AnswerBite · 2023–2024

Full case study

AI Content Platform

Redesigned the MVP web app so uploading, processing and sharing video content is transparent and scalable — cutting first-time friction, making every system state explicit and separating inputs from outputs.

Project type

Product redesign

Role

Product Designer

Platform

Landing page & web app

Year

2023–2024

Key outcomes

+38%

Successful first uploads

−27%

Drop in abandoned sessions

4.4★

Early onboarding satisfaction

Clearer states

Fewer support tickets

Context

The MVP never told users where they stood

The MVP gave first-time users almost no guidance. Uploading felt heavy, banners repeated the same message without helping, and system states were implicit.

People could not tell whether their files were processing, ready or failed. Navigation mixed content inputs (uploads) with outputs (deliverables), so results were hard to find even once they existed.

Research

Five interviews, twelve Maze sessions, one recurring sentence

Almost every participant asked a version of the same question: “did it work?”. That reframed the brief from adding features to making the machine legible.

01Users wanted continuous progress feedback during upload, not a spinner.

02Confusion came from duplicate banners and redundant confirmation steps.

03Library, Deliverables, Clips and Social read as four unrelated products.

04First-time users expected one place that lists everything and its status.

Nadia, 29

Content marketer at a small SaaS

“I upload a webinar and I want ten clips I can post today.”

Goals

Fast turnaround
Clips ready to publish
Reuse long recordings

Frustrations

No idea how long processing takes

Outputs scattered across tabs

Fails silently on large files

Tom, 36

Solo creator

“If the first upload doesn't work, I'm not coming back.”

Goals

Try it without signing up twice
Understand pricing
Edit AI copy

Frustrations

Empty screens with no next step

Repeated banner nagging

Rewrite results hard to compare

The two people the redesign is written for.

Opportunity

Explain the value, cut the friction, earn the account

01How might we make uploading and processing effortless and transparent?

02How might we organise navigation so users find content and outputs faster?

03How might we use copy and states to guide users with confidence?

Team & scope

01Designed solo with the founder and one engineer.

02Scope: MVP redesign across upload, Library, Deliverables, Clips, AI Rewrite and Account.

03Timeline: 8 weeks to ship v1, iterative rollout afterward.

04Constraints: limited dev bandwidth, no existing design system, tight deadlines.

Information architecture

Splitting what you put in from what you get out

An open card sort with nine users showed a consistent mental model: source material on one side, generated assets on the other, account concerns kept out of the way. The navigation was rebuilt on exactly that split.

Card sort — FigJam board

Things I upload

Videos

Podcasts

Transcripts

Drafts

Things I get back

Clips

Summaries

Social posts

Rewrites

Things I check

Processing status

Errors

Usage

History

Things I set

Brand voice

Plan & billing

Team

Integrations

Open card sort — clusters users formed unprompted.

Site map

App root

Sources

Upload
Library
Processing queue

Outputs

Deliverables
Clips
AI Rewrite
Social drafts

Insights

Status overview
Errors & retries
Usage

Account

Brand voice
Plan & billing
Team access

Navigation after the split.

User journey

Landing → upload → processing → review → deliverables → sharing

Mapping the journey made the drop-offs obvious: they clustered around moments where the system was working but silent.

User flow

01LandingJudges whether the tool fits in ten seconds.
Pain pointValue described in feature language
02UploadDrags in a long recording.
Pain pointNo progress, no size guidance
03ProcessingWaits, refreshes, waits again.
Pain pointSilent system, unclear duration
04ReviewReads generated clips and rewrites.
Pain pointBefore/after impossible to compare
05ShareExports to social or downloads.
Pain pointOutputs live in a different section

Journey map with the friction points that drove abandonment.

Solution

Key changes introduced

01Streamlined upload flow — fewer steps, with real progress and file guidance.

02Navigation split into sources and outputs, mirroring the card-sort model.

03One explicit state language: Queued / Processing / Ready / Failed, everywhere.

04A single banner system with concise, action-driven copy.

05Guided empty states that tell a first-time user exactly what to do next.

06AI Rewrite rebuilt as a side-by-side before/after with inline accept.

State machine

Queued

Accepted and waiting — position and estimate shown.

Processing

Live progress with a cancel option.

Ready

Outputs available with a direct link to review.

Failed

Plain-language reason plus a one-tap retry.

The state language applied across upload, library and deliverables.

Wireframes

Sources / Outputs nav

Drop zone with guidance

Queue with live states

Recent uploads

Upload & queue

Sources / Outputs nav

Original | Rewritten

Tone controls

Accept / regenerate

AI Rewrite

Core screens at low fidelity, before visual design.

Coming soonFinal UI shot placeholder

Library — final UI

Coming soonFinal UI shot placeholder

Deliverables — final UI

High-fidelity UI shots to be added — placeholders for now.

Impact

What changed

01+38% successful first uploads during testing.

02−27% abandoned sessions.

03Support tickets dropped as states became self-explanatory.

04Early testers rated onboarding 4.4★.

Reflection

What I took from it

State clarity is often more valuable than adding features.

Microcopy drives trust as much as interface changes.

Building a simple structure early prevents rework as products scale.

Tools — Figma, FigJam, Maze, Jira, Hotjar

View the original case study ↗All case studies

Next project

Music Curating Platform

Playlist Map · 2021–2022