Introduction
Just like “product manager”, the words “product”, “platform” and “feature” suffer from an identity crisis: everybody uses them, but there's not a single real shared understanding of what they are.
The level of the problem though is a lot bigger, since these words are EVERYWHERE. As a product manager, you will encounter them daily.
When you hear the words “product”, “platform” or “feature”, there's either one of two things happening:
- The person saying it hasn't really ever thought about the difference and uses them interchangeably, usually parroting what they heard somewhere else
- The person saying it has a very clear interpretation of what it means and assumes that it is the universal truth. As a result they will be very disappointed if later on a misunderstanding ensues following you not being aligned with their definition
With this article I'm going to add to the fray by specifying what my framework is for differentiating amongst the three (because yes, they are different). I unfortunately will not promise you that my interpretation is what most people share. In the two cases above, I realized that actually a lot more people have no clear distinction and use the words interchangeably because they have never thought about them.
If you want to be a great product manager, though, you will need to use them properly.
So let's get started.
Product
“You're a product manager, right?”
Now if you are, you would assume that you manage… a product?
So it's pretty important to understand what a product is.
My definition is the following:
“A product is a single end to end experience with a defined interface and aesthetic that at least one user can use to have one of their needs fulfilled or to complete a task”
Let’s break it down:
- End to end = this means that the user ONLY needs to use the product to achieve their need, nothing else
- This is the major difference versus a feature
- Experience = This is a little generic but includes everything from hardware to software to services. In general all products include something changing hands (whether it is physical or bits) which I consider an experience. Using a hardware piece is an experience. Buying a dataset from someone (in B2B case) is an experience.
- Defined interface and aesthetic = there can be two products that fulfill the same need, however if they have two different interfaces then they are two distinct products. With “interface I define any exchange surface. For example:
- Two different models of a Dyson vacuum cleaner that look different (but fulfill the same need) are two different products. Two different units of the same version of the vacuum cleaner, even though they are two distinct objects, given they look the same, are interacted with the same way and fulfill the same need, are the same product
- Each web application usually has APIs. You can use the web UI to use the product or hit the APIs directly through code. Those are two different interfaces, so it’s two different products (though effectively the UI product uses the API product, meaning that the API product serves both the user and the UI)
- User = there needs to be at least one user person that this works for, and only one user persona is sufficient for there to be a product
- This is a major difference versus a platform
- Need fulfilled / completed a task = This part highlights that the product does need to solve for a need or a task that someone wants to be done. There could be an amazing robot that just spins around in the middle of the street. If it is not fulfilling anyone’s task (it could provide entertainment maybe?) then it is not a product, given it doesn’t really solve for any need
- This is usually the difference versus a… research project! Many people in companies don’t realize they’re working on research projects instead of products… though that’s a story for another time
What are common things usually associated with products that are missing from my definition?
- Products need to have a P&L (Profit & Loss) associated to them, or in general need to generate revenue
- FALSE according to my definition
- Usually the most important products for a company would have a P&L attached to it, however, there are also many products which would not. For example, internal tools, websites, customer support portals, specific APIs etc (specific examples later). I see all of them as products though they don’t directly generate revenue individually
- Products need to serve customers outside of the company they are produced by
- FALSE according to my definition
- Products can also cater to people inside the same company. This is the case for internal tools in technology companies, or for a homegrown HR software for example
Common mistakes in defining a product:
- Single experience = Sometimes people define multiple products as a single product, ending up mixing together different experiences which are individual products in themselves. For example, the “Robinhood app” (a stock trading application) is not well defined enough to be a product. The Robinhood iOS (phone) app is a product and the Robinhood web (HTML website) app is another product. Their main functionality is the same but they are two different experiences, to the point where they actually have different features (usually features launch either first in the mobile app or first in the website, according to which feature it is).
Feature
“ A feature is a particular functionality of a product that, in itself, cannot fully fulfill the needs of even one single user, though enhances the experience”
The definition of feature is really dependent on the definition of product. A feature is a part of functionality of a product that, taken individually, is not a product. Usually, the requirement of the product definition that the feature fails to fulfill is the “end to end” (see above).
