For large SaaS companies, the process of collecting revenue from subscribers is intricate. As companies grow, their billing systems need to expand to accommodate variables such as different pricing plans, currencies, sales channels, and complex prorating and discounting logic, while supporting multiple other technology platforms and operating processes. The limitations of a billing system can hamper customer experience—and ultimately a SaaS company’s growth.

That was the challenge facing Intercom, the leader in AI customer service, in 2023. During more than a decade in business, Intercom had emerged as a leader in online customer service, with a base of more than 25,000 customers using its innovative online customer service and support tools. But as technology kept evolving, Intercom saw a new opportunity in the growing capabilities of artificial intelligence. After ChatGPT launched in 2022, Intercom committed itself to rebuilding its customer service platform from the ground up with AI technology, rolling out new tools, including Fin AI Agent, a bot that talks directly to customers to solve their problems, and Fin AI Copilot, which talks to support representatives to improve their efficiency when assisting customers.

To support the next phase of its growth while delivering a better customer experience and these next-generation tools, Intercom needed to develop new pricing models and a more sophisticated billing platform—a task that presented its own set of challenges.

Fragile legacy systems and customer service challenges

Intercom was building new products and pricing tiers on top of a fragile, inefficient legacy billing system and internal infrastructure. It had a complex mix of homegrown solutions and products from multiple vendors that didn’t always work well together. The third-party billing vendor’s API was a black box: customer-facing teams couldn’t determine the cause of the billing errors on their own. That meant they had to escalate problems to the engineering team, which would often spend weeks trying to find a solution.

This messy setup meant that attempts to develop new products or pricing models were often plagued by billing errors. And while customers loved Intercom’s products, they were frustrated with Intercom’s pricing and billing experience.

The company’s leaders wanted to treat their customers’ billing experience with the same craft and care they used to build their products, but they didn’t want to dedicate engineering efforts to building a custom SaaS billing engine. The prospect of migrating the company’s billing system to a new third-party provider was also daunting, as it, too, would be an intricate and time-consuming project. That’s when Intercom turned to Stripe, which has extensive experience in SaaS billing implementations.

In early 2023, Intercom asked Stripe to help create a modern billing system on Stripe Billing that matched the company’s advanced SaaS platform. Intercom’s goal was to rely on Stripe’s prebuilt technical features and established processes, diverging from Stripe best practices only when it was necessary to do so. That approach would allow them to iterate faster on new products or pricing models.

Conducting a feasibility assessment

A SaaS billing platform migration begins with a detailed feasibility assessment to ensure the solution fits a company’s needs today, as well as its growth plans for the future.

Stripe assembled experts from its sales, solutions architecture, product, and professional services teams to lead the feasibility assessment process. This group spent six weeks conducting detailed interviews with a long list of Intercom personnel. For Intercom, that group included members of the billing engineering team, product managers, experts in accounting and accounts receivables, data teams, technology administrators, and sales and customer service personnel.

Their conversations covered how Intercom’s teams interacted with the billing system, including their current challenges and the additional capabilities they’d need in the future to support the company’s pivot to AI-powered solutions. Along the way, the Stripe team took an inventory of all the other technology platforms that connected with Intercom’s billing process to ensure they had captured all the necessary integrations they would have to perform.

“Pricing and billing are inherently cross-functional concerns, and each team involved has its own priorities and interests,” said Lynch. “Navigating the balance between these sometimes competing interests requires a lot of collaboration.”

Drawing on those conversations, the team developed a feasibility assessment that identified important elements of the project, starting with a documentation of Intercom’s current platform architecture and the key third-party integrations that Stripe would need to support. For example, Intercom had a custom internal billing engine and a billing admin UI that allowed the billing ops team to update and edit subscriptions. Subscriber card payments were handled by Stripe. Other third-party integrations included NetSuite for accounting, Salesforce for CRM and sales quoting, Avalara for tax compliance, and a Redshift data lake.

The feasibility assessment also highlighted key billing challenges and problems that Intercom needed to solve. The company’s stakeholders were looking for an array of upgrades, starting with an improved billing experience for customers that reduced the number of confusing or incorrect invoices. At the same time, the billing system needed to be easier for internal users to work with, with features that included improved data synchronization and visibility. They also wanted to enable modern functionality, such as a simple process for managing discounting and ramped deals, automated recognition of revenue to GAAP standards, and the ability to add payment methods. Above all, they wanted to decrease the burden on engineers by reducing the number of time-sensitive issues that created frequent interruptions.

Finally, the assessment outlined a project implementation plan that provided a staged roadmap for a full migration, starting with an MVP. Intercom also wanted to limit—and clearly define—the period of time in which they would have to run two billing systems side-by-side. The plan included a proposed solution architecture for each stage, deliverable dates, and risks or threats to watch out for in the implementation process.

