The mismatch
A prompt-first interface serves users who already have intent. Amplift's core value was discovering opportunities they did not yet know existed.
AI marketing / Product redesign
Amplift AI Redesign
Bringing AI-detected marketing opportunities to the front of the product. I redesigned the home page and navigation so users could see what to act on before deciding what to ask.
- Role
- Product Designer
- Timeline
- Q1 2026 / 1 month
- Team
- 1 PM, 1 Lead Engineer
- Scope
- Home, navigation, templates
My contribution
I worked with a PM and a lead engineer. I owned research, user flows, two rounds of design and testing, and handoff. Watching users move through the product led me to change the home page without removing the workflows they already used.
Discovery
A blank prompt hid the product's intelligence.
Amplift could detect competitor gaps, trends, and influencer matches. Yet its command-bar home required users to know what they wanted before it showed them anything useful.
The navigation problem
The flat sidebar treated frequent tasks and deeper tools with equal weight. As features grew, every visit required another mental scan.
The original command-led experience.
The home depended on an initial command, leaving the product's proactive intelligence out of view.
Study how AI value becomes visible.
I benchmarked more than ten marketing platforms for how they present generated intelligence and guide a first action.
Support both discovery and intent.
Opportunity cards provide a proactive entry point; agent commands remain available for users who arrive with a task.
Iterations
Two plausible directions exposed the same gap.
The first direction led with performance metrics. The second led with an agent prompt. Testing showed that neither should displace the opportunities users came to find.
V1 / Dashboard first
Metrics explained performance but did not explain what to do next. Users still navigated away to find actionable opportunities.
V2 / Agent first
Templates helped users start, but the command bar still required intent. Users repeatedly returned to the Opportunities page.
Final direction / Opportunities first
Bring ranked opportunities onto home. Keep commands as a secondary entry point and templates as guided support.
Users kept leaving the command interface for Opportunities. That told me the home page was starting in the wrong place. I moved opportunities to the front instead of adding instructions to the prompt.
Product Design
Lead with the next useful action.
The final home surfaces ranked opportunities and direct actions. A stable sidebar supports multi-step research, content, and reporting workflows.
Opportunities lead; commands stay available.
Ranked opportunities occupy the primary area, with the command bar secondary and templates close at hand.
Prioritize spatial consistency.
The sidebar performed better for users moving between research, content, and reporting tasks.
A lighter-looking pattern added orientation cost.
The dock alternative was tested, but switching between complex workflows made its visual appeal less valuable than a stable rail.
A template name did not explain the output.
The original grid offered little indication of what a template would produce.
Rank the opportunity, not just the metric
Present what changed and why it matters beside the action the user can take.
Make templates concrete
Show output examples and reduce the commitment required to try a template.
Preserve the intent-driven path
Experienced users can still command the agent directly without making that the only way into the product.
Results
From finding an opportunity to starting a campaign.
The goal was to help users find something worth acting on and start a campaign. After launch, we looked at session data, survey responses, and support tickets.
- Faster discovery
- 5x
- Research time saved
- 60%
- Faster setup
- 3x
Average sessions to an initial influencer shortlist, before versus after launch.
Self-reported per-task savings from a post-launch survey across 40+ active accounts.
First campaign completion tracked from home to the first published result.
Discovery and setup compare usage before and after launch, not a controlled experiment. The 60% time saving comes from survey responses, not timed tasks. We have not measured a retention change.
Support tickets
Findability-related Intercom tickets fell 40% in the first 30 days compared with the prior 30 days.
What I learned
Users wanted help deciding what to do next, not another view of past performance. Their repeated visits to Opportunities made that clear.