Proprietary Excellence
Special Edition: Examples of innovative proprietary technology built by the world’s leading technology companies
🔒DoP Deep dives including Special Editions like this are available for paid subscribers. To unlock your newsletter and get access to this plus the Knowledge Series and all Deep Dives, you can upgrade below. Or you can learn more about what you get as a paid subscriber.
Hi product people 👋,
Building proprietary technologies is often cited as one of the most powerful ways tech companies can defend themselves against competition and assert their position as market leader.
Proprietary technology doesn’t necessarily mean a company owns a patent for it; it can simply mean they have built something that is difficult to copy which has the potential to generate additional revenues or boost conversion and retention in unique ways.
In this Special Edition, Proprietary Excellence, we’re going to explore some examples of how tech companies are leveraging their most talented product / engineering team members to build proprietary technology that boosts their product processes internally and provides value to customers externally.
We’ll look at some recent examples from some of the world’s leading tech companies which includes proprietary technologies for important product areas including:
How Spotify’s Confidence platform utilizes a clearly defined set of rules to determine whether a feature should be launched
A server driven UI system that simplifies design changes across multiple devices at once
Why Pinterest’s engineers decided to build generative AI models that don’t create images from scratch
How Uber’s design component management system goes beyond simply tracking component usage
…and more
Spotify’s product decision making and experimentation platform
Proprietary technologies: Confidence platform for A/B tests, feature management, recommendation system
Confidence is a proprietary experimentation platform by Spotify. The company uses it across much of its architecture for tasks like product decision making, experimentation, feature flagging and cross-platform consistency.
The technology started life as an internal tool called ABBA (their original proprietary testing platform inspired by the Swedish legends) which included experimentation capabilities and feature flagging. ABBA laid the groundwork for what would eventually become Confidence.
How Spotify uses this technology to make product decisions
While Confidence now includes a series of different proprietary technologies that it deploys across its business, one of the most powerful is its application in product decision making.
Spotify uses Confidence to assess proposed product decisions and limit the number of “incorrect” product decisions. This research paper outlines how the technology works in more detail and to supplement the study, Spotify’s data scientists have also explained in more detail how the system works.
The platform considers four types of metrics in its decision-making process:
Success metrics - metrics that you want to improve
Guardrail metrics - metrics that you don’t expect to improve, but yuo want evidence that they’re not deteriorating by more than a certain margin
Deterioration metrics - metrics that shouldn’t deteriorate
Quality metrics - metrics that validate the quality of the experiment itself
Based on the outcome of the tests for these metrics, the Confidence experimentation platform will then recommend a decision. Spotify has a clearly articulated set of rules that are used to determine whether or not a feature should ultimately get shipped.
Here are the rules Spotify uses in Confidence to determine this: