Chat bots are booming! You go to any consumer website and you will be greeted by a chat box pop up asking you if they can help you with something, like a virtual embodiment of the salesperson in showrooms who lurks behind you. Every app/website is adopting chat bots, some only because of the peer pressure too. And this has resulted in the flooding of the market with chatbots, a few of which may not even serve the purpose that a company is trying to solve. So, if a chat bot doesn’t work for you, we can’t really put the complete blame on the bot company itself. So, how do we resolve this issue?
Let’s first understand the core purpose of the chatbots: they are serving as a replacement of humans to carry out certain operations, bring in the benefits of accessibility, increasing capacity, cost effectiveness, etc. to the customer service operations. But they can largely be categorized based on their purpose as follows:
- Utility provider: Eg. Setting an alarm
- Information provider
- Customer service/ issue resolver
- Conversation Commerce/ Selling
Let’s first look at the cases where chat bots perform better than the traditional means:
- Utility provider: Chat bots are faster in this content than approaching the app functions directly. However, they can’t really compete with voice assistant here but that’s not the topic today
- Information provider: from the source cited, the retention of information from the chat bot compares similar to that of the information on the app directly. The accessibility increase, however.
- Customer service/Issue resolver: This is where Chat bots have really shined. They result is faster resolutions when you specify the steps clearly and they also bring better customer satisfaction score. If this is your sole purpose, chat bot seems like the best option in the market.
- Advertising: The lead conversation from chat bot ads are seen to be better than your traditional Google AdSense etc. So, they could prove to be a better marketing tool as well if you are able to utilise the chat bot integrations on popular social media websites like Facebook, Telegram etc.
We didn’t mention commerce in the list above because selling products still remains tough for chat bots (transaction conversion rate of 1% compared to the app’s 1.5%) to do despite making the process faster and easier to conduct. Where the chatbots are failing is in engaging the user to finally, make a purchase. The final sales of a product also include a human touch can build more trust in the user. A solution could be to add better character designs to the chatbot but currently, this remains to be seen from a sales perspective.
So, we did see that the chat bots serving a lot of different purposes. But if a chatbot would work for your business still depends on a lot of factors like your purpose, flexibility for your business, the vendor etc. But, if you are still evaluating, you might already be late to the party
Are Chatbots serving their purpose?
By Akshay Thakur
Right from the Turing test’s invention, chatbots have been one of the more visible products of AI & ML development over the years. Almost every B2C app in the market uses some form of chatbot software to deal with customer queries. The spectrum of technological complexity is diverse, starting from the basic multiple-choice question bots used by some food delivery apps to two-way speech-enabled, heavily-trained and nuanced software like Siri and Alexa. Chatbots have been an attractive option for many companies to cut down costs on service operations, customer support, and employee self-service platforms to the tune of almost 30%. But the chatbot industry is also plagued with multiple issues like spamming and cybercrime, dissatisfied customers, and legal disputes. So, it is natural to ask if chatbots genuinely serve their purpose, i.e., supplementing and enriching human interactions. Let’s have a look:
- Improving business efficiency – We all know the painful experience of talking to a human customer care service, suffering from long waiting times, inadequately trained staff and insufficient redressal mechanisms. Chatbots have brought a paradigm shift with lightning-fast response times and satisfactory resolution of queries. A lot of chatbot systems are cheap to develop, maintain and customise. The upside – massive savings on outsourcing human-agent teams, improved customer retention and obtaining big data to use customer analytics to develop strategies.
- Customer experience – As stated above, chatbots can bring down response times dramatically and solve most of the generic queries with ease – a smaller human team serves as a back-up for critical issues. All in all, the customer is pleasingly satisfied. However, implementation has been inconsistent across companies, and many of them see chatbots as a way to offload their responsibility towards their customers altogether. Untrained bots with no feature of leading discourse frustrate the customer even more, and they search for an (absent) human-agent for resolution. Also, in fields like software, where the issues are less generic, a bot will frustrate the customer since they have a hard time explaining to the bot what their exact problem is.
- Legal and social implications – With each day, it’s becoming harder to draw the line of responsibility between an intelligent chatbot and the organisation who runs it. Microsoft had to formally apologise after one of its chatbots “Tay” went rogue in 2016. A developer may include checks and balances so that chatbots don’t end up being politically incorrect – but there is no practical end to it as social norms keep changing constantly. A company deploying a chatbot need to assess the legal ramifications of such things happening and decide.
It is often said that the use or misuse of technology is what projects its image, and not what it is intrinsically. Same goes with chatbots. All the issues that have come up are due to inadequate engineering or poor application; the onus of both lies on human beings. Apart from that, chatbots have genuinely served their purpose and introduced a new dimension to the realm of interactions.
 Maruti Techlabs (2017, April 21). Can Chatbots Help Reduce Customer Service Costs by 30%, Chatbots Magazine. From https://chatbotsmagazine.com/how-with-the-help-of-chatbots-customer-service-costs-could-be-reduced-up-to-30-b9266a369945
 Dave Lee (2016, March 25). Tay: Microsoft issues apology over racist chatbot fiasco. From https://www.bbc.com/news/technology-35902104
Are Chatbots serving their purpose?
