Julie Delisle, HCP & Patient Specialist at conversationHEALTH
With many build-your-own platform options available, it may be tempting to consider developing your own conversational agent. How difficult can it be?
So you have embraced the conversational AI technology, you use it in everyday life when you’re online shopping, planning your next trip or looking for help with your financial transactions. You are now thinking about the great benefits, e.g. expanded service with 24/7365 availability, allowing users to self-serve in a human-like way, increased efficiencies, etc.
The benefits are obvious but where to begin? Here are some thoughts:
1. What are the options? There are really two — menu/button-based or open-text. They both have their applications, and their pros and cons. Menu/button-based conversational agents offer the user a choice from several options, presented in the form of menus or buttons, whereas open-text conversational agents offer a two-way dialogue that stimulates conversation with your customers.
(Click-to-chat is not an automated technology — it still requires human engagement and oftentimes 9-to-5 coverage only).
2. What are you looking to accomplish? Looking for new ways to engage with users? Insights on what their real needs are? Reduce the volume of incoming calls? Expand your reach? Deliver a superior user experience?
If you want return users, you need a seamless user experience. conversationHEALTH’s conversational agents, for example, leverage learning across all of our life sciences-specific conversational agents, allowing for understanding of industry-specific jargon and 97% accuracy in answering customer’s inquiries. That takes a lot of time, effort and learning. This will apply to other use cases too.
3. Don’t underestimate your teams’ involvement: Have no doubt — customers will engage your conversational agents, and at scale. What pressures will be placed on internal resources, both in IT and other departments that will be affected by this deployment? Who are your NLU/NLP experts? Your data scientists and engineers? Your conversation designers? Your voice specialists?
Be realistic when assessing the resources, time, and energy required to build a conversational agent. It’s like teaching and training a human — it takes serious effort!
4. Build to your industry requirements: The large majority of conversational platforms are industry agnostic. Makes sense — maximize the use of the platform. But Life Sciences isn’t just another industry. It’s a complex one, with significant Compliance, Medical, Legal and Regulatory requirements. By audience type. By market. That’s a lot of requirements to understand and people to involve.
What about Adverse Events and Product Complaints? How will the conversational agent recognize and escalate a possible adverse event? How will it communicate this time-sensitive information to the appropriate teams? How accurate will it be at doing so?
5. Speed to market: This may be one of the biggest mirages of the build-your-own option. On the one hand, it’s easy to set up a test case or two on some of these platforms. But how do you know you’ve identified the right conversations, intents and utterances? What are the accuracy and response rates of the conversational agent? How many users have you frustrated before getting to a working model?
Equally important are the many integrations of the conversational agent into the existing technology stack of the company. A conversational agent should never be an island by itself but, rather, deeply and smartly integrated into the Sales and/or Marketing mix. In addition, how is data used at the individual level for ongoing personalization of engagement, and in aggregate to re-inform go-to-market strategy. Conversational agents shed some of the richest primary data for any organization, it’s important that that data isn’t stranded in the solution.
6. What about scalability? This is where the concept of build-your-own starts to falter. Life Sciences companies are most often global powerhouses that need global solutions. Innovative solutions should also be scalable in the context of a country’s portfolio, multiple lines of business, or better yet, global expansion. And of course compliant!
Fundamentally, Life Sciences companies core competencies are product discovery, development and delivery — not software development or technology optimization. As such, build-your-own platforms may be helpful for internally-facing MVPs (minimally viable products) but not for externally-facing proof-of-concept work. And ultimately, certainly not scalable to the needs of globally-minded Life Sciences companies.
In short, I guess you can see my bias to buying a platform over building, and working with a partner who can help you from first project to scaling. As Forbes states in Don’t Blame The Bot, Blame The Brand “It’s inevitable and undeniable. Chatbots and AI technology are the future, and the future is here”.
We love educating about conversational AI and believe bringing great ideas together is always the start of something exceptional. As you launch into 2021, book a session with one of our conversational AI experts to learn how you can best leverage digital solutions to engage HCPs, patients & consumers.