adtech businesses with high login rates are the ones that benefit from ITP regulations.

For’s Graph product, this means:

  • Safari browser: Cookies are cleaned as users no longer contact the 
tracking domain in the last 30 days. Thus, 3rd party tracking is not 
  • Chrome browser: has made all necessary changes to enable 
tracking of cookie IDs with Chrome 76. Deprecation of third party cookies in Chrome expected in 2023.
  • Firefox browser:  Enhanced 
Tracking Protection is installed automatically for Firefox users. List of blocked tracking domains: 
  • Other major browsers have not yet announced any tracking prevention 
measures (eg Opera, Internet Explorer, etc.).

We do remove strange outliers to avoid skewing of the performance metrics. The cut off will be discussed with the client. For example, if we find 20 cookies connected to one hashed email address (HEM), then it’s most likely that it’s not a single user. In a 30 day window, there are usually 1 or 2 MAIDs tied to a HEM. If there are 20 or more, then it’s probably erroneous, a bot,
or a shared account. We only want users that are potentially commercially viable for our clients so we remove the outliners. 

Additionally, Roqad advises each client to pre-filter and cleanse the delivered data as much as possible to avoid running into skewed results.

Get in touch!

Interested in learning how the Identity Graph can help your business grow?

Let's Stay in touch!

Subscribe To Our Newsletter

Roq-news: where we share trends and developments that we’re seeing in the world of digital advertising.