Skip to main content

Frequently Asked Questions

How is Objectiv different from product analytics tools?

There is a big gap between what data scientists want their data to look like and what data actually looks like when it comes from the current generation of trackers.

Objectiv’s tracker collects clean, well-structured data that’s ready for modeling without any additional cleaning and transformation. It embraces the open taxonomy for analytics, which is our proposal of a common way to collect, structure and validate data. This eliminates the recurring discussion on what to track and how to track it, and enables data scientists to quickly build advanced models by reusing parts of models others have already built.


What is the open taxonomy for analytics?

The open taxonomy for analytics is our proposal for a common way to collect, structure, validate and model data.

Just like the taxonomy of the animal kingdom does for animals, the open taxonomy for analytics describes analytics events in a class system. Each common event type has its own class, which defines its properties, requirements, restrictions and relationships with other classes. It also includes context classes, which describe the context in which an event took place, rather than the type of event that took place.

Check out the Taxonomy section of our Docs to see what it looks like, or try Play with Objectiv to see how it's used

Why an open taxonomy for analytics?

When it comes to product analytics, we noticed that most data teams have similar goals (optimize conversion, increase engagement, reduce churn, etc.) yet their data and models all look different. Models are often built from scratch. Someone in your field has probably already written a very similar analysis, but data teams keep reinventing the wheel because there is no common way to structure and model data. This leads to isolated knowledge and no meaningful way for data scientists to collaborate.

We believe a widespread adoption of the open taxonomy for analytics would be beneficial for the data science community on multiple levels:

  • The recurring discussion on what to track and how to track it can be eliminated because the modeling requirements are carried by the design of the taxonomy.
  • The classification of events & contexts enables trackers to collect validated, well-structured data that is ready for modeling without additional cleaning or transformation.
  • Reuse parts of any model from any source as long as it embraces the taxonomy.


Does Objectiv work for mobile and web?

Objectiv tracker works with any JavaScript-based web and mobile application. It is built on top of Objectiv’s core tracker, which has been designed to be platform & device agnostic and have minimal dependencies. You can combine data from various platforms and devices in your analyses without additional transformation. Check out the roadmap for upcoming tracker development.

Do I need to replace my existing analytics tracker?

No. You can typically instrument Objectiv’s tracker and leave your existing tracking instrumentation alone and both will work fine. You can start small and expand as you go.

Do I need a tracking plan?

In most cases, the recurring discussion on what to track and how to track it can be eliminated because the modeling requirements are carried by the design of the open taxonomy. It ensures that you collect the required data to effectively build models for common analyses (conversion, retention, engagement, etc.). The taxonomy can be extended to also cover less common analyses.

Why do I need to tag sections and events in my UI?

While most trackers simply track everything by inserting a simple script, the data they collect is often incomplete or overcomplete, unstructured and ambiguous. Significant gruntwork is typically required before it can be used for modeling. Objectiv asks the frontend engineer to tag events and sections in the application’s UI. This creates a contextual layer for your tracker that serves multiple purposes:

  • The collected data becomes rich and descriptive. You can pinpoint exactly what event happened, which type it was, in which context it happened and from which location in the UI it was triggered.
  • Your tracking instrumentation becomes debuggable. You’ll be able to catch tracking instrumentation errors early on because it is being validated as you’re developing it.
  • The collected data no longer requires significant cleaning and restructuring before it can be used for modeling because it was collected in a structured manner and validated at the first step of the pipeline.


What does Objectiv’s tracker data look like?

Rich, descriptive, clean and well-structured. Check out our our sandboxed notebook to play with Objectiv. It uses the real untransformed data that was collected with an unaltered version of Objectiv's tracker.

Why is Objectiv SQL-based?

SQL is widely used, versatile, not overly complex, and with developments in cloud technology, SQL databases have become very powerful and scalable. It’s supported by all major cloud providers and easy to self host.

Objectiv Bach, our modeling library, features an SQL abstraction layer that enables you to use Pandas-like operations on your full data set, combining the modeling power of Pandas with the versatility of SQL

Which Pandas-like operations can I use to build SQL models?

Objectiv Bach has support for an initial set of Pandas-like operations. For all supported operations, check the reference in the Modeling section.

Can I run these models in production?

Yes. You can use for example DBT to run these in production. We plan to offer seamless integration for DBT in the future. Check out the roadmap for details.

Does Objectiv’s tracker collect personally identifiable information? (PII)

Yes, but not before the user has given explicit consent. In order to identify returning users, Objectiv creates a unique user ID. Objectiv only tracks what is strictly required in order to model effectively.

Does Objectiv store or share any user data?

No. Objectiv’s data is strictly first party and we do not have access to the generated event data. You’re free to choose how and where to store your data and have full control over who gets to see and use it.

What’s the default expiration date on Objectiv’s cookies?

One year. You can change the expiration date in your Objectiv configuration.

What kind of license does Objectiv use?

Objectiv is licensed under the Apache License, Version 2.0 . It is a permissive license whose main conditions require preservation of copyright and license notices. Contributors provide an express grant of patent rights. Licensed works, modifications, and larger works may be distributed under different terms and without source code.


Does it work with my stack?

Objectiv currently requires a PostgreSQL database and works in any environment where you can run containers. More databases will be supported in the future. Check out the roadmap for details.

Does it scale?

Objectiv’s architecture is built to scale. It is only restricted by the capacity of your databases. Collection is stateless and can be run in parallel. Objectiv currently uses PostgreSQL as a database backend and features an SQL abstraction layer that enables you to run Pandas operations from your notebook on the full dataset. We have plans to support other databases as well. Check out the roadmap for details.

Can I run it on-premise?

Yes. We don’t have any plans for a hosted version for the near future. Check out our quickstart section of the introduction for installation instructions.


Is it supported properly?

We’re well-funded and are actively working on the project on a daily basis with 8 full-time team members. Check out our Github repo to see our activity.

What’s your business model?

Objectiv ultimately wants to monetize with paid offerings to users of the ecosystem that we believe will emerge around the open taxonomy for analytics. Objectiv’s core library and the taxonomy are and always will be open source and free to use.

How do you plan to develop the project?

We want to make sure the taxonomy enables collection of data that meets the requirements of the data science community for effective modeling. As such, we will be looking to create a continuous feedback loop with said community to ensure it meets their needs. As widespread adoption is a goal, we will put effort in seeing that it can be used for a wide range of DS use cases.

Do you plan to develop a taxonomy for other areas?

The current version of our taxonomy is built for product analytics. With it, we’ve built a solid foundation that can be used to create taxonomies for other fields, e.g. Payments and CRM. If you’ve got a lot of experience in a particular field and would like to contribute, please join us on Slack.