If texting takes more time than clicking on a webview, why is it better?” tweets Connie Chan (Partner at Andreessen Horowitz)
You might be wondering why a partner at one of the best known VC firms is tweet-shaming (an arsenal of tweets to be precise) “Chat”. Well, to understand this we’ve to understand the journey that the “Chat” has taken as it ventured into the business communication world. So, lets get on it…
Once upon a time there was a myth that “Chat” is the future of business communication and “Chatbots” were at the helm of it. The popular opinion came from the assumed progression of customer support where companies thought that Live Chat will translate to Chatbots. We can’t blame the stakeholders who believed in it as the buzz around it was unfathomable in early 2015. From tech journals (like this TechCrunch article) to renowned futurists (article by Benedict Evans) were victims of the “Chatbot” mania that was over-hyped even by the standards of hype-cycle.
The hasty progression towards Chatbots happened because everyone realised that Live Chat itself was not enough. The limitations were more obvious than you’d think:
1. Troublesome Timezones:
You would have heard a lot about making your customer support 24/7. The reason for that is simple. Your customers don’t sleep or wake up the same as you. If your business is spread across timezones, it is an operational nightmare to treat customers from all geographies consistently.
2. Scalability Shenanigans:
You no longer need Paul Graham to tell you that “YOU NEED TO GROW FAST”. As you scale your company, your website traffic is bound to increase and so will the opportunity to acquire more customers. But you will find it difficult employing more and more customer executives to chat with your customers. It reduces making business sense. And it comes wrapped with high training costs and high attrition rate. Using Live chat to scale is nothing but a cash burning scheme.
Alright! We get it! setting up an efficient Live Chat is a hassle. But what’s wrong with chatbots. They definitely seem to be one of the greatest invention of this century. Why are businesses moving away from them as well?
To put this into context, lets hear the tale of a startup in India that went through the entire hype cycle of using conversational UI for businesses.
HELPCHAT -> TAPZO:
“We once passionately believed that chat is the new universal UI”
Back in 2015, businesses were really stoked about adopting chat as it was being widely used in the P2P environment. Helpchat ventured to build a Chatbot to enable communication between consumers and businesses. They took the “Live chat” route in order to train their future bot. Consequently, they hired an army of customer chat representatives to build their database. By June 2015, they were one of the biggest “Consumer to Business” chat apps in the world doing more than 70,000 chat sessions per day.
“We thought the users would get trained to use chat”
The customers loved the chat experience as well as the idea of a human being helping them with their queries. But the company soon faced a renowned roadblock in the B2C industry: “CUSTOMERS WERE NOT COMING BACK”. They tried everything possible from optimising first response & average response times to building better dashboards for monitoring chats. They even made the final progression towards the much anticipated “Chatbots”. But the problem seemed to worsen after that. Eventually, the NLP (Natural Language Processing) based chatbot irritated users to no end. The reasons of failure were why most (if not all) Chatbots fail:
1. Holy grail of Messiness:
Misspellings, SMS Lingo, abbreviations, autocomplete errors, no regular sentences…the list goes on. If you’re asking the customer to type her problem, she’ll do it her way (yea…”her life her rules”). Even after building a huge data set, the Chatbot was not able to decode these complex message requests from customers. Expecting the bot to respond with a legitimate answer to their queries would be day-dreaming (if not a fairy tale story).
2. I speak a different tongue:
Once vernacular languages come into play, it becomes extremely difficult to train the bot. The complexity is incomprehensible given that the training data set will be different for each language both qualitatively and quantitatively (depending upon your target geographies). It was not a piece of cake that can be solved by plugging in a library.
3. Do you understand, though?
In case of NLP, one size doesn’t fit all. Everyone has a different perception and they use different words/phrases/sentences to convey their queries. Interpreting them in a way desired by the user and fetching relevant results from a database still remains a daunting task for any well-trained Chatbot.
“It became obvious that the problem was chat”
Tapzo realised that it was trying to force fit “Chat” in the consumer to business communication space. The customers tried it out of curiosity but never returned. The problem was quite fundamental and is usually referred as: “People on the internet will take the path of least resistance”. They finally pivoted to Tapzo and had to fire their entire chat team. Thereafter, they built the product from ground up keeping the user as a priority and came up with the tap-friendly interface. This has resulted in immense growth of Tapzo in terms of partnerships, users and revenues. Operation revenue went 60% up in Financial year 2016-17.
This journey is very well articulated by the founder and CEO of Tapzo, Ankur Singhla, in its official press release.
Please note: Amazon Pay has acquired Tapzo (for $30-40 million) to push their digital payments business
The story culminates into the Kingdom of “Chatbots” being demolished and concept of Chat dying a slow death in the world of business communication. User data has repeatedly shown that Conversational UI or chat hasn’t been able to provide a delightful experience when users meet businesses online.
Airim’s analysis further justifies this as only 7% of our users prefer Chat interactions. It is also interesting to note that 76% users prefer to click pre-suggested options based on intent. But what does it imply?
Is guessing User Intent going to be the next big thing in business communication?
Is it possible to engage visitors with a tap-friendly interface where all they have to do is click rather than typing their queries?
And could AI actually be used to optimise this experience?
We have our own views on this and will keep you posted about them. Till then, do give us a flavour of how has your experience been while having online conversations with businesses.