Machine learning and artificial intelligence will have a radical impact on enterprise software and the changes are already underway at a frightening pace.
That’s the view of Adrian McDermott, president of products at Zendesk. He argues much of what enterprise software systems do is relatively repetitive and can be improved by machines.
“We are slowly but surely taking every aspect of our product and figuring out how we can use AI to make it better,” McDermott told Which-50 between sessions at the company’s local user conference in Melbourne last week.
“For most software companies, Zendesk included, we are facing the same challenge which is eventually all developers will need to know how to build a machine learning model. It is going to become one of the algorithms that we use to solve problems in the same way we use databases and search engines.”
Founded in 2007, Zendesk builds customer service software. It has more than 7,200 paid customer accounts in Australia, and a research and development centre based in Melbourne working on building machine learning into its products.
Zendesk unveiled its first ML product in 2015, called Satisfaction Prediction. Based on analysis of all the customer support tickets a company receives, the feature predicts how likely a ticket will be resolved with a positive or negative outcome.
“Pinterest use this to route tickets; tickets that have a low predicted satisfaction rating are sent to a specialist team,” McDermott said.
The company has also used ML to power a feature known as Answer Bot which matches customer queries to help desk articles with the aim of resolving customer queries without involving an agent.
From there, the machine can identify the recurring customers questions across the system and suggest content that should be added to the knowledge base. Getting a lot of questions about vegan ingredients? Zendesk’s Content Cues will suggest updating an article on product ingredients.
“We are able to use AI to get people to enrich the knowledge base, which then drives up the self-service ability of their system and drives down their support cost,” McDermott said.
Think like a growth marketer
Artificial intelligence will also be applied to making customer support more predictive, rather than waiting for a customer to make contact after something goes wrong.
McDermott argues customer support needs to embrace a growth marketing mindset to anticipate customer needs and provide a better overall experience.
“One of our hypothesis is that our buyer is owning more of the customer journey and is becoming more sophisticated. They have a right to the same technology that the most sophisticated growth marketers are using,” he said.
“Those tools should not just be the purview of people who deal with initial conversion and the new customer funnel. They have the right to use those tools for existing customers. It isn’t just customer marketing. It is a way to build better experiences with the brand through nurtured support.”
To advance that ambition, Zendesk acquired a marketing automation company called Outbound.io in 2017. The product has been rebadged Zendesk Connect, a proactive messaging tool which allows support professionals to create dynamic customer segments all past customer interactions.
For example if ecommerce orders are delayed, Zendesk Connect can be used to build a customer segment based on transaction history and send out an apology message with a discount code.
Changing customer base
Zendesk went public in 2014 with an annual revenue run rate of $100 million. In just the last four years, annual revenue run rate has grown to $500 million, half way to its goal of $1 billion by 2020.
Nearly 40 per cent of Zendesk’s revenue now comes from larger enterprise customers. According to the company’s recent earnings, last quarter Zendesk closed 60 per cent more deals with an average contract value of $50,000 or more compared to a year ago.
McDermott said the changing customer base has occurred in two ways – one from existing customer scaled rapidly and by deliberately targeting enterprises.
“We signed up a lot of customers who grew a lot, people who signed up with three agents became 10,000 agent customers. We are proud of the fact they didn’t have to throw us out half way through. We built a product that scaled to those use cases and can grow with them,” he said.
The company also used the funds raised from its 2014 IPO to build a base of enterprise customers. From a product perspective that involved adopting a set of hosting, security and governance features primarily focused on large companies, as well as features designed to manage thousands of agents across different locations.
Zendesk also established a team of account managers dedicated towards enterprise customers.
“The way those accounts are managed on an ongoing basis is radically different to your smaller online customers, who have very different expectations.”