Lessons from Shopping for Intelligent Assistants

AI_IAI recently had the opportunity to help a small business evaluate intelligent assistant solutions. This particular company sells products online and they’re not large enough to have a call center. As their business grows, they’re finding themselves spending a lot of valuable time answering basic questions from customers and suppliers. They were seeking an intelligent assistant / virtual agent to take over as much of this question answering as possible. As I helped this company in their search, I came away with two lessons that might be valuable for other companies in a similar situation.

Facts First, Conversation Later
When most people hear about intelligent assistants, they assume that these assistants are akin to chatbots. They think of them as chatty and highly conversational, engaging the customer in a back and forth dialogue. They also have the notion that it’s easy to set up a chatbot that can just talk to their customers for them.

In fact, most of today’s industrial strength commercial intelligent assistants are really sophisticated search engines that are given the thin veil of a human personality. The personality is mostly as simple as a person’s image. For voice-enabled assistants, the personality is reinforced by a human voice. But the focus of the assistant is on search, not on conversation.

These high-powered intelligent assistants utilize proprietary tools to interpret a user’s question and then piece together the best possible answer based on related information from your website and other content sources. This technology can be much more powerful than simply matching on keywords. It can leverage semantic models that improve search results. For example, if a customer searches for delivery times and your website only talks about shipments, the search tool will know that delivery times are related to shipping information, enabling the intelligent assistant to give your customer a correct response.

As you can see, an intelligent assistant solution will only be successful if you’ve invested time in building out your knowledge base of FAQs. Does your website contain answers to all the typical questions that are being asked? If the answer is yes, that’s great. You’re ahead of the game. If the answer is no, then focus on building out your FAQ database. The best thing you can do for yourself and your customers or other stakeholders is to document all the key information about your product and/or services. If you are an online retailer, it probably won’t be possible for you to document all the questions related to every single one of your products. But you can answer questions about general policies like setting up and editing account information, payments, shipping, returns, and discounts. You can probably also document answers for FAQs related to specific product categories.

While you’re evaluating intelligent assistant solutions, go ahead and publish your FAQs. Maybe you can get constructive feedback on them. You should get in the habit of constantly tweaking your knowledge content. To get the most out of your future intelligent assistant investment, you’ll need to continually find the gaps in your content and fill those gaps with answers that the assistant can provide.

Don’t Be Penny Wise and Pound Foolish
The second lesson I learned is that commercial intelligent assistant solutions can come with a hefty price tag. These solutions are Software as a Service, meaning that your web servers will be connecting to the vendor’s complete solution in the cloud. Pricing is based on the number of customer interactions with the assistant on a monthly basis.

If your company has to field tens of thousands of incoming questions a month, you might be looking at a cost somewhere in the range of $350K to upwards of $1 million a year. That cost will likely include services required to set up and deploy the solution. If you’re a smaller company and expect fewer than 5000 inquires a month, you may be able to get a great solution for less than $1000 a month. You’ll likely need to add on at least another $5K for set up and implementation.

Before pursuing such an investment, it pays to ask yourself what the goals are for your business. In the case of the company I was assisting, they were shooting to dramatically increase the scale of their global sales force, requiring them to train and interact with a much larger volume of resellers and suppliers. Their current method of answering questions as they came in via Skype or other methods simply wasn’t scalable. Getting the right intelligent assistant solution in place could become a key element in their success strategy. For that reason, it made sense for them to consider a larger investment.

If you are serious about pursuing a solution, there are great intelligent assistant providers out there who will be willing to work with you. Document your requirements as best you can, so that you’ll be ready to give the vendors good information when you begin engaging with them. You’ll need to have an estimate of how many questions you think the assistant will be asked each month.

When you see the price quotes, remember what you’re getting. These are search powerhouses that are typically wrapped in a whole host of other valuable services. Most include complete monitoring of all interactions with the assistant, insightful analytics, and other data to help you improve the assistant over time. They may even include support for live chat, so that conversations can be seamlessly escalated to your human call agents.

If you really want to grow your business, and if the ability to answer lots of incoming questions is key to that growth, consider an investment in a proven, commercial intelligent assistant solution. Over time, it should prove to be much less expensive and more readily scalable than going down what may appear to be a cheaper path for the short-term.



Categories: Conversational Intelligence, Intelligent Assistants, Articles

Tags: , ,