Define
Problem
Business owners know reviews build trust and drive bookings, but they don’t have time to write thoughtful replies in their brand voice for every comment.
Solution
We designed Edge Autoreply to help business owners save time while staying engaged with their customers. Each new review becomes an opportunity for meaningful interaction: the system generates a draft AI reply that can be quickly edited, regenerated, or posted with a single click.
Target audience
Small business owners, multi-location businesses, and administrators who oversee daily operations
Research Process
Brainstorm Session
Before diving into research, our EPD (Engineering, Product, Design) team came together to align on the questions we needed to answer, potential technical limitations, and the risks we should keep in mind. The best ideas come from collaboration! We used a Miro board with sticky notes so everyone could add their thoughts, then grouped questions and concerns into themes to identify patterns and decide where to focus using anonymous voting by applying dots to the board.
Competitor Analysis
Edge can stand out by giving businesses real control over auto-replies, letting them decide if a review should get a reply after a set number of hours and choosing exactly which star ratings trigger automation. It can also go further by using brand guidelines to craft highly customized responses and by allowing teams to create and manage their own templates so every reply feels authentic and on-brand.
Learning from Mistakes
In the past, I made the mistake of jumping straight into polished, end-to-end vision designs. When I shared them with developers and PMs, they often had to be cut back to an MVP, which meant I’d redesign everything again to match the actual scope. Putting too much detail into early designs slowed us down instead of speeding things up, and it sometimes felt overwhelming to the team. I’ve since learned that starting with an MVP design makes it much easier to build alignment and momentum. By creating the MVP with PMs and engineers, we can bring in technical and product input early, reduce rework, and give everyone a sense of ownership.
Keeping this in mind, our EPD team worked together to define what needed to go into the MVP and what could wait for a future version.
What our users say?
A one-week in-app survey through PostHog showed why users don’t always respond to reviews and helped gather contacts for follow-up qualitative feedback.
Let's get to know a user persona
Edge users run all kinds of businesses. Some have small, boutique shops, while others manage large, multi-location businesses. No matter the size, they face the same challenges and share the same goals. That’s why I’ve combined them into a single user persona: Christine.
Design System
Using the MUI design system, which we adapted to match the Edge brand style, I built the future design. Here is a small part of the design system related to the star rating:
user flow
Creating a user flow makes it easier to see how our customers move through a product, spot where they might get stuck, and design a smoother, more intuitive experience.
.png)
Responsive Design
Keeping user persona Christine in mind, I built the MVP design. Since Christine is a very busy business owner, I added stronger visual indicators to show the status of online review responses and highlighted reviews that need more attention (1–3 stars).
Not Quite Right. Let’s Iterate!
Problem
It’s not clear whether you’re deleting the customer’s review or the AI’s response to it.
Solution
Place CTAs and the status chip inside the AI reply card, making it clear users aren’t deleting the customer’s review. Change the CTA from 'Delete' to 'Decline' for clarity.
Problem
For the MVP, reviews got an automatic reply after 3 days if no action was taken. Customers want their team to manually review and approve responses so nothing is missed and they can learn from how AI Reply responds to negative reviews.
Solution
We removed auto-reply functionality from the MVP, allowing customers to manually review and approve responses before we introduce more advanced configuration options.
What is next?
Customers were very excited about the upcoming AI Reply feature, and it has been included in the upgrade plan. Even in its current state, it delivers value by reducing the need for manual replies. AI-generated responses are sent via email, and customers can approve them directly from the email or through the app.
Looking ahead, we want to enhance the feature by allowing customers to upload brand guidelines, create templates, and configure advanced options, such as response hours and automated handling of specific star ratings via toggle buttons.
One of the biggest challenges customers face is that staff often copy and paste directly from templates without making changes, which makes responses feel robotic. AI-generated replies can help by creating customized responses based on templates or brand guidelines, ensuring replies feel more natural.
View more work
