Improving a 3rd party integration

Silo is an enterprise resource planning software that helps distributors in the food supply chain streamline and connect their operations, logistics, and finances.

Context

Users who have their accounts payable and accounts receivable connected to Silo often use QuickBooks to store all their accounting information. The initial sync between the two platforms required a Silo team member to walk users through the integration set up because the process could be confusing and show premature errors. With a major backend update to the way Silo handled QuickBooks information, it was important to improve the integration process so it could scale and users could set up the integration independently.

Primary Problems

Silo is a robust platform that includes accounting and payment processing. Users can sync their QuickBooks account to Silo so everything stays connected, but the initial integration process was confusing, error-prone, and had product-specific requirements that users were unfamiliar with. Users who tried to set up the integration on their own would get premature errors and often needed to reach out to support to understand what was happening. This was not a scalable solution and needed to be addressed before a major backend update was released.

Issues with the current experience

• Unintuitive: unexpected patterns in UX and UI increased cognitive load on users

• Error-prone: the system would throw errors prematurely causing users to reach out to support

• Overwhelming: 30+ inputs show at once when users at setting up the integration

Goal

The goal was to streamline the initial integration experience, make it scaleable, and reduce cognitive load so users could complete the entire set up independently

This project came with a couple constraints: we could not make changes the general settings UI, and we needed to preserve confusing legacy categorization of assets that was scattered through the platform.

Product Goals:

• Improve the QuickBooks <> Silo integration in anticipation of increased users.

• Help users understand nuanced differences of how each platform categorizes information.

Out of Scope:

• UI updates to the overall settings page and layout

• Changing confusing legacy categorization of assets that was scattered throughout the platform.

Competitive Analysis

I conducted competitive analysis to understand what best practices exist for connecting two platforms.

Some questions I explored while analyzing:

• What does a user do to trigger the beginning of the integration?

• How many steps does a user need to take from start to finish?

• How many integrations are available?

• How does a user discover this integration exists?

• How difficult is the integration to set up?

• How does a user know it’s working?

• How well does the integration work?

Design Process

When connecting the integration, users had to fill out a very long list of inputs on an already very long settings page. Since we could not change the number of inputs that needed to be filled in or automatically fill it in, I explored finding ways to reduce the cognitive load on users to make it feel shorter and lighter.

Expandable card

I tested a version using expandable cards to make filling out the inputs a more focused experience. Once a user is finished filling out one section and moves on to the next, the prior section collapses. Users could still see how much information was required, but could decide if they wanted to return later to finish.

Separate page

I tested out a version where the integrations are broken out to their own page, but received push back on this since Silo only had one integration available at the time. We reserved this option for a future iteration when there is more than one integration available.

Modal

Breaking the steps out into a modal gave the experience focus and allowed us to use progressive disclosure to reveal more information as it became necessary.

Other changes made

• changed the disconnect button to red, to signal a destructive action

• added expense overrides to the settings pages so all changes can be made in a single place

• changed the switch to connect to be a button, signaling to the user that there are more steps to activate the integration.

Expense Overrides

For every expense users create in Silo, accounting users must add a ‘Silo expense category’ to the transaction so they can accurately track their expenses. When there is not a Silo category that easily matches their QuickBooks accounts, users have to create an expense override.

Previously, a user was only able to create an override from their Chart of Accounts (CoA) page from a buried drawer. This was only possible to create retroactively and could add complications to users bookkeeping. We chose to include these step in the initial integration process so the information could be as accurate as possible from the beginning.

This was a legacy feature that was now too embedded in the product to remove at this time, so I needed to find a way to educate users and prompt them to set overrides during the integration setup.

Since this was an optional step with a limited number of overrides that could be created, we added it to the relevant workflows with the option to add more lines as needed. Once the set up has been completed, users can easily edit their expense settings.

Solution

We finalized a stepped modal flow based on which accounts are being synced to Silo. Each flow allows a user the customization that is needed to fully integrate their accounts payable or receivable between Silo and Quickbooks. Users have greater control over their accounts, are able to find and edit them more easily, and UI elements are consistent with expected behavior.

Results

The results of this project were not able to be collected, but I would collect the follow metrics to determine its success and future improvements:

• Reduction in integration sync errors

• Reduction in support tickets to set up

• Increase of users actively using the Quickbooks integration

• Average time to complete the integration set up

Learnings

Collaboration, simplifying complexity

This project taught me important lessons in cross-functional collaboration and the value of each stakeholder’s input. Through the collaboration with a product specialist, I gained insight about user expectations in certain parts of the platform and tried to find ways to educate complex but useful actions within the platform.

This project was last minute and had a rushed timeline so I was unable to conduct any user testing. I believe the project could have benefitted from some user research and testing would have allowed for a stronger deployed solution.

If I could go back and improve this project, I would change some of the UI elements to have greater consistency with the rest of the platform. I’d also push for the integrations to have their own page since I anticipate more integrations becoming available. It would give users an easier way to navigate the overall settings.