Necessary vs incremental features
Some features are necessary for the product, meaning the product can't exist without them, and some are additional / incremental, meaning that the product could actually exist without the specific feature, though the feature SHOULD (not always the case…) make the product better.
For example, in Robinhood (the same trading application mentioned above), the “buy” button is a necessary feature. You cannot buy a stock, which is one of the main tasks that the Robinhood app allows you to do, without the buy button (well, technically you could just through the APIs that the app uses, though that is the API product, not the app product). The AI summary of the news surrounding a stock (which was launched in 2025) is an incremental feature, given you could fulfill the need of buying a stock without it (as all Robinhood users did for many years when the feature didn’t exist).
Platform
This one is my favorite given that I consider myself a Platform Product Manager!
“A platform is a single product that serves more than one user persona or other product fulfilling more than one distinct need and that enables other products to exist because of it”
- Single product = It is key that the platform is a SINGLE product. Often products and companies are mixed up. A company can have a few customer facing products which are internally served by a single internal platform. That doesn’t make the customer facing products also platforms
- People tend to consider Lyft and Uber as platforms, but according to this definition they are not, given they have distinct apps for drivers and riders, meaning they have different products for each user. Internally, they could have a platform layer that serves multiple teams that help run the apps, though the apps themselves are not platform products
- More than one user persona = most products have a lot of users, though the characteristic of a platform is that it has to serve at least two entirely different types of users
- Fulfilling more than one distinct need = similarly, different types of users necessarily have different types of needs. To some extent this is a repetition of the previous statement, though it’s helpful to mention both users and needs to reiterate the point
- Enables other products to exist because of it = platforms are things upon which other products rely on. This is also a duplication of the “more than one user” statement above, given a downstream product, in broad terms, is also a user of the platform. A platform should have at least two distinct products built on it. They could be from other companies (customer facing platform) or from the same company (internally facing platform)
More extensively, there are companies that are defined as platforms. However, I don't think that is a distinction that can be made, given the boundaries of a company are fairly undefined. What makes the above definition of products so powerful is that it clearly specifies its boundaries (in the form of an interface). Taking a product, then you can define whether the product is a platform or not.
From my perspective, defining whether something which is not a product (for example, Uber or Lyft), is a platform or not is irrelevant and not productive.
Most of the features of a platform rely somewhere in the gray area between necessary and incremental. Given the many uses (and users) of a platform, usually removing one feature will not make the platform useless, though it might make it lose a few of the user types. That’s where a good product manager comes in; they should decide on the tradeoff between complexity and effort of adding a new feature and customers that it can win.
Are AI agents users?
Lately AI agent use has been scaling. A relevant question is whether AI agents are users or not. Given they are fairly independent entities, even though they usually would be working for a person , I do consider agents to be “users”. There already are and increasingly will be products which are entirely dedicated to being used by agents. MCP servers are an example.
Products, features, and platforms built for agents have different types of surfaces and aesthetics but they fulfill the same needs that a user may have. Already I notice a lot of documentation pages are being written to be primarily consumed by agents.
Examples
I’m going to provide a set of examples around the iPhone, given it is something that I imagine most readers are familiar with. Given I have limited space, I may cut some corners and be overly simplistic, though hopefully you get the idea
- iPhone (the hardware device) is a PLATFORM that serves on one side the end users interfacing with iOS and the apps through a touchscreen and on the other serves the apps that are running on iOS and use the sensors on the phone
- That’s why it has a touchscreen FEATURE for the user and a GPS FEATURE for the apps. The end user doesn’t use the GPS directly, the app does. The GPS feature is not necessary since the platform still works for a lot of users without it
- The end user uses the maps app (which is a PRODUCT) to see where they are (which is one FEATURE of the app) and to get directions to where they want to go (which is another FEATURE of the app). The maps app itself relies on the GPS feature of the iPhone to function.
