Intercom Analyzes Customer Data at Speed and Scale to Save Sales Teams Thousands of Hours

A frustrating customer service experience is something no one wants and almost everyone has encountered, and those negative experiences live long in the memory, damaging reputations and impacting the bottom line. Addressing this issue is integral to the success of any company. Faster resolutions mean happy customers, and happy customers mean happier service agents —  all of which can breed brand loyalty, employee satisfaction and genuine competitive advantage. 

Intercom is an AI-first customer service platform, combining a powerful AI agent called Fin with a full suite of tools to help clients resolve queries quickly, at higher quality and lower cost. In all, it provides a better support experience for customers, agents and leaders alike. 

Data is integral to Intercom’s service offering. Not least, it powers the company’s AI agent, generating answers and handling millions of customer interactions. But there are other uses too, primarily enabling the company’s internal sales teams to monitor clients’ resolution rates, deflections and customer satisfaction to optimize service performance. Previously, however, a range of disparate tools and an outdated data stack meant these vital customer insights were time consuming and complex to come by. 

Today, Intercom uses Snowflake’s AI Data Cloud, running on Amazon Web Services (AWS), to provide its sales teams with vital insights into customer behaviour through a unified Sales Cockpit. This Cockpit unlocks a number of use cases, from showing reps which customers they should be focusing on, to optimizing the performance of those customers’ service agents. Using it, Intercom’s reps save thousands of hours every week previously spent searching for information, and can rapidly apply actionable insights to improve service delivery. The resulting savings are estimated at $1.4 million.

Built in a matter of weeks and deployed on the Snowflake AI Data Cloud in days, Sales Cockpit is a fully featured web app for Intercom’s sales teams. It provides a single point of contact for reps to get customer information, using automatically generated insight decks and other documents so they can make better decisions for their customers.

Intercom’s data scientists also use the Sales Cockpit’s LLM analysis and optimization tooling to improve Fin’s performance for customers. The result is a truly one-stop shop for insight and service optimization, which has seen a great response from sales reps and added value for Intercom’s customers. 

“With Sales Cockpit, we’ve been able to create a more sophisticated user experience and give reps the tooling to carry out their own LLM analysis and understand their customers better,” says Louis Ryan, Senior Director of Data Science at Intercom.”This visibility makes identifying areas for improvement far easier.” 

Sales Cockpit has seen huge efficiency gains for the company’s sales teams. Whereas previously it took about 40 minutes to send a personalized outbound email, it now only takes around 30 seconds for an AI agent-driven workflow to research a contact, after which a sales rep can review the results. In addition, customer insight decks that once took 3–4 hours to produce are now ready in less than 10 minutes, with around 500 prepared each month — all requiring no manual effort. 

These improvements have led to a time saving of more than 2,000 hours a month which equates to an estimated total benefit of around $1.4 million. And these figures can be expected to grow as more users are onboarded. Currently, Sales Cockpit serves sales reps working on existing business, 35% of whom are daily users, with 55% accessing the application weekly. In the future, the company plans to expand access to the tool across the organization.

Gaining a complete view of customers is vital to giving them what they want, but it’s not always easy. Prior to using Snowflake, Intercom’s growth had outstripped its Redshift data infrastructure. The company experienced significant issues, particularly the proliferation of data tools and dashboards. 

“If a sales rep wanted to search for accounts, research them and prepare for a QBR or meeting, it took eight or nine different tools and dashboards,” says Ryan. “This was not the right experience for our users; we needed to provide access to information in a way that was less disjointed.” 

The solution was to create workflows tailored to individuals and job roles, and to incorporate AI for core workflows. And that had to be supported by a robust, flexible data platform.

Building the Sales Cockpit on top of Snowflake data and deploying it with Snowflake Container Services gave Intercom several advantages. Firstly, it meant there was no need to move any data, which removed any security or governance roadblocks. Secondly, it enabled the solution to be developed and rolled out quickly, with minimum engineering effort. 

“The deployment of Sales Cockpit on Snowflake Container Services took a single engineer a few days’ work,” says Ryan. “Not having to worry about things like permissions or deploying it onto our own internal architecture saved us at least a month’s worth of time.”

The result? An entirely more efficient and effective way of serving up data. With the ability to build solutions like this quickly and easily, Intercom now has data scientists working as end-to-end builders — identifying issues and inefficiencies within the company’s GTM organization, designing and implementing solutions, measuring their impact and applying iterative improvements.

“We’ve gone from building dashboards to experiencing a whole new world of building products,” Ryan continues. “Without Snowflake, this would have required longer cycles, different roles like product managers, engineers and data scientists, and of course, lots of investment. It’s a fundamentally more empowering way of working.”

Looking to the future, Sales Cockpit will play an important role in driving greater efficiency for Intercom’s Research and Development and Go To Market teams. 

“Sales Cockpit is our primary vehicle for those initiatives,” says Ryan. “And we plan to keep up with everything Snowflake releases to continue improving it.”

As part of this plan, Intercom intends to use Cortex Search to retrieve unstructured data in a new agent and is conducting a hackathon to explore these capabilities. 

“Those capabilities will ensure people get the answers they need, while also helping our data scientists avoid basic questions and focus on more strategic work,” says Ryan. “We expect it to be a major efficiency gain.” 

Not only will this help Intercom’s reps understand customer needs more succinctly, it will ultimately help those customers deliver service interactions that stand out — for all the right reasons. Those that resolve issues more quickly, build brand reputations and breed true customer loyalty. It’s a future full of promise for customers, service agents and business owners.