agnieszkamichalska

When Media Monitoring moved from Prowly into Semrush's AI PR Toolkit, the easy path was to carry the existing tool over as-is. I pushed for a full redesign instead, treating the migration as a chance to rebuild the experience around what PR teams actually needed rather than what already existed.

Before changing anything, I audited the existing tool to understand what was working and what wasn't, drawing on direct feedback from the PR specialists and agency users who relied on it every day.

I redefined the core user personas and mapped the end-to-end user journey from spotting a trend to reporting results. A value proposition canvas clarified what the tool needed to deliver, which fed into a new information architecture, new user flows, and a clear list of what to keep from the old tool and what to leave behind.

With the research and structure in place, I designed the new version of Media Monitoring end to end, translating the personas, journey, and information architecture into a redesigned interface.

The old query setup was a major source of friction, so I redesigned it into an advanced query builder with simplified boolean search. Query completion rose to 73.3%, up from 48.7% in the previous Prowly-based version.

I moved key metrics onto the first page so teams see their results immediately, with individual mentions organized into their own tab. Alongside this, I built easy reporting so PR teams can share metrics and mentions with stakeholders without manual work.

Monitoring data doesn't just surface what's happening, it points to what to do next. I built a feature that turns tracked mentions and trends directly into campaign ideation: PR teams can generate campaign recommendation cards straight from their monitoring data, each one pairing a story angle with the media targets most likely to cover it and the supporting assets needed to pitch it. Teams can start from a trend they've spotted and get matched campaign ideas, or start from a campaign goal and get matched coverage opportunities. Once a campaign is underway, the same loop tracks goals and results against the original recommendation, closing the gap between watching the market and acting on it.

The redesign turned Media Monitoring from a tool PR teams tolerated into one built around how they actually work: a faster, clearer query builder that lifted completion rates from 48.7% to 73.3%, metrics and mentions that no longer require digging, and reporting that takes seconds instead of manual compilation. Users who engage with it retain at more than 3x the rate of those who don't, and it's compelling enough that almost 60% of users on its plan tier chose that plan specifically to get access to it.

Semrush had already shipped an AI Visibility tool, letting marketers and PR pros answer 'are we visible in AI?' But it stopped there – it couldn't explain how that visibility could be influenced or improved. Without knowing which media outlets large language models actually trust, users had an interesting metric but no strategy. That gap is what AI-Cited Media was built to close: turning AI presence into something PR teams could actively shape.

We designed AI-Cited Media as a focused, topic-based experience, and in a later iteration connected it directly to outreach. Users enter a topic, market, and optionally a brand to see which outlets and journalists AI models cite most, each scored by AI authority. From there they can open a journalist's or outlet's full profile: beat, reach, audience demographics, contact details – select the right contacts, and generate personalized pitches. What started as a way to answer 'who should I pitch' became a flow that takes users from insight to a drafted email in a few clicks.

Before shipping broadly, we ran early concept testing and usability research to check the product would actually make sense to the people using it. Ten marketers and PR professionals went through 45-minute sessions with an interactive prototype, focused on first impressions, task understanding, and decision-making – not just reactions to static screens.

The value proposition wasn't landing on its own: several participants described the tool without mentioning AI at all, even after finishing the flow, and most gravitated toward concrete outputs – outlet lists, journalist recommendations, pitch ideas over abstract metrics. Confidence rose with in-context explanation, but most people skipped instructions and explored instead. That reshaped the product: renaming it to foreground media influence over visibility scores, simplifying language and tooltips, and shifting the flow toward outlet discovery → journalist selection → outreach.

AI-Cited Media quickly became a mainstream part of the AI PR Toolkit: roughly half of all users engage with it every week, and since connecting the flow directly to outreach, 24% of users who start a search carry it all the way through to a personalized, AI-drafted email. That confirmed the core bet: people don't just want to know who to pitch, they want the tool to help them actually do it.

Through sales calls and support feedback, digest notifications kept surfacing as one of the least valuable parts of the product. They lacked context, offered no prioritization, and didn't meet the time-saving bar enterprise users expected. Users could see what happened, never why it mattered, a gap that hurt both satisfaction and conversion.

The solution paired AI-driven analysis with a full redesign of alert setup and digests: AI summaries and spike analysis that explain unusual trends, a simplified setup flow for mention types, topics, and frequency.

I led the experience from problem definition and concept validation through flow, UI, and post-release analysis. Partnering with the PM on scoping and metrics, working with developers on feasibility, and actively shaping the AI prompts behind the summaries. I built interactive Figma prototypes for validation and engineering handoff, and helped define the Mixpanel events used to track adoption.

Two goals anchored the work: drive enterprise conversion and retention through feature differentiation, and lift alert and digest engagement metrics.

Trial users setting up alerts rose 173%, paid users with active alerts grew 38%, email open rate climbed 12.7 percentage points, and click rate rose 94.6% nearly doubling engagement across the funnel.