Over 100 million people visit Envato's marketplaces every year. It's one of the world’s leading marketplaces for selling digital assets – everything from website themes, music tracks to illustrations and photos. You might have heard of them: ThemeForest, AudioJungle and GraphicRiver to name a few.
Big marketplaces need a lot of content. To keep a level of quality the marketplaces have content guidelines. But when getting over 2 million pieces of unique content a year, how do you review them all?
This is where the review team comes in. They review every single item, to make sure it meets the quality bar, before letting the item being published for sale.
To do this, without hiring hundreds of people, Envato needed a really efficient review tool.
Envato gets thousands of items that need to be reviewed every day. The volume is increasing a lot year-over-year. The existing review tool was slow, ineffective and the reviewers dreaded using it.
To keep up with the growing influx of items, Envato needed to greatly improve review capacity without greatly increasing the cost by hiring new reviewers.
We built and shipped a new review tool from scratch in 6 months. The tool is focused on speed, and that focus really paid off: Reviewers now get through 4 times as many items a day.
Besides the added speed the tool also...
Below is a prototype of the tool.
The tool (and prototype) are not responsive. The reviewers always work from a big desktop.
If you're on a mobile, try a desktop computer to play with the prototype.
As the main designer on the team, my responsibilities were very wide. They included the whole design process: user research, product direction, prototyping, visual design, user testing, and working with the developers on the implementation.
We built the product with an incredible team, which consisted of a product manager, 6 developers and me. The broader team also included a handful of reviewers and operations managers.
I had never reviewed any items. I knew very little about the review process before starting this project. In fact, the whole team didn’t know a lot about reviewing. So we started by learning.
We needed to understand:
We started by interviewing the review team. We wanted to learn about the current review tools, their biggest challenges, and understand what the different reviewers’ workflow looks like.
We discovered a few insights we had to address:
After getting a fundamental understanding, we started with mockups and early prototypes. Throughout the whole project, we worked closely with the reviewers. Every time we had a new idea (could be a flow or feature), we would bring it to the reviewers. This process helped us quickly narrow down what the tool needed to do, and the best way of doing it.
To lower the communication barrier we set up a Slack channel for reporting bugs, getting feedback on new features, and sharing general ideas on the review tool.
The close collaboration was very successful. Reviewers who have worked at the company for +5 years told us that they've wanted to collaborate for a long time, and this was the first time it was happening.
The new tool would fix a handful of different challenges. Both business challenges and issues with the current review tools.
Slow and therefore not scalable
Envato's library was growing very fast. Doubling every year, and forecasted to continue that trend. Having an ineffective review tool wasn't an option anymore. We needed to increase review capacity, without just hiring more reviewers. Hiring more reviewers would be too expensive over time.
Bad reviewer experience
The reviewers hated the previous tool. They had 3 reasons:
Contributors deserve better
Contributors experienced very long wait times, before knowing if their item had been accepted to the marketplace. It was also hard for contributors to get specific feedback and understand why an item might have been rejected.
The previous tool had a legal problem as well. The tool didn't allow doing the required technical check. It meant content with a legal risk could be published on Envato's marketplaces. Getting rid of this risk would be a big win.
The previous review tool. It's very busy and had a lot of functionality the team did not need.
We ended up delivering the new review tool in three phases:
1. MVP focused on the core review experience
For the first version, we focused on getting the core review experience right. This is the part where a reviewer needs to understand a certain item, see all preview images, and download the file to check everything is in order.
When the reviewer has made up their mind, they need to be able to action that item. There are 4 potential outcomes of a review:
The main part of the tool - reviewing an item.
2. Bulk reviewing to increase speed
After shipping the MVP version, we instantly started working on the next feature: Bulk reviewing.
Reviewers often only need a quick glance at a preview image, to tell if the item should be rejected. This is where bulk reviewing comes in handy.
Instead of reviewing one by one, the reviewer sees a grid of images. In this view, they can reject batches of items. The reviewer uses this to quickly deal with the clear rejections, so they can focus most of their time on the good and great items.
Bulk reviewing is used to quickly deal with the clear rejections, so the reviewer can spend most of their time on good and great items.
Reviewer can ask creators to do small changes by providing feedback. After the feedback is implemented the item can be published.
3. Revising a review
We're human and we all make mistakes. That's why the next feature we implemented was allowing reviewers to revise a decision they recently made.
It's quite simple. We implemented review history so the reviewers could look through the work they've done. If they see an item they want to change the decision on, they simply click it, and redo the review of that specific item.
Review history, for reviewers to check their work and correct mistakes.
Reviewer can see past review decision in bottom right corner, and revise the review if a mistake was made.
We built and shipped the new review tool in 6 months. We achieved our goal and much more:
The next step is to support other item types. The tool is specifically built for Graphic items. We're in the process of building a similar tool for stock video.