top of page
Multimodal Expression
Context
Overview
A homeowner notices a problem, but doesn’t always know how to word it or who to hire. Today, our search experience depends on keywords and expects users to know the exact pro they need. When that assumption breaks, customers choose the wrong service, pros receive mismatched leads, and both sides lose time and money.
Existing search experience
Goal
Our team wants to transform the way homeowners express their needs so they can confidently decide what to do next.
In order to do this, we must increase the depth and quality of customer queries to ensure accurate matches with qualified pros. Rich queries = 4+ words or an image is added.
The next opportunity is to support multiple forms of expression and leverage AI, shifting the burden of decision-making away from the user and onto the system.
Past experiments
Learnings
We have tried a few different versions so far...
Q2 2025
Q4 2025
User problem
However, users still struggled to adopt and move beyond the mental model of keyword search.
Business problem
This means customers are dropping off upfunnel, leading to lower quality matches.
Core hypotheses
1) They’re confused and/or uncertain how it works
-
They do not know what to type
-
They think they’re in the wrong place because it doesn’t look familiar
-
They’re confused by the prompt language
2) They know how it works but choose not to engage
-
It feels like too much work
-
They’re skeptical of it being AI-driven
HMW make it easier to get started?
Design
New design
This new iteration focuses on:
-
Clear feature education to help users quickly understand how the experience works
-
Personalized, conversational headlines (H1) that feel relevant and welcoming
-
Dynamic project examples tailored to the user’s home data
-
Rich, animated input fields that invite interaction and signals AI
-
A more intuitive interface for combining multiple modalities (e.g., text, photos, and voice)
Personalized and conversational H1
Juicy, animated input box with modalities
Education on how we are using AI
Dynamic examples based off their home
‘Course-correcting’ smart text to encourage richer details
Value prop to explain why their details matter
Flexibility to move forward with short queries
Insights
Product evolution
While there’s more work ahead, we’re seeing encouraging early signals.
4+ words
20%
Photo adoption
3%
Abandonment
10%
Future iterations
As we continue evolving this experience, the next step is making it more engaging, intuitive, and inspiring for homeowners. We’re exploring ways to reduce friction while bringing more visual guidance into the journey.
Key areas we want to explore next include:
-
Leaning into a more playful, visual-forward experience to make the interaction feel lighter and more inviting
-
Introducing curated images and visual examples to help customers quickly recognize and articulate what they need
-
Creating low-friction entry points so customers can start a project with minimal effort or uncertainty
-
Highlighting stronger value propositions earlier in the experience to build confidence and clarity
-
Adding moments of delight through thoughtful animations, micro-interactions, and personality
bottom of page