How to create ultra-precise user segments with multiple levels of filters and properties?

Problem

When sending a survey to a specific user segment, product teams could only apply filters based on a single event, without the ability to specify contextual details.

Solution

I designed an extension to the filter system that allows users to select an event, a related property, and an operator, while providing real-time feedback on audience size, all without cluttering the interface.

How to help product teams analyse their data more intuitively and instantly?

Problem

Screeb’s users, primarily Product Managers, have grown used to conversational interfaces that let them find insights quickly through natural language. This type of interaction is a major time-saver.

To meet these expectations, Screeb wanted to launch a new AI feature: a Product Assistant capable of analysing any dataset across the platform.

This raised several design challenges:

• Sometimes, the user’s question requires the AI to scan all available data to produce a relevant answer.
• Other times, the user is referring to a specific dataset (e.g., a survey or a product page). → How can we help the user clearly define the scope of analysis expected from the AI?
• Finally, to support adoption and meet user needs, it was essential to make the AI’s reasoning transparent and easy to follow. → How can we offer visual feedback on the AI’s logic and calculation steps without overwhelming or confusing the interface?

Solution

The Product Assistant suggests useful prompts based on context (to drive adoption), answers natural language questions, adapts to the current page (context-aware), displays its reasoning to build trust, and proposes next steps.

How can we surface relevant data before users even think to ask? How can we help them save time by avoiding repeated queries for the same answers? And how can we encourage Product Assistant discovery?

The dashboard widget was designed as a direct response to these questions. First, it proactively brings relevant insights to users, without them having to search, by displaying key data the moment they land on Screeb. The homepage becomes a gateway to product knowledge.

Second, the widget helps users save time: instead of typing the same queries or reopening the assistant to find a previously answered question, they get instant access to updated answers to recurring topics.

Finally, the widget serves as a trigger for Product Assistant adoption: the assistant’s icon is embedded in the widget, along with a CTA that encourages users to dive deeper or ask a follow-up question, gradually building a new habit in their product workflow.