- You now also see ads on the maps app, in the form of featured destinations or pins. So it is a platform? I would say not, since the ads providers submit the ads on an ad portal, not on the app directly.
- Ads portals are complicated PLATFORMS, though would take me to long to explain here why
- How do developers actually create the maps app? They use the iOS SDK which is a PLATFORM, because it serves the needs of many different developers that create many different types of applications with wildly different functionalities
- To learn about the functionalities of the SDK, developers need to read the documentation of it, which is a PRODUCT. Even though the developers are creating many different applications, they interface through the documentation through the website to understand the functionality of the SDK, which is a single purpose. So it is a PRODUCT (which has no direct P&L for Apple, though I’d say is a fundamental part of the ecosystem). Why is it not a FEATURE of the SDK? Because I can read all the documentation and learn everything about the iOS SDK even if I never actually use the SDK. Learning about it could be a single and worthwhile objective for a user
Necessary vs incremental feature examples
Not all features are born equal. The easiest way to understand whether a feature is necessary or incremental is to see what happens when it doesn't work. Even that, though, is not a binary distinction, given some features cannot work for a while even though they will render the product useless in the long term. A few examples from my career:
- At Mythical, we lived through a few AWS outages (Amazon Web Services, the backbone of a lot of digital products where cloud functionality lives on). Even though we weren't on AWS, A lot of our providers, including the login providers, were not reachable. As a result, When trying to access the website you would be stuck on the login page and we would not be able to log you in. The product was effectively unusable, just like the majority of internet apps during that time. Logging is a necessary feature for an online marketplace.
- At Lyft, my team once discovered a bug for which, on Android phones, a particular prompt panel that would ask you whether you wanted to get a nicer ride was not displaying when it should have. That bug had been around for a few years and no one had discovered it before. Clearly a prompt panel that asks you to upgrade your ride is an incremental feature. I'm sure no user missed that (though Lyft might have missed some revenue because of it).
- Also at Mythical, a few weekends after we had just launched the first version of a marketplace, we realized that users were not able to cash out. That had been working fine for the first few weeks but suddenly had stopped working. The marketplace was still usable and transactions were still happening. However users were starting to complain about the inability to cash out. It was indeed a major issue and we quickly identified the solution and fixed it (on Saturday morning, these issues always happen on the weekend). So are payouts a necessary or an incremental feature? They are a case of a non-immediately necessary feature. If word had gone out that we did not have payouts, then likely the whole marketplace would have stopped working. So even if the marketplace can function without it, ultimately a major need of a marketplace user (getting paid for what they are selling) would be not addressed without the feature and the product would stop being successful
Conclusion and how these definitions aid in scope assignments in companies
Now that you’ve read this, a worthwhile exercise is to pick something that you know well and identify whether it is a platform, a product and which are its features. Also identify the users of the product and the need that it is fulfilling for the user.
With this structure, you should start seeing clear boundaries of ownership, which is what, in companies, is assigned as “scope” of product managers and teams.
Initially companies create teams and assign them (with product managers) to a specific product or feature (remember that a platform is a product, so for scope assignment there’s no need to break the division out).
As companies are small, product managers tend to have to cover more surface area and might have multiple products assigned to them. As companies grow and depth / precision becomes more important, bigger and bigger teams are assigned smaller and smaller scope, such that instead of products, a single product manager and team might end up only being responsible for one feature.
That is one of the reasons that working in bigger companies is alienating to some and overall has less opportunity for career growth. As companies grow they want all their features that work to be best in class, so they hire the best people they can find that are specialised in that one single feature, paying them handsomely. However, once hired on a feature that already kind of works, the team has nothing really to do, so they end up either not working hard, or making up more features and needs that don’t really exist, adding bloat to the product overall.
That’s a story for another time, I’ll make sure to write a follow up to that.
My suggestion is that, as a product manager, you want to ensure that your scope is enough to keep you motivated and to keep you learning, so aim for roles which at least cover a product and not just a feature.