Feature Discovery
Feature discovery is the process by which users become aware of product capabilities they have not yet used and begin engaging with them. It is the bridge between a feature existing in the product and a user actually knowing it is there, understanding what it does, and incorporating it into their workflow. A feature that is built but never discovered delivers zero value to the user and zero return on the engineering investment that created it.
This sounds like a trivial problem but it is not. The average SaaS product ships dozens of features per year, and most users interact with a small fraction of the total feature set. Pendo's State of Software report found that the average SaaS product has 44% of its features rarely or never used. This is not always because those features are bad. It is often because users never learned they existed, or never understood why they mattered for their specific use case.
Feature discovery sits at the intersection of product design, user experience, and communication strategy. It requires the product to surface capabilities at the right moment, in the right context, with enough explanation to make the value obvious. Sending a mass email about a new feature is awareness. Getting a user to adopt that feature in the context of their actual work is discovery.
Why it matters for SaaS
Feature discovery directly drives the metrics that determine SaaS business health: retention, expansion revenue, and customer satisfaction. Users who discover and adopt more features are stickier because they have invested more in learning the product and have more workflows that depend on it. Product insights help teams track which features are being discovered and which remain hidden. They are more likely to upgrade because they experience the value of premium capabilities firsthand. And they are more likely to become internal champions who drive adoption across their organization.
The economic impact of poor feature discovery is substantial. If a product team spends six months building a collaboration feature but only 15% of users discover it in the first quarter after launch, the ROI on that engineering investment is a fraction of what it should be. Multiply this across every feature release and the cumulative waste, not in the features themselves, but in the unrealized value from low discovery, becomes one of the largest hidden costs in a SaaS business.
For PLG companies, feature discovery is especially critical because it powers the natural expansion that the business model depends on. A user who discovers advanced reporting upgrades to a premium plan. A user who discovers team sharing invites colleagues, driving seat expansion. A user who discovers integrations connects the product to other systems, deepening lock-in. Every feature discovered is a potential expansion trigger. Every feature missed is an expansion opportunity lost.
How it works in practice
Feature discovery happens through multiple channels, and the most effective strategies use several in combination. The most common approaches include in-app announcements, contextual tooltips, onboarding checklists, release notes, and email campaigns. Each has different strengths and weaknesses.
In-app announcements and modals reach users at the moment they are engaged with the product but carry the risk of interruption fatigue. Contextual tooltips appear when the user is near the relevant feature and have higher relevance but lower reach. Onboarding checklists are effective for new users but fade in relevance after the first few weeks. Email campaigns reach users outside the product but have notoriously low engagement rates for feature announcements, with open rates often below 20%.
The most effective approach to feature discovery is contextual and behavioral. Instead of announcing features to all users at the same time, the product identifies when a specific user would benefit from a specific feature based on their current actions. A user who manually exports data every week gets a prompt about the scheduled report feature. A user who creates the same type of project repeatedly gets introduced to templates. This approach respects the user's attention by only surfacing features when they are immediately relevant, which dramatically increases adoption rates compared to broadcast-style announcements.
Feature Discovery vs Feature Adoption
Feature discovery and feature adoption are sequential but distinct. Discovery is the moment a user becomes aware that a feature exists and understands what it does. Adoption is the sustained use of that feature as part of a regular workflow. You can have discovery without adoption, the user saw the feature, tried it once, and never returned, but you cannot have adoption without discovery.
The gap between discovery and adoption is where product teams should focus their attention. High discovery rates with low adoption rates signal a feature that users find interesting but not valuable enough to incorporate into their workflow. This might mean the feature needs refinement, the use case is not compelling enough, or the learning curve is too steep for casual exploration.
Low discovery rates, on the other hand, point to a distribution problem rather than a product problem. The feature might be excellent but buried in a menu that users rarely visit, accessible only through a keyboard shortcut nobody knows about, or documented in a help article that no one reads. Solving low discovery is often cheaper and faster than solving low adoption because it is a communication challenge rather than an engineering one.
How Floe approaches this
Floe transforms feature discovery from a broadcast communication problem into a personalized, contextual experience. Instead of showing every user the same tooltip about a new feature, Floe's AI agent recognizes when a specific user would benefit from a capability they have not yet used and introduces it naturally within their workflow. The introduction is not a popup or a banner. It is a conversation that explains why this feature matters for what the user is doing right now.
This contextual approach achieves dramatically higher discovery-to-adoption rates because the feature is introduced at the moment of maximum relevance. A user does not need to remember a feature announcement from last week's email. They learn about the feature at the exact moment they have the problem it solves. And because the AI agent can walk them through the feature in real time as part of onboarding, the gap between "I learned this exists" and "I know how to use it" collapses from days to seconds.
FAQ
Why do users not discover features on their own? Three primary reasons. First, cognitive load: users focus on the tasks they know and do not explore menus looking for new capabilities. Second, feature sprawl: as products grow, the interface becomes crowded and new features get lost among existing ones. Third, the cost of exploration: trying an unfamiliar feature requires time and mental energy with an uncertain payoff. Most users make a rational decision to stick with what they know rather than invest in discovering what they do not.
How do you measure feature discovery? Track two metrics: discovery rate (percentage of active users who have seen or interacted with a feature for the first time within a given period) and time-to-discovery (how long after a feature launches before the average user encounters it). Compare these across features to identify which are surfacing naturally and which need active promotion. Segment by user type to see if discovery patterns differ for new users versus long-tenured ones.
Should you force users to discover features through mandatory walkthroughs? Almost never. Mandatory walkthroughs interrupt the user's workflow and create resentment, especially for experienced users who know what they are doing. The most effective discovery mechanisms are opt-in and contextual: available when helpful, dismissible when not. The exception is during initial onboarding, where a guided introduction to core features is expected and appreciated. Beyond that first experience, feature discovery should feel like a helpful suggestion, not a mandatory detour.