Ironically, a lot of people think that they have never used a chatbot, that they have never experienced it, when in fact, they have. Because Siri and Alexa and all those voice- activated speaking bots, those are actually audio chatbots.
Kelly Noble Mirabella Spiderworking
PC20190122 P19dakshesh@iima.ac.in 9057234717
To understand Chatbots’ efficacy in providing customer support (one of their many usages), and how much utility customers ascribe to them, I conducted a survey. The survey received 80+ responses, and its findings form the basis of this article.
Are you able to make out
Are you satisfied with the
Would you rather chat whether you’re chatting
service you receive when
with a ‘Human Being’ than with a ‘Bot’ or a ‘Human
you chat with a ‘Bot’ for
a ‘Bot’ when seeking Being’ when seeking
customer support? customer support via Live
0% 50% 100%
As evident, 89% of respondents feel that Chatbot is not able to closely mimic a human being, 77% are unsatisfied after seeking customer support from a Chatbot, and 88% would rather chat with a human being than a bot.
Understanding how Chatbots work
Simply put, Chatbot is a machine that chats with you, while mimicking human conversation as closely as possible. To do so, it relies on knowledgebase, past interaction history, logics, and machine learning. The following diagram summarises working of a Chatbot.
Human Input Knowledgebase
Rules and Logics
Are Chatbots Serving Their Purpose?
What Chatbots Have Been Getting Right What Chatbots Have Been Getting Wrong
Customer Insights: Chatbot interactions can yield precious customer insights which otherwise require significant resource investment. MHAssistant, a Chatbot, is used by Malaysia Airlines to talk to travellers and understand their preferences. 7
Large scale usage: Chatbots can step in when human resources are limited. UNICEF’s Chatbot ‘U-Report’ interacted with 13,000 Liberian children to find out if teachers were coercing students into sex for better grades. 86% students were found to be sexually exploited. 2
Remote Assistance: Humans cannot be everywhere. For people with dementia who had no companions, Endurance launched a Chatbot which can talk to them about hobbies, movies, music, news, etc. 3
Better Sales: Roof Ai, a chatbot helps real-estate marketers automate interacting with potential leads and does lead assignment via social media.4
Machine Learning Can Go Awry: In 2016, Microsoft’s Twitter Chatbot ‘Tay’ learned expletives from Twitter users and tweeted problematic statements like “Hitler Did Nothing Wrong.” 1
Emotions cannot be learnt: AI can only go so far. Support executives know customer emotions, Chatbots do not. When situations go south, Chatbot with pre- programmed responses might not be able to handle them.
Specific solutions not available: Chatbots are designed to handled first-level queries. If a customer has specific need, which is not present in Chatbots’ knowledgebase, it can cause customer dissatisfaction.
Lack of context: Even with detailed conversational trees, it is impossible to understand the context. Below is Domino’s Pizza Chatbot lost in pre-programmed responses. 5
This article recommends following three steps to improve Chatbot experiences:
- Train for Empathy: Contextual customer sentiment training for Chatbots during production based on
- Acknowledge Limitations: Chatbots shine when handling low-complexity situations. Keep them away from
advanced interactions. 3. Go Hybrid: Chatbots should evaluate if the interaction is going well, and if not, should immediately escalate
to human advisor.
Chatbot market is to grow to USD 9.4 billion by 2024 at 29.7% CAGR6 . As with all technology, Chatbots are currently a mixed bag. With some improvements, they are bound to have diverse usage in customer satisfaction, sales, marketing, and industry research.
Words used: 497
1) Peter Bright – Mar 24, 2016 2:28 pm UTC, and Resolute Ars Praetorian. “Microsoft Terminates Its Tay AI Chatbot after She Turns into a Nazi.” arstechnica.com/information-technology/2016/03/microsoft-terminates-its-tay-ai- chatbot-after-she-turns-into-a-nazi/
2) Shewan, Dan. “10 Of the Most Innovative Chatbots on the Web.” WordStream, 26 Feb. 2020, www.wordstream.com/blog/ws/2017/10/04/chatbots.
3) “A Robot-Companion for Senior People and Patients with Alzheimer’s Disease.” Endurance Robots, endurancerobots.com/azbnmaterial/a-robot-companion-for-senior-people-and-patients-with-alzheimer-s-disease/.
4) “Customer Experience for Real Estate.” Customer Experience for Real Estate | Roof AI, roof.ai/.
5) Cummins, Emily. “The Worst Chatbot Fails: Conversational AI vs. Chatbots.” Netomi, 13 Aug. 2020, www.netomi.com/the-worst-chatbot-fails-and-how-to-avoid-them.
6) “Chatbot Market.” Market Research Firm, www.marketsandmarkets.com/Market-Reports/smart-advisor-market- 72302363.html?
7) SnatchBot. “How to Get Valuable Customer Feedback Online with Chatbots.” SnatchBot, SnatchBot, 12 Sept. 2019, snatchbot.me/amp/blog/114/customer-feedback-with-chatbots.