Let's make this CLIRE [ kleer ].

Connect Data

Before we start our cooperation, we first need to collect and organize several types of information. The main actors in this step are the Data Officer on your side and your dedicated Assetario Success Manager.

User Data #

User data encompasses two general categories of data, which we clean, aggregate and combine on your behalf in order to compile a fully transparent User profile minimizing possibilities of human or machine error caused by misinterpretation. After setting up the ingestion pipeline, we require you to push all of new user data in daily (or shorter) cycles in order for our models to learn from the most recent actions.

Analytics data #

Often called "event data", these are timeseries database records denoting actions of individuals interacting with your app during the course of their usage. Some typical behavioral data subcategories we leverage for a wide variety of use cases are:

  • user identifiers - GAID, IDFV, IDFA (where possible), or a custom ID, please encrypt user identifiers using a cryptographically secure method, for example a salted sha256 hash function - we never process any personally identifiable information of your users to ensure their privacy is protected
  • activity - session started/ended, push notification interactions, ...
  • monetization - ad viewed, IAP viewed/bought, subscription bought, ...
  • progression - tutorial completed, level completed, currency received, first shop visit, ...
  • social - joined a club, referred a friend, connected a social profile, supported friend, ...

Full specification of each app varies and its understanding is subject to interpretation, always to be validated and confirmed either through manual testing process or in communication with the assigned Data Officer on your side.

Behavioral data is also the most volume-heavy category, often spanning gigabytes to terabytes worth of database rows. We recommend defaulting to one of our integrated partners if possible:

If your data warehousing technology isn't listed above please contact us, our very own engineering team is more than capable (and happy) of adding your storage in a matter of days.

Contextual information #

Includes data sources which describe environmental parameters rather than information related to the user's playthrough. Mostly relevant for personalization of first time user experience, the subcategories of contextual information we distinguish are:

  • device information - device OS type & version, memory size, display resolution, ...
  • acquisition (on Android) - country, paid/organic, CPI or CPA, source parameters, ...
  • retargeting information (cross-promo) - source app & parameters, source user profile, ...

This data may be provided in a denormalized form - attached to behavioral data as parameters of certain events (such as registration or first session), or directly by granting us developer access to your third parties' management accounts.

IAP catalogue #

Usually managed by your liveops designers, IAP catalogue includes definitions of most or all of your IAP bundles, whether it's starter packs, best value deals, seasonal bundles or anything else. Automated integration is required for any personalization use-cases, as we optimize for long-term impact and require parameters such as:

  • bundle contents,
  • availability / exclusivity,
  • and others...

As these definitions typically live in CRM systems, we provide some out-of-the-box integrations to get you set up faster:




Firebase Engage

If your CRM system isn't listed above, or if you use an in-house solution let us know so we t streamline the experience for you.

Identity directory #

This dataset represents a way for you to forward us any additional aggregated information about the players that you have readily available and may include:

  • player ID mappings,
  • tester flags,
  • cheater flags,
  • custom segmentations,
  • custom AB testing attributes.

Data definitions #

In the initial stage of our cooperation we always ask to share any readily available technical documentation and interpretation guidance relating to behavioral events or contextual parameters, typically in a form of a document, spreadsheet or a more formal specification - for example a JSON schema.

Data definitions represent a living document which is to be kept up-to-date at all times to avoid operations issues. For best experience, we recommend notifying your Assetario Success Manager about any major remote changes or releases ahead of time, so we can facilitate a seamless transition.

Learn From

After connecting your data and understanding your data specification we move on to the next step, which is designing models for the specific personalization use-cases you requested. Our data scientists train our platform's models for your specific game using past data as a part of the process.

Pre-training ingestion #

  • Amount of data depends on retention curve, conversion, monthly new users and data consistency.
  • The exact time period varies by app, 3 - 5 months of past data usually suffice.

Feature engineering & feature selection #

  • We start with more than 150 features in our machine learning algorithms at the beginning, adding and removing features automatically as our algorithms adjust to your optimization goals.
  • Our models evolve free of charge, as we adjust to the latest trends and changes you make in-app.

Behavioral modelling #

  • Our AI-driven approach automatically finds similarities between users through contextual and behavioral data from the past. It’s like running thousands of informed AB tests simultaneously, without the downsides.
  • We help you decide the optimal course of action on a per-user basis, automatically and in real time.


The process of value inference comes to play once we have the pre-trained models. By inference, we mean predicting or recommending specific values on a per-user basis, in batches or real time.

Types of values #

From the perspective of usage, there are two types of output values we work with in general - predictions and recommendations. The difference between the two is explained below.

LTV predictions #

Predictions allow us to estimate an attribute of a cohort or an individual using behavioral and contextual data, e.g. customer lifetime value (LTV) after only 3 days since installation.

