UX RESEARCH · UI DESIGN · 2022

Redesigning the YouTube Search Experience

ROLE
UX Researcher · UI Designer
TIMELINE
Fall 2022
TEAM
User Experience Class Midterm Assignment
SKILLS
User Interviews, User Journey Mapping, Competitive Analysis, User Flows, Low- & High-Fidelity Prototyping (Figma)

Overview

I use YouTube's search bar constantly, and just as often I leave it not having found what I was looking for. Search for almost anything and the results page hands you back three different things stacked on top of each other: the videos that actually match your search, a "People Also Watched" row, and a "For You" feed of recommendations — with no clear line between them. I wanted to know whether that was just my own pet peeve or a real, shared problem, so I ran a small research study to find out, and used what I learned to redesign the page.

A competitive analysis of YouTube against Vimeo, TikTok, and Twitch — looking at how each platform's search and discovery model handles the same tension.
A competitive analysis of YouTube against Vimeo, TikTok, and Twitch — looking at how each platform's search and discovery model handles the same tension.

Research

I interviewed 10 students about how they actually use YouTube search day to day, then had five of them walk me through a real search, start to finish, while I mapped their experience as a user journey — what they searched for, how they felt at each step, and what they expected versus what they got.

The five journeys covered very different searches — a skincare video, box-braid tutorials, a knitting machine setup, scary story videos, home tour videos — but the shape of each journey was strikingly similar.

Ivy's journey: a specific search (“Vogue Skin Care”) succeeds immediately, but her mood drops as soon as she reaches “People Also Watched” and “For You” — she describes the suggestions as “irrelevant” and “overwhelming.”
Ivy's journey: a specific search (“Vogue Skin Care”) succeeds immediately, but her mood drops as soon as she reaches “People Also Watched” and “For You” — she describes the suggestions as “irrelevant” and “overwhelming.”
Gbemi finds what she wants only after refining her search twice — then “People Also Watched” surfaces a weight-loss method she has no interest in, and “For You” mixes in videos about “people dying,” which she calls “distracting.” She scrolls past several more irrelevant sections before giving up.
Gbemi finds what she wants only after refining her search twice — then “People Also Watched” surfaces a weight-loss method she has no interest in, and “For You” mixes in videos about “people dying,” which she calls “distracting.” She scrolls past several more irrelevant sections before giving up.
Lucia has to search twice more, adding increasingly specific keywords, before she reaches results she actually wants — and even then, “For You” serves her a sewing video and an apology video with no connection to knitting at all.
Lucia has to search twice more, adding increasingly specific keywords, before she reaches results she actually wants — and even then, “For You” serves her a sewing video and an apology video with no connection to knitting at all.
Araceli's search goes more smoothly overall, but even her one dip comes at “Previously Watched” — a section she didn't ask for and calls “not the most helpful.”
Araceli's search goes more smoothly overall, but even her one dip comes at “Previously Watched” — a section she didn't ask for and calls “not the most helpful.”
Milo's core complaint: the platform's biggest, most-verified channels crowd out the smaller creators he's actually looking for, and “For You” recommends gaming and cooking videos with no relationship to his search for home tours.
Milo's core complaint: the platform's biggest, most-verified channels crowd out the smaller creators he's actually looking for, and “For You” recommends gaming and cooking videos with no relationship to his search for home tours.

This is the opposite of knotless braids.

What the journeys had in common

Every single journey dipped at the same two moments: "People Also Watched" and "For You." Regardless of what someone searched for, these sections consistently broke the thread of what they came to find — introducing unrelated topics, resurfacing videos they'd already seen, or leaning on an algorithm that, as more than one participant put it, they simply "didn't understand." The searches themselves usually worked, sometimes only after a student added more specific keywords than they should have needed to. It was everything appended below the actual results — treated as part of the same page, with no separation — that consistently cost the platform their trust.

I mapped that repeated pattern into a single user flow to see the whole loop at once: a search leads into "People Also Watched," which either lands or sends someone scrolling into "For You," which either lands or sends them scrolling further — and if nothing lands, they don't abandon YouTube outright, they retype the search with different keywords and run the entire loop again before ever considering another platform.

