How Intercom cut software cost capitalization from three weeks to three days

Intercom is an AI customer service company powering support operations for over 30,000 businesses worldwide. The company's engineering organization consists of hundreds of team members, all of whom work to grow its main products like Helpdesk and Fin, an AI customer service agent.

Juci Kulloi is a Staff Strategic Operations Manager, and her team sits at the intersection of Intercom’s R&D, finance, go-to-market, and legal departments. She works closely with Oran O'Dowd, VP of Engineering at Intercom, to align on quarterly cost capitalization efforts, which have historically been one of the most time-consuming and manual processes at Intercom — and one of the first areas where Span delivered measurable, almost immediate relief.

CHALLENGE

Intercom needed to replace a painfully manual cost capitalization process

Every quarter, Kulloi had to collect data from more than 20 engineering managers. Each EM manually mapped their team members' time to projects in a Coda spreadsheet that wasn't connected to GitHub or any engineering system. The work was prone to human error on multiple levels.

"It was essentially asking them to duplicate work they were already doing somewhere else, but in a specific format," says Kulloi. "And then there were always issues with the data because it was not automated in any shape or form."

The process took roughly three full weeks each quarter, including the weekends and late evenings often needed to meet deadlines.

During this time, not only were teams at Intercom working to collect all the data needed, but they also had to manually check for errors. Failing to do so would leave Intercom open to everything from financial blind spots to difficulties in prioritizing projects for the months ahead.

SOLUTION

By automating cost capitalization, Span shed a new light on engineering investment at Intercom

Span directly addressed Intercom's most pressing challenge by automating cost capitalization workflows, which in turn made engineering investment visible in one place.

Span connected directly to Intercom's GitHub data to automatically capture engineering work activity. Rather than asking EMs to manually log team members’ time, Span ingests pull request data and uses AI-assisted mapping to categorize work against a defined project list. When Intercom reviewed their Q4 data, roughly 50% of engineering work was already mapped to the correct projects before any EM was involved — up from 0% in the previous manual process. The remaining work requires lighter-touch categorization and an EM approval workflow where managers review and confirm pre-populated data rather than building it from scratch.

Built on the same data foundation, Investment Mix gives engineering and operations leaders a consolidated view of work and time distribution across projects, work streams, and teams. Before Span, this information was scattered across team-level cycle plans, individual spreadsheets, Coda docs, and GitHub data that had to be manually assembled. Once the cost capitalization data was flowing automatically, Span could surface this in one place — showing the scale and trajectory of engineering investment over time, including when projects are ramping up or winding down.

RESULTS

Span helped Intercom transition from manual reporting to operational clarity and efficiency

Cost capitalization workflow reduced from three weeks to three days

What previously took close to three full weeks of the quarter for Kulloi now takes roughly three days at most, including the EM approval workflow. Additionally, the pre-mapping work that Span conducts for teams reduced EM reviews at Intercom to 10 to 15 minutes each. The hours of post-processing to fix data quality issues have largely disappeared because the data is pulled directly from engineering systems.

This means that Intercom’s engineering and R&D teams no longer have to dedicate an entire cycle goal to cost capitalization efforts.

Eliminated manual overhead and institutional risk

In many cases, Kulloi was the single point of contact and the person who most understood how the data was assembled end to end. That created real institutional risk, particularly for a process tied heavily to audits and financial operations.

With Span, the data is tracked in a system, categorized consistently, and historically preserved. If Kulloi ever needs someone else to step in, they can easily navigate the data at first glance.

A clear view of where engineering effort actually goes

Span gave Intercom's operations and engineering leadership something they had never had in one place: a clear picture of where engineering effort is going across projects and work streams. Before Span, understanding the true scale of a project required pulling data from separate systems and manually piecing together a view.

With Investment Mix, Kulloi can see which projects are the most time-consuming, which are winding down, and where they can reallocate efforts. These insights feed directly into headcount allocation discussions, grant applications, and decisions about when to wind down or scale up investments — conversations that rely heavily on accurate data-driven evidence.

“I often sacrificed a weekend, a late Friday evening, or a bank holiday to get the data mapped on time. That's no longer needed."

– Juci Kulloi, Staff Strategic Operations Manager

Looking ahead

With the cost capitalization process now automated and backed by reliable data, Intercom subsequently receives a broader view of their engineering work and investments. The team continues to make full use of Span to gain clearer insight on engineering investments.

"As we scale the engineering org and shift to fully AI-first, end-to-end development, having a system of record for how work maps to priorities is going to be foundational,” says O’Dowd. “Span is becoming part of how we operate, not just how we report."

Everything you need to unlock engineering excellence

GET A DEMO