We predict the value of a measurable KPI, where the output can be either

  • An exact number - predicted LTV = $132.99,
  • A segment (also called class) - predicted spending group = high spender ($100+),
  • A ranking (relative scale using historical data) - predicted spending power = 8/10.

Predictions serve for the purpose of a baseline understanding of a user, which can then be acted upon in any way you'd like, as the meaning of the output value is always strictly defined. As an example, LTV could be defined as "total net revenue accumulated by a single user over the whole course of their usage".

Notice that the output specification does not rely on any app specifics at all, because it is an industry-standard KPI which does not require interpretation, only details.

iOS 14.5 and beyond, App Tracking
Transparency (ATT) and SKAdNetwork (SKAN) #

With Assetario, you don't have to sacrifice quality for privacy anymore. Our solutions are fully integrated with SKAdNetwork, designed to be platform-agnostic and only use anonymized user identifiers when it's absolutely neccessary.

Curious to learn more? Tell us about yourself and download our iOS 14.5 readiness deck!

IAP recommendations #

Recommendations, on the other hand, take a bit more. Besides just understanding, our service is also part of the actionable.

Instead of providing a predicted value of a measurable KPI, we:

  • first define success criteria (e.g. 5% LTV uplift)
  • then recommend specific, remotely configured IAP offer IDs to be shown during any in-app interaction involving a purchase possibility.

Every IAP type (also called monetization track) such as

  • Pop-ups / Banners
  • Subscriptions
  • Starter Packs
  • Shop Offers

and many others in your app are treated separately and predicted using a unique machine learning pipeline, putting emphasis on systemic treatment effect and long-term impact on KPIs like LTV, retention or conversion, whether that is Day 14 or Day 360 for you.

Integrations #

To help you set up faster, we have several no-code integrations. The only thing left for you is to give us a set of valid access credentials to your CRM:

If you'd like us to send the outputs anywhere else just let us know, we'll gladly extend the list.


For direct API access, see our API documentation on swaggerhub. You're a single API call away.

Contact us for an access token and an interactive tour!


Although ease of integration is one of our core principles, we completely understand that some problems require a human guiding hand.

Keep the heartbeat going #

We believe that communicating changes clearly and ahead of time is critical for liveops success.

Sharing is caring #

Share what works for you with us, so we can help build on a strong foundation.

The Secret Sauce #

We will always leave the offer design to you. The bundles that you create - their content, price, discount ratio, placements and such, are all up to you. You are the secret sauce.

Content strategy #

Share your content strategy, content delivery cycle plans, design principles, intentions and your future goals. We will pitch in with interesting analyses, observations and suggestions based on data.

Seasonal and special events #

A seasoned liveops designer should and will take advantage of special events, as will our platform.


We can learn a lot from the journey, but without a goal we're just wandering. It's important to have clear goals and look back - both at failures and successes.

Don't trust us, trust the data #

We always evaluate ROI from our services using relevant benchmarks and AB test results. We ask the right questions, and strive to hit the perfect balance of critical thinking and constructive solutions. Let's not reinvent the wheel, but instead explore together!

Testing in production #

Related only to recommendation solutions, as predictions don't have any direct impact on users.

Together, we:

  • start with an AB test on a subpopulation of players,
  • test and then launch the integration,
  • make sure to monitor performance closely,
  • analyze impact on cohort-level KPIs.

We always provide a standard monitoring overview with uptime and all the must-have KPIs - contact us if you'd like a demo. As most games have weekly performance cycles, we recommend grouping results by 7-day intervals, or longer.

As soon as you see the results and deem them confidently satisfying, you just say the word and the personalized group will be seamlessly scaled up according to your directions.

You are in control #

We never touch a user without your explicit directive. Every aspect of the cooperation needs to be communicated and handled together, as a team - so we can do our best for your users.

Reach out a helping hand! #

Assetario always provides a dedicated Success Manager to guide you through the whole integration process, day-to-day operations and impact evaluation. We believe that in order to provide a great experience the human factor can't be replaced by automation, no matter how sophisticated.

We pride ourselves in our ability to communicate in a timely, precise manner and center our focus on delivering real value, not only "numbers".

Our core principles

Quality at Scale

Individual approach that scales with you.

Ease of Use

No SDK or complex usage guides, as simple and fast as possible.


Performance evaluation is a cooperative process, not just supplier informing demand side.

Evolves With You

Adding new features for genres or monetization tracks as we grow, no extra payments required.


Use only what you really need, we integrate with your existing solutions seamlessly.

Let’s Talk!

Nobody wants to be average! Our AI-driven solutions boost value for you AND your users.

Thank you, we'll get back to you shortly!
There's been an issue, please write to us at matej@assetario.com.