Current State.
The Warehouse Management team focuses on situations that happen without warning such as a store outage, emergency, or catastrophic event. If a store can no longer operate the live operations team has to be alerted so they can shut down a particular store, multiple stores, or even entire zones depending on the severity of the disruption.
Meet Tim, the Warehouse Management team lead.
They depend on buggy tools to accomplish complex tasks, and with the amount of work they handle, creates a very frustrating journey to accomplish their goals.
Using the product vision as our north star, we set out to not only consolidate tools, but also create a streamlined disruption for unplanned disruptions.
A local retailer is experiencing a POS outage that is affecting every store in the area.
Lets explore what the process of intervening on unplanned disruptions looks like.
Workflow.
First Point of Contact
Malik, Instacart Shopper
"Hey, I am shopping at this store and I can't check out. The cashier said it was a big outage across the city. I'm not sure what I should do..."
Riley, Instacart Care Agent
"I am so sorry to hear that! Let me get that taken care of for you. I need to fill out a form so this may take a second, so, please, bear with me..."
If a store is experiencing issues, typically the catalyst to intervention are shopper who reach out to Instacart’s customer care team through calls, chats, or e-mail. The care team agent will create a Jira ticket or one is automatically generated through Zendesk. The Jira ticket is sent to the Live Operations inbox.
Riley, Instacart Care Agent
"I am sending this form over now. Thanks for being patient!"
Tim, Instacart Warehouse Management
"Looks like we got another store outage request. It looks like we got a few requests for this area. I think I should loop in my team..."
Live Ops is Notified
Once the live ops teams is alerted to the issue, they can begin to process of intervening. The process of gathering and documenting information starts to ensure the right interventions are in place.
Tim, Instacart Warehouse Management
"Hey guys, looks like we might have a bigger issue in one of our zones, lets hop on a call and figure this out..."
Tim, Instacart Warehouse Management
"Every store that this retailer owns is experiencing issues. I think we need to focus on figuring out the issue...."
Intervention Begins
Information is copied and pasted from the ticket and into more spreadsheets, various communication channels, and store configuration tools. Once the intervention is in place, the information is then distributed manually to stakeholders and shoppers.
Tim, Instacart Warehouse Management
"Team, looks like there is a major outage in the city. We need to shut down all our fufillment for the retailer in that area..."
Tim, Instacart Warehouse Management
"Hey everyone, we are having a severe disruption. Here are the details..."
So, why doesn't this work?
The current intervention procedure to transfer information from different sources is slow and burdensome. When it comes to unplanned interventions, we know every second counts, and with so many steps, seconds can have an extremely high cost.
Efficiency.
The amount of obstacles between a disruption happening and the operations team being able to intervene is far to high. The time it takes for team to be alerted can cost Instacart hundreds of canceled orders, as well as lost revenue.
Complexity.
The workflow for unplanned disruptions is very complex. The care agents have several different tools that live in different places. When talking with the lead of the customer care team she emphasized how they already have too many tools.
Most of these actions are also manual, which makes it harder to communicate consistent information between teams. It becomes a bit like a game of telephone.
Workload.
The amount of obstacles between a disruption happening and the operations team intervening is far too high. The time it takes for the team to be alerted can result in hundreds of canceled orders, lost revenue, and very unhappy customers.
This process is frustrating for all parties involved.
Nobody can predict when an unplanned disruption will happen, but the moment an event begins the impact is felt immediately. From customer and shoppers, to retailers and care agents, in the event of an unplanned disruption, store intervention becomes priority for all parties.
Typically, by the time the Live Ops team is even alerted about a disruption, there is already negative impacts actively happening. When the volume of work reaches this level, other members of the larger live ops team will jump in to help.
With so many tools, procedures, and stakeholders in this one journey friction points and hiccups are not only more likely, but inevitable. If one component of this balance if off, the entire process will suffer.
With so many tools, procedures, and stakeholders in this one journey friction points and hiccups are not only more likely, but inevitable. If one component of this balance if off, the entire process will suffer.
How can we create a cohesive end-to-end experience, simplify processes, and reduce negative impact in the present?
Shifting focus from future thinking to present day solutions, I began the process of creating a viable product that targets present day issues based upon the Extreme Event Management Command Center vision.
Design Process.
Evaluate & Conceptualize.
What should we build?
To begin the design process, I took time to analyze what I already about the team, their work, and multiple points of contact that go into the intervention process. Through conversations with the team, I was able to find key insights into what was truly an immediate priority.