With the feasibility assessment and project implementation plan complete, the Intercom and Stripe teams were in a strong position to begin the migration.

Managing implementation

To manage a project of this scale, Stripe assembled a team of experts pulled from the company’s talent around the world. The team featured members from the data engineering and finance teams in San Francisco, as well as implementation consultants in the US and Dublin, integration engineers in London and Dublin, and a Dublin-based engagement manager.

The Stripe implementation team opened up several channels of communication with Intercom stakeholders to guide the process. They exchanged thousands of messages on Slack, held weekly and ad-hoc engineering implementation check-ins on Zoom, participated in multiple cross-functional in-person workshops at various stages of the project, and committed to weekly project steering committee meetings to ensure the project remained on schedule.

This approach fostered a two-way exchange of information that helped the teams move quickly and accurately, despite all the moving parts. For example, Intercom provided regular updates on the ongoing development of new pricing models and their expected timelines, keeping the Stripe team looped in on details that would affect the implementation. In addition, Intercom’s development team sat down with Stripe to fully explain how the company’s tech architecture worked and to highlight product features they had added since the feasibility assessment process—developments that could have otherwise knocked the billing launch off course.

On the Stripe side, meanwhile, the team used its experience to flag potential roadblocks in Intercom’s architecture before those issues became unpleasant surprises during implementation. Stripe also shared its experience with SaaS billing to recommend integration methods and best practices for operation.

All of these efforts enabled the Stripe team to develop an approach that simplified Intercom’s billing process. The new architecture eliminated the third-party layer between Intercom’s billing engine and payment processing, streamlined data transfer to other third-party software programs, and enabled real-time billing changes that previously had to be managed by customer service or operations personnel through the billing admin UI.

Proposed architecture:

Completing third-party integrations

In a billing platform migration, establishing integrations to all of a company’s critical technology systems is as important as developing the billing architecture itself. Getting these connections right the first time ensures that the new billing system doesn’t disrupt any other business functions.

As the implementation team worked through the list of Intercom’s key third-party platforms, they used a mix of custom integrations and prebuilt connectors to optimize performance and make the most efficient use of developers’ time. For example, Intercom was able to maintain its current Avalara setup by using a prebuilt Stripe connector, AvaTax for Stripe, that listens to webhook events from Stripe Invoicing to calculate and remit taxes based on the country where an invoice is issued and the customer’s country. Intercom also used the Stripe Fivetran connector, built by Fivetran, to quickly plug all Billing data into its existing data lake with no integration effort.

Although Stripe provides a ready-to-use Stripe Billing Connector for Salesforce CPQ, Intercom chose to use the Stripe API to build a custom integration directly with Salesforce. Building this integration layer with Salesforce ensured that customers who came to the company through the self-service website and those brought on by the sales team followed the same subscription experience.

Going live

Prior to launching the new subscription options, Stripe met with Intercom’s operations team to orient them on the internal Stripe Dashboard. They reviewed how the new billing platform integrated with other internal systems and showed them how to evaluate potential issues and respond to customer-facing requests.

Then in November 2023, Intercom went live with its new pricing models on Billing. The upgrades were significant: Intercom gained the ability to implement line-item discounts and to accept ACH payments from self-service customers—an important option for bigger customers that come through the self-service website channel. In addition, managing and deploying pricing changes is now faster, easier, and less error-prone.

Since then, Intercom has continued to follow the roadmap outlined in the integration plan, methodically moving existing customers into the new products and pricing models. This thoughtful approach is intentional, as Intercom wants to ensure that customers fully understand the value of the new products and the difference in subscription options and prices before upgrading their service. As of July 2024, Intercom had migrated 20% of its subscription customers and 12% of its revenue onto Stripe Billing.

As the implementation progresses, Intercom will be able to migrate additional billing functions, such as usage-based billing, to Stripe. And by eliminating the third-party layer between payments and billing, customer-facing teams can easily log in to the Stripe Dashboard and resolve most issues without having to escalate to the engineering team. For their part, the engineering team is no longer mired in keeping-the-lights-on work. As a result, they can focus on building and refining Intercom’s core AI-powered customer service tools.

Already, Intercom has seen its AI tools transforming the customer service experience. In one year, the company’s Fin AI Agent increased its average customer question resolution rate from 26% to 51%, with some users resolving 80% of customer questions via Fin AI. That success rate gives human customer service agents more time to work with customers on solving harder issues—helping boost overall customer satisfaction scores. Intercom’s Fin AI Copilot is automating the task of scouring through support documents and previously resolved queries to provide customer service agents instant answers to tough questions, which is helping users improve efficiency by 31%.

With these initial successes, Intercom expects to accelerate its pace of product development and innovation. And it can do so with confidence because the company’s new billing platform is sophisticated and flexible enough to support the ongoing evolution of its subscription models.