Monitoring & Alerting For Multi-Channel Sellers
Large multi-channel ecommerce sellers find it a hard task monitoring details in their sales and stock across their many marketplaces, resulting in missed opportunities.
After speaking to our ListSmart customers we came across many with this problem. As we were well placed to create a new product to help, we decided to investigate the scale and value.
Surveys & Competitive Analysis
I started with producing a survey to gauge the size of the problem and discovered ways sellers worked around it. Using LinkedIn I ran a marketing campaign for the survey, targeting Heads of Ecommerce and similar roles. During this time, I researched the competitive landscape and looked at which marketplaces we need to consider adding. I found that there are multi-channel management tools which have the primary function of managing sales and inventory. However, many sellers are missing the level of detailed analysis and alerts which are needed.
User Interviews & Personas
Working in collaboration with a User Researcher, I carried out interviews of large multi-channel sellers. I created quick mockups and prototypes of ways we could resolve the problem and ran users through them. Over a couple of weeks, I ran several iterations, finally narrowing down on a few core issues sellers had. From the user feedback of our ListSmart customers and the user interviews, I produced personas for our target market.
"Marketplaces like eBay and Amazon are a mystery – it’s so hard to get a complete picture. They’re like a black box."
Large e-commerce seller – Customer Research
The research showed
• Busy operational e-commerce teams are time precious.
• Multi-channel increases the workload multiple times over.
• Existing tools focus on charts, not pinpointing the problems.
• Data from multiple sources is required to show one consistent context.
• Sellers would like to save money by throwing away existing expensive and clunky channel management analytics solution.
Sellers want to know when
• Categories are the best and worst performing quickly
• Top sellers are outselling them
• Sale volumes drop
• The sell-through rate drops or rises
• Listings with long listing durations are underperforming
• Items with significant stock levels have low sales
Quick digestible alerts
Sellers are time constraint on what metrics they would be able to review, analyse and action each day. As a result, it was clear that we needed to present a high-level view of all their marketplaces with information prioritised to allow it to be digested quickly.
The solution was a limited number of alerts. Which are a short sentence with all the information needed, to be able to quickly scan and decide if it is something that needs looking into.
Low & high performing items
Alerts are divided into low and high performing, with low presented first as sellers monitor these closely. Initially, I showed only the top category for alerts where listings had dropped in sales and views, as to limit the list. However, feedback from sellers showed that we needed to allow them to see more categories for these type of alerts. For later versions we expanded it, allowing a seller to look at the top 5 categories.
An early working version of the app was used by a small group of sellers. Using in-person testing, feedback forms and analytics from Heap, I was able to collect insights into what was working and where the application was lacking.
Each alert has a modal to quickly access which listings have changed and the corresponding metrics. I included a download button for a CSV file of this table. This was a short term solution to saving the list of items that needed a later review. Ideally Catalogue Monitor would integrate directly into a management tool for quick access to amend data.
In the header, I had mocked up variations of a seller score to help gauge store health at a glance. I decided to demote the feature due to the build estimate, being larger than other features.
Product Owner role
As the Product Owner, I was in regular contact with the engineering team in Cork, Ireland. Setting the priority of user stories in each sprint and reviewing their output. As they are remote, our daily standups and monitoring the sprint-board was vital in understanding the progress of the product. In addition, I made fortnightly trips to meet in person and created worked examples of data breakdowns.
I collaborated closely with the sales team to get them confident with the features of the product, but also gave them visibility on the product roadmap. I sat in on calls and demos with customers to hear their questions and feedback. This feedback helped with understanding what was working, but also, as the Product Designer, to see how I could improve the user experience.
Previously, we had only integrated with eBay, so adding Amazon and Shopify each came with their own complexities. For example, we had to create a new category tree to normalise the data, as all three marketplaces have different methods.
Manage your marketplace
Connect and manager marketplaces integrated into Catalogue Monitor. Each site has its own authentication process with a description to help step a user through. Alongside, I included a feedback survey on which other marketplaces users have to help me priorities the next one to integrate with.
A feature requested by users was finding an individual product by SKU code. These codes are unique to each seller and therefore need to be text input. Products may be on multiple marketplaces so there is a summary to each site.
Catalogue Monitor was a SaaS application sellers can connect their eBay, Amazon and Shopify marketplaces to. Using the daily alerts, sellers were able to monitor all their stores for key metrics.
After 12 months of developing and improving the product, the customer base was not growing fast enough to support. As a result Catalogue Monitor was sunset.
Products that rely on data trends, like Catalogue Monitor, require historical data to be able to provide a trend. We started with 7 & 30 day trends, which meant that a user would signup and connect their store, but no alerts would show for a week. We added a 3 day trend to reduce the wait but the feedback from the customers showed it had little value. Having a stronger email campaign that ran alongside the product to help educate users over the first week after signing up might of helped.Accessibility
In retrospect, using the turquoise text colour has poor legibility and fails the Web Content Accessibility Guidelines (WCAG). The colour was an aesthetic choice, rather than functional and led me to research how I can improve my designs to be more accessible in future projects.
Finding product market fit is hard!