DoP Deep - How are companies using machine learning?
ML-powered product features from top companies like Strava, TikTok, Stripe and more
🔒DoP Deep goes deeper into the concepts and ideas that are covered in the Weekly Briefing to help you learn lessons from the experiences of top tech companies. If you’d like to upgrade to receive them you can do so below. Or you can find out more about what you get as a paid subscriber here.
Hi product people 👋,
In the weekly briefing we mentioned that fitness app Strava has recently unveiled a series of new AI features and one of these in particular is worth exploring in more detail. “Digital doping” is the practice of users uploading fake runs using e-bikes and other vehicles to jump to the top of leaderboards. To combat this, Strava developed a new feature using machine learning that can quickly identify and remove the runs.
Machine learning is a subset of AI and while in previous deep dives we’ve focused more broadly on AI features and strategies, in this Deep Dive, we’re going to look specifically at how the world’s leading product teams are using machine learning to shape their roadmaps and features.
We’ll explore specific examples from top tier companies with some new use cases you might not have considered before. You can use these examples to help you inform your own product’s ML strategy, too.
Coming up:
How this analysis is structured
A deep dive into some new ML-powered features from specific companies including: Strava, TikTok, Stripe, Spotify, Snapchat and other new companies you might not have heard of
How to introduce machine learning in your organisation
Resources and tools for further reading
The full list of companies and machine learning features - with links to each one
How this analysis is structured
For this Deep dive, we’ve picked a mix of machine learning powered features from top tier companies and new startups to give you some inspiration for your own roadmap.
And to help you digest them in a more structured way, we’ve categorised each of the features using attributes including the following:
Machine learning application - how is ML applied to the feature? We’ve intentionally picked a wide variety of ML applications to ensure this deep dive is as thorough and varied as possible to ensure you get a varied sample of how companies are using ML
How it works - a description of how the ML product feature works
Industry - including B2B, B2C, SaaS, healthcare, advertising, fashion and others
Link - a specific link to each of the features to help you learn more about it
Even though ML is a subset of AI, machine learning itself and how it is used by companies is so broad that it can be sub-categorized further.
To help structure this, the machine learning application section is broken down further into the following use cases:
Anomaly detection - this involves using machine learning models to identify unusual patterns or outliers in data that deviate from the norm. In product, anomaly detection could be used to automatically flag and alert on suspicious activity, errors, or potential issues. Traditionally, this is used by finance companies and banks to detect malicious behaviour but there are more innovative ways to use this too. We’ll cover these.
Recommendation systems - by learning patterns in user behavior and preferences, machine learning enables highly relevant and personalized recommendations. Companies like Spotify and Soundcloud use ML to power their music recommendation engines, for example.
Generative AI - generative models in machine learning can create new data (like images, text, audio etc.) that resembles the training data. Products could use this to generate realistic content, designs, or suggestions. We’ll explore some of the ways companies are doing this.
UX enhancements - machine learning can be used to personalize and optimize the overall user experience. For example, by learning user preferences to provide customized recommendations. We’ll explore some ways companies are doing this.
Predictive analytics - machine learning models can be trained on historical data to predict future outcomes. Products can leverage this to provide forecasting capabilities, estimate demand, proactively identify risks, or offer intelligent decision support.
Security enhancements - machine learning can augment cybersecurity by learning to detect threats and anomalous behavior. Products may use it to enable intelligent threat detection, automate security policies, or adaptively respond to emerging vulnerabilities.
Now that you’ve got a pretty good idea about how we’ve structured the analysis, let’s take a look at some of the ways top tier companies are using ML to power their features.