OSP Review Moderation Tool
Cost saving
Operational improvements
User satisfaction
Summary
Ocado's review moderation was outsourced to a third-party provider costing over £550k annually, with projected costs rising to £750k as more partners joined the platform.
This project aimed to design an in-house moderation system to replace the external service, reduce operational costs, and support Ocado Group's goal of becoming cash-flow positive by FY26.
My Contributions
• Led end-to-end UX design from ideation to delivery
• Co-developed research plan and discussion guide
• Conducted interviews with existing moderators
• Created prototypes and validated concepts with stakeholders
• Produced detailed documentation for engineering implementation
Role
UX Designer
Platform
Internal tooling | B2B
Team
1 PM | 2 Engineers
Duration
5 Months
The Problem
The existing system relied on Ocado's partnership with PowerReviews which created several issues.
Cost
This partnership was costing upwards of £550k a year to maintain.
Parity
Ocado was maintaining two separate moderation systems.
Scale
In the future we wanted to included user uploaded images in review, with a partner this wasn't possible.
Research & Understanding the System
As mentioned previously Ocado was maintaining two separate review moderation systems, one they had built years ago called Kale which was no longer maintained and PowerReviews which was a partnership with another business.
I started this project by mapping out the previous state of both system's information architecture to not only understand their current process of review moderation but also to be aware of possible constraints which we could address in a redesign.
From this I was able to see the differences in the two software. The next step would be to talk to users on how their experience is when using either PowerReviews or Kale. For this I approached a UX Researcher in my team to assist in the creation of a research plan and discussion guide with two main objectives;
Experience
How is the current experience of review moderation?
Is there anything that you are missing currently?
Usability Test
Allow the interviewees to experiement which a prototyped version of an OSP Review Moderation System
Key Insights
From interviewing current users we were able to full grasp how they currently use both Kale and PowerReviews, understanding their patterns of working and current pain-points which could be address in a ground-up redesign.
Additionally I found that their reception to the proposed direction of the redesign would need some changes. Although their overall feelings were positive there was evidently some things that needed ironing out
"I prefer this method. Not having to click into it and then choosing a reason then clicking back out. This is just such a quick way of doing things."
"Seeing whether it's published or not published, that's really useful. I like that."
"feels a lot smoother than Kale immediately… very smooth and very easy to navigate"
"Currently If the words aren't flagged in our report then they are missed, unless someone in the company flags it"
Design Goals
Following on from the user interviews I was able to identify key areas for improvement and create the outline for what would become the review moderation system.
Solution
The final design focused on simplifying reducing decision time, consolidating tools, and allowing for future scaling into a single intuitive product.
Reducing time
Having select boxes and the ability to accept or reject would speed up triage time significantly. No need to click into reviews with all the information readily available.
Consolidating tools
Having all you need to moderate on a couple pages makes for a much simpler information architecture.
Scalability
Making sure that future features such as image moderation could be completed easily and fit in well to the solution.
Validation
Following on the previous rounds of designs I once again set out to validate these with users to understand their thoughts and reduce further friction when they would migrate to the new system. There were a few main takeaways from these sessions.
There are clearer workflows, with flagged reviews all in one place, no need to parse through alternate pages.
There no more clicking into reviews, this allows for a quicker moderation process (6s on average from 14s).
It's much more efficient, with quick access to moderation tools and the ability to search for reviews all on one page.
Outcome and Impact
