Put simply; Conversational AI gives machines the ability to process/analyze speech so that they can interact with people in a human-like way.
Using NLP (Natural Language Processing), Machine Learning, and oodles of data, Conversational AI is designed to interpret meaning (from the human) and respond appropriately (to the human.) When used correctly, it’s a highly personalized way to communicate with customers throughout their journey – from their first spark of interest to long after their completed transaction(s).
Conversational AI includes chatbots, voice assistants, automated communication (email and text), and the like. It can be used across all channels – mobile, web, Voice, live chat, SMS/text, messaging apps, messengers, etc.
The machine learning part of Conversational AI is a HUGE differentiator from the run-of-the-mill, predefined responses from yesteryear. Machine Learning makes Conversational AI fluid and far more robust than conventional automation tools. Plus, by using and analyzing large numbers of past conversations, it learns at scale and then continually evolves/grows.
3 FREQUENTLY ASKED QUESTIONS ABOUT CONVERSATIONAL AI
What are the benefits of Conversational AI for marketers?
Conversational AI provides the ultimate customer experience at the lowest possible cost when used effectively.
Conversational AI saves money. It reduces staffing costs and training and makes it easier and more efficient to scale, especially after hours and holidays. It also frees up your (human) reps to work on higher-value tasks.
Saves time. Provides fast customer responses (no long wait times!) Conversational AI doesn’t get sick or go on vacation. It’s available – and responsive – 24/7/365, no matter the volume.
Creates continuity in customer responses. Dynamic, personalized communication from one source typically results in increased customer service satisfaction scores and more consistent branding.
Pinpoints where users get stuck. You’ll get a better handle on where, when, why, and how your users are struggling by tracking and analyzing user behavior and the questions they ask throughout the process.
Gives you unfiltered user feedback. Allows you to hear directly from your prospects and customers – at scale – without the inherent bias of things like social media pile-ons.
Increases follow-through and shortens the sales cycle. Let’s face it; salespeople cherry-pick the good stuff, which means the not-as-great stuff falls through the cracks. With Conversational AI, every user gets the follow-up they need/deserve based on your unique hierarchy rules.
Augments your Sales and Customer Service staff. Conversational AI can help with simple tasks like scheduling and follow-up emails. More importantly, you can train your programs to go through ALL the leads/conversations and rank them based on propensity to buy/quote/inquire/act. This makes it easy to route just the ones ready to close and/or those that need special attention from your (human) reps. Plus, it can cement positive brand recognition.
Often increases AOV (Average Order Value.) Perfect for providing users with dynamic recommendations, it’s also great for upselling and cross-selling. Plus, there’s more opportunity for the use of consistent personalization.
What are the drawbacks of Conversational AI?
It’s still a toddler. A young toddler, in fact. Conversational AI’s infancy status means it still has much to learn, especially about accents, dialects, sounds (gestural and background), tone, sarcasm, slang, and such. Plus, there are variations between text and voice that aren’t yet handled well, especially regarding emotions.
The other challenge (besides security and privacy) that nobody likes to discuss is that there’s a massive chunk of users who aren’t all that good at interacting with it. You see this in live usability a lot – people will go to a site expecting to have a specific question answered. Despite repeated attempts, it’s just plain impossible for them to get an answer and/or escalate the process. Granted, they often get frustrated very quickly (and even more easily!), but this is because the “instant help” process is oversold and underdelivered. This will improve as users become more familiar with the process, and the technology improves.
Incidentally, Escalation Programs are essential. Unfortunately, too few companies take the type to develop them properly. Escalation Programs allow users alternative ways to communicate with you, depending on their issues and satisfaction levels. Some companies do this manually (a call center rep watches x number of conversations and steps in when they see the user is repeatedly failing). Others use automated systems/AI to move users from one “level” to the next. The most effective Escalation Programs start manually and evolve to AI solutions.
How do I start to use Conversational AI in my marketing efforts?
When appropriately used, Conversational AI gives you valuable insight into your customers and prospects, the market in general; sentiment; product demand; buying propensity; customer objections, and more. You can use it for everything from identifying key data points to cross-selling, upselling, and tracking your customers purchasing patterns. You can also use it for content building, Search Engine Optimization (SEO), lead generation; ranking based on propensity to buy; email marketing; reputation management and tracking; and more.
The best way to start?
First, figure out two things:
Where and how do your customers want to interact with you?
What questions do they have about your products and services?
For most companies, it’s going to be easiest to start with your inbound emails and/or live chats and any information you may get from your contact center representatives to develop your list of Frequently Asked Questions (FAQs.)
Identify the top questions. List each question and its best possible answer. Remember, the stronger your foundation is, the more solid your AI will be long-term, so please develop good, complete answers/responses for each question. Be sure to work with your SEO/traffic teams on these. It helps.
After you’ve developed your first batch of FAQs, implement them accordingly into whatever conversational design tool you’re using or the system you’ve built.
Yes, many systems have embedded tools that will list your FAQs for you, and it’s a good idea to use them to spit out a list for you; however, learning how to do this on your own will pay off down the road. (With AI/ML, you don’t need to know precisely how the sausage is made, but you should have the gist…. and definitely the list of ingredients!) Incidentally, there are lots of other things you can use Conversational AI for – live chats; search bars; “no cart” order taking; upselling and cross-selling, etc., and it tends to work best if you start with something quick and easy but more importantly, something you can watch evolve and know whether it’s doing it correctly as it does. No matter what your vendor tells you, Conversational AI is not set-it-and-forget-it.
It’s important to remember that although your objective is a humanlike conversation, you need to train the system first. Once trained, you can teach the AI to ask follow-up questions, probe for further information, rank the users based on their propensity to act, and so on.
Currently, the companies that use Conversational AI best are the ones that combine AI with humans (call center reps, for example) to provide an optimal customer experience.
Are you using Conversational AI in your marketing efforts? If so, do you have a tip you’d like to share? Tweet @amyafrica or write firstname.lastname@example.org.
A Down-and-Dirty Definition for Marketers. (Read more about these here.)