Though I understood that the initial map vision was the ideal solution for the team, through conversation I was able to bring the team together to make near term compromises. Given the restraints, we unanimously agreed that to make the greatest impact in the present we should focus on two key areas:
Unification
This team monitors weather events across the North America and Canada. These events can range from a snow storm in Ontario all the way to a major hurricane in Florida.
Unplanned Disruptions Workflow
This team focuses on finding local events that could impact fulfillment, for example, a marathon or a parade.
How should we build it?
To begin working backwards from the design vision, I looked at what we could do to consolidate programs. Some of the tools that the live ops teams, as well as the customer care team, use live within Instacart’s Partner Platform which I saw to be a great opportunity to unify tools.
I worked on the Instacart Partner Platform team who is responsible for building Instacart’s Partner Platform (IPP) experience, which is the hub for all things Instacart, both internal and externally. I knew that one of the most important goals for our team was to begin bringing all internal functions for all teams to IPP. The overall team vision is to make IPP the main internal tool company wide, no matter the team or role, within the next 2 years.
Soluition: Phase 1.
Iteration & Finalization.
Building the Workflow.
We consolidated 65% of the most crucial internal tools, such as Jira and Zendesk, that were cornerstones in the intervention process into IPP. This created a space for all the live ops internal tools to live and grow.
We needed to make changes to the IPP information architecture so that it could accommodate the new additions. Originally, the tool would live within the 'operations' section of IPP. Since operations within IPP were for 3rd party entities to perform tasks that were not directly related to monitoring and intervening, I proposed creating a new section called Marketplace Ops.
Within the Marketplace Ops tab are the three main actions that can be performed: submitting intervention requests, managing requests, as well as creating and managing interventions.
Submiting Requests.
Submitting and managing requests with Jira became a messy process that was not equipped to handle Instacart's specific needs. The customer care team had too many tools and the volume of requests put a strain on Jira's capabilities.
Since the care team was already familiar with IPP, I thought it was the perfect opportunity to build the request form directly into IPP. This reduces the number of tools they use and creates a more seamless experience.
Before, the care team would have to research and find information to ensure theat they are submitting the correct requests for the correct store, leaving room for error. Instead, we created a simple form that is kicked-off by a dynamic internal search.
The care agent can search for multiple stores at once, and create on single request form for all of them in three steps.
Managing Requests.
The request is submitted to the Live Ops team and lands in their inbox within IPP. Once alerted, Tim and his team are able to act quickly and confidently.
The unplanned disruptions designed to give the users a birdseye view of the most important information for them to act on. At the top of the page are metrics that provide insight into negative or positive trends based on the volume and type or intervention requests.
The inbox is a database of submitted tickets that provide various at-a-glance information regarding the request details.
When the user clicks on a request, a side panel opens to reveal the details of the submission. From there, they can make informed decisions and initiate interventions.
Interventions.
Interventions is possibly the most important aspect to Live Op's role at Instacart. We wanted to ensure that they can act quickly on high priority issues without them leaving the experience to perform other tasks.
In this experience we wanted to provide an easy way for the team to intervene on issues immeditaly. The side panel has an intervention button which launches the intervention experience.
From here you can begin to alert stakeholders, document actions, and create a single source of truth for this disruption.
In the even that more information is needed or needs to be reviewed, we created a full details page where users can directly change store fulfillment configurations.
We created a home for Live Ops that is usable in the present but capable of handling the future.
Wrap Up.
The Final Product.
As the final product began to materilize, I made sure to go throuigh every step of the user flow with the Live Operations team to ensure that each design decision made was tailored to their specific needs.
We were able to create a place for Live Ops work to live and evolve over time. In the beginning this particular team felt very underserved even though the impact of their work impacted the entire company.
Iterations.
This process went through several rounds of refinement before the final product was agreed upon. We collaborated on paring back the product vision into a MVP that could be immediately implemented. I was able to guide the team, as well as product partners, through the process of defining what features are nice to have and what features are non-negotiable.
The final version addresses the most immediate issues that are causing the most issues within the team's workflow. The resulting product is lean, flexible, and fills in the current gap's in the team's current process.
Compromises & Future States.
Working backwards from such an ambitious vision came with compromises from all sides. Ideas such as an internal messaging system and more complex interaction patterns could not be supported in the current IPP structure.
Though these versions were not used for various reasons, they sparked conversation for the possibilities of what IPP can be capable of.
Instacart: live ops vision
( 01 )
Instacart: Live Ops - phase 1
( 02 )
Amazon: GROCERY SHOPPING W/ ALEXA
( 03 )
DRMA: DEMYSTIFYING THE SKINCARE ASILE
( 04 )
I Design Stuff.
