Two moments inside larger launches, at the level where the judgment actually happened: the problem, my reasoning, what I rejected, and what changed after it shipped.

Read the full case studies →


Diagnosing a leaky funnel for a new app

Instagram for TV login screen • MVP → Post-launch update

THE PROBLEM

We were seeing a big drop-off from download to login on the TV app after the MVP launched. The login screen’s headline anchored on a co-watching value prop, which tested well in research as a top use case for IG on TV. We also knew from research that people were wary of adding their account to a shared device, so the screen offered account creation too — leaving two competing CTAs with nothing to explain the choice and no clear hierarchy between them. The first instinct across the team was that this was a copy problem.

Instagram for TV login screen at MVP launch. Headline reads "Reels are made for watch parties," with "Log in or create an account" beneath it. A large QR code fills the right side, with the activate URL and code below, over a blurred animated background.

Login screen at MVP launch. Value prop scoped to specific use case. Create account not framed as privacy option. Competing CTAs and instructions. All on top of an animated background image.

MY APPROACH

The opportunities to refine content were clear, but I didn’t think that was the core issue. If you put a giant QR code on the screen, most people will use it to log in — and ours was on top of an animated background image. During QA tests, my phone camera struggled to recognize the code and it turned out other team members experienced the same thing. My hypothesis was that switching to a static image would move conversion more than any headline would. But we didn’t have any data on why users were dropping off, and mine wasn’t the only theory.

Instagram for TV login screen after the post-launch update. Headline reads "Everything you're into, now on TV," with "now on TV" in a handwritten script. Beneath it: "You can log in with your current Instagram account or create a new one to watch on TV with others." The QR code sits at right, over a static photo of two people watching TV in a bedroom.

Post-launch update. Headline anchored on core IG value people already believe in. Clarified existing vs. new account options to address privacy concerns. Changed animated background to a static image.

It could be the value prop wasn’t resonating, or maybe it wasn’t clear you could set up an account just for TV. So I explored alt headlines built on other value props, getting the team aligned on one that reflected Instagram’s core brand positioning, and added context to better explain the account options. We also replaced the background image, then shipped all the changes at once.

ALTS EXPLORED

We considered a wide range of headline options within different messaging territories. These are just a few of the directions we passed on, and why.

Alternate headline explored: "Instagram is now on TV." Login screen otherwise unchanged, with the same subhead, QR code, and background.

At first, the team wanted to eschew the value prop for a simple announcement. But users only see this screen after downloading the app, so I advocated against what felt like empty language that doesn’t build momentum.

Alternate headline explored: "Nothing else like Instagram on TV." Login screen otherwise unchanged, with the same subhead, QR code, and background.

Building on that approach, we looked at an option that still leaned into the branding while also emphasizing differentiation — no other streamer has content like Instagram. But it didn’t say why: personalized content based on your interests.

Alternate headline explored: "Something good is always on." Login screen otherwise unchanged, with the same subhead, QR code, and background.

This less direct, more ad-like headline signals unique value and creates intrigue via the Instagram advantage of endless short-form entertainment. Better as a campaign line, though. Here, a clear tie to the core product people love is best.

OUTCOMES AND NEXT STEPS

QR scanning was seamless in team dogfooding after the changes, which I think supports the background theory. But we didn’t isolate variables and test results were still pending when I left the team, so the actual impact of these changes on conversion is unknown.

Everyone agreed the content updates addressed the other theories, and we had more product and design improvements lined up — including a significant user flow/IA change I proposed — to continue optimizing this step in the funnel.

More of my work on Instagram for TV →


Teaching AI responses to meet the moment