Build The Right Thing: A Feature Strategy Guide by Spotify Sr PM, Dejan Krstic

Product Strategy

combination of 4 types of product work

1.Feature strategy
Feature strategy is one of the pillars of the broader product strategy work.
The feature strategy focuses on improving our ability to create and capture value.
2. Growth strategy
A growth strategy on the other hand focuses heavily on maximizing the value proposition of a product.
Effective growth strategies link acquisition, retention, and monetization. It is important to move away from funnel thinking to growth cycles where there is entry, action and exit that will fuel the next cycle, and so on.
3. Product market Fit expansion strategy
A product market expansion strategy attempts to add value in two ways: 1. adapting products to new additional markets, 2. adding new additional products to overcome saturation, such as market saturation.
Market capture reaches a natural limit, or product saturation, when the product becomes fully optimized for its use case.
4. Scaling.
When you invest in supporting processes infrastructure and strategies that support previous levels of strategy, scale and technology, platformization of technical depth management, UX monetization, process scaling, process improvement, value stream mapping assessment, user scaling are key pillars.

Feature Strategy

improve the ability to create and capture value.

Features can create value in three ways

  1. Acquisition. Acquiring new users

  2. Retention. Retaining existing users

  3. Monetization. Monetizing users

It is critical to evaluate and improve existing features as this will inform your future build strategy and help you develop new features and continue the post-launch performance evaluation cycle of new features in order to achieve functional product compliance.

Evaluate performance after launch

Source of Insight:

  • Strategic Insight

  • User Insight

  • Data Insight

Qualitative Evaluation

Qualitative assessments are very important because they help you with the following questions

User Problem Serverity

User Category

  1. Core Users

    Users for whom your feature design to create value

  2. Adjecent Users

    Users who get some value from your feature, but for whom it was not designed

  3. Non-Adjecent Users

    Non-users of this feature

Feature Pre-Mortem

These are just guidelines and you should define results and metrics that are specific to your company and product. You can test and compare your assumptions with actual results after implementation. This will help reinforce the idea of the product.

Before you roll up your sleeves. you have to check that your ideas are pressure tested.

Developing New Features

Feature Release Method

  • Experimental Release

    Focused on running experiments to validate key assumptions in your feature to learn and shape it over time

  • MVF Release

    A fully functioning version of the feature is released, but with the minimum functionality, in order to validate the core value proposition.

  • Phased Release

    Release a more robust version of the product which is broken down in phases to allow features to be released when ready.

These methods do not compete with each other. The feature must go through each of these release methods based on the ambiguity spectrum.

Ambiguity Spectrum

If you have high uncertainty and not enough information from the sources, then the experimental release is your choice, as you need to find out which product with the right features will solve the biggest problem for our users.
If you have high confidence and have tested your hypotheses with other releases, then a phased release is your choice as it focuses more on building a feature or product in the right way to provide added value.

Experimental Release


  • Learn and shape the feature iteratively with user behavior


  • List all assumptions to believe in the success of the feature

  • Identify assumptions, turn them into experiments that can be validated by users

  • Refine features with every experimentational outcome

Hypothisis – driven validation

We believe that clearly defined incremental experimentation leads to faster learnings and deeper insights than writing detailed specifications.

Step-by-step guide:

  • Identify your assumptions

  • Reframe assumptions as “hypotheses”

  • Rank them in order of importance

  • Design appropriate experiments

  • Conduct the experiments

  • Synthesize your learnings

  • Act

What is an Experiment?

  • Validating business logic in a spreadsheet

  • UI sketch on paper

  • User flows in Figma

  • Low/no-code prototype with a purpose to validate a specific aspect of the product

  • Anything that makes us learn

MVF Release


  • Validate core user value prop of features using minimal embellishment and functionality


  • Minimize the number of platforms the feature is built for

  • Minimize the number of integrations the feature needs

  • Minimize design and engineering resources needed

Phased Release


  • Building features more fully as ambiguity is low and confidence is higher on the final state of the product


  • Simple Features

    Fully building the feature and releasing it

  • Complex Features

    Break features down into phases which can be released independenly

Evaluate Features

Retention Score

Evaluating Features

This exercise helps us evaluate our features, as calculating the retention rate is the first step, followed by plotting the evaluation against strategic importance.
With this approach, we can evaluate our features and a great tool for this is the feature matrix.

Feature Matrix

As for the features of our project, you have to ask yourself if they are worth investing in or if we should abandon them as they are important to the overall success and health of our product.
If you allow too many features to bleed into your product, the core value proposition and vision can easily become blurred.

Dejan Krstic, Spotify Sr PM