Background

A lot of people land on Facebook from outside the app, through a search result, a shared link, or an embedded video. They are not logged in, may not have the app installed, and have not chosen to be on Facebook for the session. They are here for one thing: the video they clicked on.

The product question for the off-platform video work was how to turn that single-purpose visit into something more, without making the visit itself feel like an obstacle course. Every prompt to log in or open the app is friction, and the same friction that helps one user convert pushes another to leave for YouTube or TikTok.

I worked on the web experience for off-platform video, across both mobile web and desktop.

Problem and opportunity

The off-platform funnel is larger than any one team. It has three layers:

• Findability at the top: how a Facebook page surfaces in a search result.
• The page itself: how a logged-out visitor experiences the content they came for.
• The conversion question: when to ask the visitor to open the app, log in, or save their state.

Different teams focused on different parts. A central team owned the findability and SEO work, including taking on the broader scope and stepping in where product teams were not leveraging the opportunity their logged-out experience created. My team owned its own surface, and within that the conversion question was the part I focused on most. This case study is about a narrower slice inside the funnel: the conversion problem and how I worked inside it.

Role

I worked on the off-platform video web team as a product designer, owning the design of conversion patterns across mobile web and desktop. That meant the full set of asks the product could make: prompts to open the app and prompts to log in. The work spanned audits and experiment proposals, the design of new upsell patterns, and shipped web experiences that made each of those asks feel more contextual. Engineering, product, and research partners were tightly involved, and the work also influenced adjacent teams.

Building the product

The three jobs

There were three jobs the team kept circling back to:

• Improving the logged-out experience so visitors get the value they came for and form a positive impression.
• Satisfying user intent so people leave the page having actually done what they came to do.
• Encouraging further exploration: opening the app, logging in, or staying long enough to chain into more content.

These three jobs are not equal partners. Most of the levers available to my team, such as upsells, login prompts, and app-open prompts, sit on the third job. They barely touch improvement of the logged-out experience itself. And there is a real tension inside that asymmetry: how much should you improve a logged-out experience if your job is to convert? Make the page too good and the visit ends with the visitor satisfied off-platform. Make it too constrained and Facebook becomes a walled garden, and visitors leave for competitors whose logged-out experience is more generous.

The question worth holding is how to lower the walls without giving up the goal of bringing people in. Improvement has a ceiling when the goal is conversion, and recognizing the ceiling shaped how I scoped the work: focus the design effort where it could actually move the conversion ladder, and resist the temptation to keep polishing the page for its own sake.

The incrementality ladder

If improving the page is bounded, the conversion work has to be precise about which conversion it is asking for, because not all conversions are equally valuable or equally likely. The work organized around a ladder:

• Open the app is the strongest outcome. The user goes from a session to an installed surface that can serve them long-term.
• Log in on the web is the bridge when opening the app is too much friction. The user is now known to Facebook, future visits start logged in, and the next ask for the app becomes easier.
• Satisfy intent in the moment so they leave with what they came for, ideally with a positive impression, ideally returning later.

Most upsell systems collapse this into a single binary: prompt or do not prompt. The ladder makes the design problem more honest. At each rung, the question is not whether to ask, but what to ask for given everything you know about the visitor and the moment.

Aggressiveness, and why it is a spectrum

A second axis crosses the ladder: how aggressive each ask is. An aggressive prompt converts more in the moment but degrades trust and the chance of return. A soft prompt protects the experience but converts less. There is no globally correct setting, only correct settings for specific moments.

I worked on a model that treated aggressiveness as a scale rather than an on/off, shaped by timing between prompts, the ratio of visible content to blocked experience, frequency, screen real estate occupied, and how dismissible the prompt was. Different combinations of these inputs produced very different visitor experiences, and the design work was matching combinations to moments: aggressive only when intent was already satisfied, lighter when the visitor was still in the value-receiving part of the session.

The headline lesson from this work was that upsells are tolerated when they connect to clear value and arrive at the right time. They are rejected almost universally when they feel like a tax on getting to the thing the visitor came for. That sounds obvious; in practice it is the hardest part to hold across surfaces, because each team wants its prompt to be the priority one.

Saved login as the bridge rung

The single piece of work I am most attached to was rethinking saved login. Saved login is the second rung of the ladder: the user has been recognized, they are not being asked to sign up again, and they are being given a one-click path back into an account that already exists. Designed well, it is the least intrusive ask on the ladder and the highest converting, because the work has already been done; the visitor just needs to be reminded.

I designed updated header patterns for the off-platform video experience that surfaced saved-login states clearly across surfaces, kept the standard and saved-login versions visually consistent so a returning visitor saw something familiar, and reduced the friction of the actual reauthentication moment.

The saved-login approach also informed adjacent work on other teams.

Where it landed

Saved-login and contextual upsell patterns shipped across the off-platform video web experience on both mobile web and desktop. The aggressiveness framing became part of how the team made decisions about when and how to prompt.

Reflection

This was the onset of my growth-design work at Facebook, and it set the pattern. The visible job was designing upsells. The actual job was deciding what to ask for, when, and at what cost to the experience the visitor came for. I carry that into the rest of my work, like the growth work on the notifications team: be contextual, and at every conversion moment, look for the friction that can be removed so the ask feels like part of the experience rather than a tax on it.