The world of product management is pretty diverse. While people often slice product management roles by things like Industry and B2C vs B2B, a more useful categorization is Systems vs Feature product managers.
Feature product managers are the most common type – when someone new to product management thinks about the role, they often refer to these. I would loosely define a feature product manager as anyone that is building a user-facing feature and has a user interface. This could be building the search experience on TripAdvisor, the Reels experience on Instagram, or a dashboard for business owners that use Intuit Quickbooks. While the TripAdvisor and Intuit examples seem different (B2C vs B2B), from a product management skill-set standpoint, they’re fairly similar – both require you to understand what your customers are looking for and build a user interface/interaction that solves that need.
Systems product managers are the ones who typically build systems, which are either products themselves, or are foundational systems powering user-facing products:
- A classic example of a system that is a product is ad attribution. If you’re advertising through Google, you want to be able to measure the impact of your ad campaigns. An attribution product would take an ad exposure data stream (for example, when was an impression served to a user), an advertiser conversion data stream (when a user made a purchase on an advertiser’s website), match these two data streams and produce a result for the number of purchases driven by ads. Note that this isn’t a “behind the scenes” product. Attribution is one-half of what makes any advertising product work. Without attribution, you don’t acquire or retain advertisers, especially large enterprise advertisers.
- The second type of system is a foundational behind-the-scenes product like the recommendation algorithm for Facebook News Feed or the ad targeting algorithm when you search on Google. These are highly critical systems that are at heart of making user-facing products succeed.
- The third type of system is an internal system – for example, an internal billing system within TripAdvisor which enables revenue collection from TripAdvisor advertisers.
With the emergence of more technical systems over the past decade, there’s been a steep increase in the number of systems product management roles. Given the specific skill set required, it is much harder for hiring managers to find the right candidates (validated by my hiring experience + that of other hiring managers I have worked with), and this presents a unique opportunity for candidates with the right skills. To understand why, let’s look at:
- Similarities between systems and feature product managers
- Differences between systems and feature product managers
- Getting into a systems product management role
The above examples of different systems products (ad attribution, feed recommendation, ad targeting), show us that there are a meaningful number of similarities between the two roles.
Systems product managers need to understand their customers deeply. It’s a common misconception that systems product management is about aggressively optimizing a single metric (eg. number of likes or number of ad clicks). While it is true that metrics are a big part of what systems optimize, a deep understanding of users is required to identify what metrics to pick. For example, if you pick number of clicks as your single metric to optimize an ad targeting system, you will end up with clickbait-y ads getting the most clicks and advertisers not seeing any real value, and therefore churning.
Similarly, if you build an ad attribution system, you need a deep level of customer understanding to figure what attribution product to build. For example:
- Do your advertisers want to measure online transactions or do they want to measure how many people walk into their store?
- Do they want your attribution product or do they want you to support a third-party provider?
- What is an attribution logic that you can effectively sell to an advertiser (eg. give all credit to the last click vs give credit to all impressions and clicks before a transaction)?
Answering these foundational questions requires systems product managers to have a deep level of customer understanding, just like feature product managers.
Systems product managers still need to collaborate cross-functionally. The set of stakeholders you collaborate with could be different. For example, to upgrade your ad targeting logic, you would work closely with a data scientist. To build an attribution system, you would work closely with customer success and sales, both to understand the advertiser and to take your product to market.
Systems product managers also have an iterative process to build products. A feature product manager may begin with a simpler MVP user interface to validate a hypothesis (eg. whether more personal content leads to higher engagement in News Feed), and a systems product manager would take a similar approach for an attribution product. This MVP, for example, could take two Excel sheets as data streams and use a simple Jupyter notebook to generate an output that can be shared with advertisers.
While there is overlap in terms of core skills between the two types of roles, there are also differences.
Systems product managers often work on more technical products and need a strong understanding of technical and data concepts. Let’s take the example of building an attribution product. As the product manager you must be able to predict potential challenges. For example,
- The quality of data streams might not be great,
- When you are trying to “match” two data streams, getting to high match rate can be a challenge,
- If you are collecting two data streams, one from mobile and one from web, the user IDs you receive from these streams might not be compatible.
You are not expected to be as technically savvy as an engineer or as data-savvy as your data scientist, but you should be able to have a reasonable conversation about topics with either of them and be able to (over time) build high judgment around these issues.
Systems product managers are typically more analytical and data savvy. Not to say that the role does not require creativity – it does – analytical skills are a huge asset. For example, let’s say you receive an escalation from a disgruntled user about receiving a bad ad. How would you go about debugging this? It would typically require you to understand how the ad targeting system is designed, do some analysis yourself, and work with an engineer or data scientist if you can’t figure it out yourself. You are also generally expected to do some analysis on your own and not rely on an analyst/data scientist every time you need to dig into something (this is also generally good practice for any product manager).
Systems product managers are able to take business context and customer need, and translate it into a systems product. This translation takes the form of either being able to envision different technical pieces required to create a product (for example, an attribution system requires collecting data streams, matching data to figure which ads drove conversions, and producing an output report card), or being able to identify a set of metrics that the system should optimize for (for example, say Facebook decides to focus on enabling quality personal conversations, so the News Feed algorithm could optimize for numbers of comments with people you have categorized “close friends”).
Getting into systems product management roles
The core skill that would help you differentiate as a candidate is experience working on systems or data problems. If you have built a technical system in a past job (as a product manager or in a product-adjacent / engineering/data science role), that’s a big plus and gives hiring managers confidence that you understand what you’re getting into. So shows any data-heavy work you have done. Something more straightforward like an analysis of an A/B test probably doesn’t make the cut but if you have for example changed the logic / algorithm for a certain feature or product (even if on a small scale), that helps.
Another critical dimension depending on the role is whether you have built systems that impact user / business outcomes. A billing system that collects revenue from TripAdvisor advertisers is absolutely valuable and business-critical, and it could help you break into a subset of systems roles, but it’s probably a jump to go from there to building a News Feed recommender system. What might be a smaller jump is building the underlying data system that is used by the News Feed recommender (for example, a data pipeline that generates features used by the model). So it’s helpful to think about what your ideal role is and then figure the path to getting there.
It also helps a lot to transition within your company. If you are a data scientist, engineer or feature product manager that has built trust with the Product org, you might be able to transition to a systems product role internally.
Systems product management roles are fascinating, particularly if you are technically inclined, and will continue to grow in number with the emergence of more technical products and the growing momentum behind artificial intelligence. There are a variety of systems product management roles you could take on, depending on your skill set and interests. It’s a good exercise to look at job openings and try to get a feeling for what types of products interest you, and then embark on your journey to get there. There is high demand and less supply of high-quality systems product managers, and it can be an incredible career opportunity to tap into.
If you like my writing, do check out my newsletter, Unpacked. I write weekly in-depth analyses of current tech topics, delivered to you in the form of a 10-minute read. Best, Viggy.