The past year has seen artificial intelligence making impressive inroads into industries like healthcare, real estate, network security and even potentially democratize election information. So it may come as no surprise that AI is also changing how things are done in the world of sales and marketing, helping to automate time-intensive tasks — such following up on initial customer inquiries or cultivating potentially lucrative sales leads — and passing that lead onto a human when the time is right.
That’s what California-based company Conversica is doing with its virtual sales assistant, which uses a cloud-based “conversational artificial intelligence” platform to help companies better engage with customers and generate more revenue.
AI with its Own Name and Email
The company says that its aim is not to replace humans, but to team AI’s efficiency and cost-effectiveness with the human-side skills of establishing trust and deepening a company’s relationship with a customer. Here, AI is yet another tool for salespeople and marketers use to manage, make sense of and act upon vast volumes of data.
AI-enabled automation as seen in Conversica’s technology would allow companies to contact every single lead that crops up, with unflagging diligence, around the clock. To give clients the impression that they are in dialogue with a person, Conversica’s intelligent sales assistant uses AI and machine learning to create emails or SMS text messages calibrated to engage a potential customer in a continuous conversation.
While it may not be human, Conversica’s virtual sales assistant is assigned its own name and email address, allowing it to autonomously contact, engage and ‘nurture’ leads before handing them off to a human at a predetermined point during in the process.
Increasing Lead Engagement
Technology such as this can ease the difficulty of following up on every single lead. Each instance of contact with the customer, or “touch” as it’s called in industry parlance, costs money and time, so human employees may, therefore, engage a potential lead only a couple times before moving on to the next one.
In contrast, AI can keep the cycle going, while also personalizing the conversation with the client. According to the company, its virtual sales assistant can help sales teams implement best practices at scale, such as boosting the touch rate to 7 or 10 for each lead, while freeing human salespeople to pursue only the most promising leads.
Conversica has helped businesses like IBM, Epson, Talend and Cake increase their lead engagement by 33 percent, reduce the cost-per-lead, and generate more than $8 billion in sales revenue, thanks to this persistent approach, according to the company.
To gain some deeper insights into how the platform works, we asked Werner Koepf, senior vice president of engineering at Conversica, to provide further details on their virtual assistants.
Could you describe in more technical detail how Conversica’s AI works?
Werner Koepf: In essence, we use artificial intelligence and machine learning to power the conversation between our “virtual assistants” and our customers’ leads. In all, we use four “intelligences” in our process.
First, we use natural language generation (NLG) to create the emails that the assistant sends to engage a lead in conversation and to continue that back-and-forth conversation. Second, we use natural language processing (NLP) classifiers that discern insights from the messages that came back from the lead. Third, we run those insights through our inference engine to determine the right thing to do next (like continue the conversation, update the customer relationship management [CRM], notify someone, turn over to a human, stop messaging, etc.). And, fourth, we introduce human intelligence in the exceptional cases where the AI is not able to interpret human responses with a high degree of confidence.
How is Conversica’s AI designed to scale up?
The first aspect that makes our AI scalable is that it is a layered approach using natural language generation, NLP and the inference engine, allowing us to make improvements independently to any one of those three layers. Secondly, our AI is based on a very elegant and modular concept of a two-way conversation, allowing us to easily add additional building blocks to extend our conversations to cater to new use cases or create additional virtual assistants.
What were some of the challenges in developing this kind of AI?
As always, the hardest part is the execution. That is, getting things to actually work in our customers’ complicated sales and marketing environments, which include particular technologies (like CRM and marketing automation), particular processes and of course, people.
By doing this over many years and millions of back-and-forth conversations, we have developed a tremendous amount of intellectual process and best-practice knowledge for building modular layered systems based on AI. But, that could just be a science experiment, if not applied to solving real business problems for our customers. The integration of our AI into our customers’ technologies and processes was the hardest part to solve, followed closely by training the machine over millions of conversations to accurately understand all the ways humans say yes, no, maybe, maybe later, not right now, etc.
How does Conversica’s AI integrate with CRM and marketing automation systems?
For our flagship AI Sales Assistant, we are working with our customers’ leads and need to integrate with their lead management systems, such as CRM and marketing automation, to both retrieve the lead records and to update our progress on contacting and engaging them. We have standard integrations with many CRM and marketing automation systems (like Salesforce, Marketo, Eloqua etc.) so that our customers are able to get up and running quickly.
As an example, let’s suppose a new lead comes into Salesforce from a customer requesting a demo on the website. That lead is added to a campaign in Salesforce that syncs the lead with a conversation in Conversica. The Assistant starts the outreach to the lead over email, sending updates back to the Salesforce lead record as the conversation progresses. When the lead agrees to a meeting and the Assistant has verified the phone number, the salesperson is alerted and the information is again updated in Salesforce, often also creating a task for the salesperson to follow up. All of this happens completely automatically with no human intervention.
Why aren’t more companies using AI in their sales strategies? What are some reasons they should?
We see AI being used more and more in sales and marketing, and we find that it helps to categorize the uses into two classes: Advisory AI and autonomous AI. Advisory AI analyzes and makes recommendations for a human to implement, such as recommending which sales opportunity you should work next given the likelihood to close, whereas autonomous AI actually takes over a task and does it without human intervention, like our Assistants that contact, engage and qualify leads. Both are important to boost the effectiveness and efficiency of the sales process because some tasks are best done by humans and some are best offloaded to AI.
What other developments does Conversica have in store for the near future?
We continue to enhance our AI Assistant with new capabilities, expand into new industries with different conversations and add new assistants that automate additional business conversations. For example, we are equipping the assistant to converse over new communication channels, such as messenger applications, and adding new languages in which she can carry on conversations. We continue to enhance our dashboards and reporting so our customers have better insight into what their assistants are doing and how it’s improving their key metrics. We also always testing new AI algorithms, new conversation styles, and new personality traits to boost the engagement the assistant has with leads. It’s an exciting time to be in the AI business.