The user flow distilled from the interviews and journey maps — the same “search → People Also Watched → For You → refine and repeat” loop showed up regardless of what anyone was originally searching for.
The user flow distilled from the interviews and journey maps — the same “search → People Also Watched → For You → refine and repeat” loop showed up regardless of what anyone was originally searching for.
Synthesizing the competitive analysis against the interviews and journey maps.
Synthesizing the competitive analysis against the interviews and journey maps.

The competitive analysis pointed at why. YouTube's greatest strength — a massive, 2-billion-user library with more original content than any competitor — is also the source of the clutter: with that much to recommend, the platform defaults to showing all of it at once. Vimeo and Twitch, by contrast, stay legible by staying niche. TikTok solves discovery differently: its "For You" algorithm is specific enough, and separated enough from search, that a creator with 100 followers can still reach a wide audience — something participants said felt nearly impossible on YouTube, where the algorithm favors channels that are already large and verified.

The Redesign

The fix I landed on didn't require inventing anything new — YouTube already solves this exact problem on its home screen, which cleanly separates "Home," "Shorts," and "Subscriptions" into their own tabs. I proposed carrying that same pattern onto the search results page: a Results tab holding only what matches the query, with For You, People Also Watched, and Previously Watched moved into their own tabs beside it — always reachable, but never mixed in with what someone actually searched for.

Low-Fidelity Wireframes

I started by sketching the tab structure directly onto the existing search results layout, working out which tabs should exist and, just as important, when each one should and shouldn't appear.

Early wireframes for a returning user, sketched directly onto YouTube's existing layout — tabs sit where the algorithmic “People Also Watched” and “For You” rows used to live. A first-time user sees no “For You” tab at all, since there's no history yet to base suggestions on.
Early wireframes for a returning user, sketched directly onto YouTube's existing layout — tabs sit where the algorithmic “People Also Watched” and “For You” rows used to live. A first-time user sees no “For You” tab at all, since there's no history yet to base suggestions on.

From there I moved into cleaner low-fidelity mocks and wrote out the logic behind each tab, so the structure held up before I touched any real visual design:

The core structure: a query lands on a Results page grouped into four tabs. “Results” are videos that directly match the search; “For You” is algorithmic and only appears for logged-in, returning users.
The core structure: a query lands on a Results page grouped into four tabs. “Results” are videos that directly match the search; “For You” is algorithmic and only appears for logged-in, returning users.
“People Also Watched” surfaces videos adjacent to the same query; “Previously Watched” resurfaces what the user has already seen — and, like “For You,” only exists for a user with viewing history.
“People Also Watched” surfaces videos adjacent to the same query; “Previously Watched” resurfaces what the user has already seen — and, like “For You,” only exists for a user with viewing history.

High-Fidelity Prototype

For the final prototype I rebuilt the flow using YouTube's own design library in Figma, so the redesign would sit inside YouTube's real visual language rather than looking like a foreign concept — the same components, type, and spacing a YouTube engineer would actually ship.

Searching “knitting machine for beginners”: the Results tab holds only videos that match the query — no algorithmic content mixed in.
Searching “knitting machine for beginners”: the Results tab holds only videos that match the query — no algorithmic content mixed in.
Switching to the For You tab surfaces YouTube's usual recommendations — sewing tutorials, an unrelated finance video — content that would have previously cluttered the results page but is now one deliberate tap away instead of eight rows down.
Switching to the For You tab surfaces YouTube's usual recommendations — sewing tutorials, an unrelated finance video — content that would have previously cluttered the results page but is now one deliberate tap away instead of eight rows down.
The Previously Watched tab keeps the videos this user already watched in their own space, so “have I seen this?” is never a question the Results tab has to answer.
The Previously Watched tab keeps the videos this user already watched in their own space, so “have I seen this?” is never a question the Results tab has to answer.

It's a small change, but the research made the case for it directly — the goal isn't to remove recommendations, which every participant still valued, but to stop asking someone to sort "did I search for this, or did YouTube recommend it to me?" on every single row of the same page.

Reflection

What I learned