When John Collins joined LivePerson in 2019 as senior vice president of quantitative strategy, his mandate was to automate the back-office function and create systems to collect, clean, and connect all the data the company was generating to drive growth and improve margins.

A year later, after he had been promoted to CFO and had two quarters in the top finance seat under his belt, the company went from negative margins to meeting the rule of 40: achieving 40% growth and profitability. 

“We achieved that partly through revenue acceleration, but another big component was bringing down G&A as a percentage of revenue by automating so many workflows,” Collins said last week in a CFO Thought Leader podcast.

The company, which provides an AI tool for businesses to text back and forth with their customers, took a chance on giving him the CFO job, Collins said. He had never held the position before, but the quick margin improvement his work helped drive validated the decision and the resources it gave him to build out the automation project, Collins believes. 

“The results spoke for themselves,” he said. “The impact we had in just the first six months before I took over the CFO role, and in the subsequent quarters — I’ve been in the seat for a year now — was so rapid and impactful that [analysts] got on board with the vision pretty quickly.”

Data models and decisions

To execute on the automation plan, Collins pulled together data scientists and engineers into what he calls the data models and decisions (DMD) team. They designed and implemented the systems for collecting and processing the company’s front office data — sales, revenue — which help fuel its growth. But, equally importantly, they designed and built the systems for collecting and processing back-office data, which helps fuel its margin improvement. 

“These are additional metrics that I'm focused on because I have a vision for reshaping … what internal operations does for the business and the kind of data advantages we can generate to propel our organization ahead of the competition,” he said. 

Those metrics include how much staff are tapping into the company’s data lake to improve decisions and performance. 

Top talent

Years earlier, both as an MIT MBA student and as head of his own investment company, Collins had been involved in building data systems, so he tapped his network of data scientists and engineers to staff up his DMD team.

“There are not many highly skilled data scientists that work in the back office,” he said. “Same for engineers. Most of these people have their choice of jobs [and prefer to] work on product development for the core of a business that is put out in the market. That’s where the interesting and desirable work has been.”

But he was able to attract them, he said, by giving them the opportunity to turn back-office automation into a value-driver for the organization.

“The vision I’ve set is so novel and appealing we’ve been able to attract the same caliber of talent in science and engineering that our core technology arm under the CTO is attracting today,” he said.

The team also likes how quickly it gets rewarded for its work.

“The work we’ve done presents many rapid wins,” he said. “We have stakeholders coming to [us] saying, ‘Please free me from this repetitive, manual workflow.’ In many cases, because of the data lake architecture we’ve built, and the data we’ve onboarded and cleaned, and connected, we’re able to deliver those automations in mere days.”

Initial improvement

The first improvement Collins and his team made was enabling the ability to track customer usage of the platform. 

“Part of the problem was due to dirty and disconnected data,” he said. “Most large enterprise customers had dozens of sub-accounts through which they access the platform that weren’t necessarily mapped to the parent account. In addition, attempts to forecast customer usage, which is critical for billing, go-to-market, product development, strategy, were basically simple averages of incomplete data. But despite the simplicity of these models, wrangling the data was manually intensive and error-prone.”

The team deployed automations and leveraged time-series analyses and data features relative to specific products and industries to map accounts to their relevant parents in the data lake.

“The strategic value-add here was not only saving the time of the people who were responsible for the manual work, but also revealing revenue leakage, because of those manual oversights,” he said. “We weren’t billing for all of the usage because we didn’t know all the usage without this enhanced data cleanliness.”

The improvement also enabled finance to implement usage-based billing automations, which saved significant time for the billing team and provided upsell opportunities to the go-to-market team responsible for developing the company’s enterprise base.  

“The ripple effect from this one automation, from connecting and cleaning this one subset of data, had a really profound effect across the organization,” he said.