The chatbot hype is still growing and, every day, new assistants are unveiled on the world’s leading messenger platforms. At my company, however, instead of wasting resources on a building a complicated “smart” chatbot, we developed a simple, rule-based one. As we can now see from practice, it was the right decision.
1. AI chatbots suck at sales
Despite the development of complex AI bots like Luka, there is still no success story to show that AI and machine learning can sell. On the one hand, there is a hunger for bots that sell, especially for ecommerce companies. Imagine if a client could message the app and the bot would choose the perfect items. Satisfaction and sales would flourish. That’s a picture-perfect dream, but life is more complicated.
That’s why a recommendation system is very valuable. Lots of indie projects and large companies like Amazon are developing their own engines to advise users on what to buy, hoping to help boost conversions and time spent on a website. It’s worth mentioning that these are important metrics for media projects, too. However, such systems are very familiar with their users thanks to purchases, searches, views, and the personal information available on social media.
Companies like Amazon and Microsoft have the vast resources to develop smart AI for their products.
Chatbots don’t have this information — they rely on what the user says. The reality is that modern technology is not advanced enough for a robot to understand people and convert potential interest into a sale. Otherwise, big players like Amazon or Facebook would already be doing so.
2. You can improve client support with simple answers
Client support is one of the most popular uses for chatbots, and it can work really well. Luckily, to get good results, you don’t have to invest a fortune.
Here’s a simple example: When we were developing our marketing agency, RockinRobin, we noticed that clients asked many of the same questions. So we had to choose to either hire more support staff or ease the workload. We chose the latter and developed a Telegram chatbot that is able to answer those popular questions.
Our bot doesn’t have fancy-smart functions yet. It shows questions that it can answer straight away and provides links to helpful articles. Essentially, it’s a knowledge database of information that you can find in a FAQ, but with the convenience of having it delivered to your messenger inbox. At least, that’s what our statistics show.
Within a few months of launching our chatbot, the number of support inquiries decreased by 20 percent. If the chatbot doesn’t know the answer, the user can request a consultation with a flesh-and-bone marketing expert. Thus, we also increased the number of leads.
3. Developing a smart bot will break the bank
Developing Luka cost millions of dollars in investments, which is neither an easy nor small feat. Realistically, were you to attempt a smart chatbot on your own dime, you’d have little chance of success.
Quest.ai, a smart chatbot for online stores, is a textbook example. Its developers wanted to create a tool that would raise conversion rates for ecommerce projects. But it turned out to be more complicated than originally thought, and they failed to attract investment. As a result, the project was frozen.
Now, don’t think that it was run by rookies. One of the creators is the founder of VC.ru, Russia’s top startup and a major technology resource. He was involved in, for instance, the success story MSQRD. It’s also important to remember that it can happen to you too! It’s extremely unlikely that you’ll be buried under a pile of offers from investors and incubators. Rather, you’ll probably run out of funds and have to shut down your smart chatbot project.
4. Simple chatbots can attract attention to your company
Chatbots are “in,” and that’s why many companies develop them to both solve problems and draw attention. We’ve met startup founders who are sure that you can only attract interest if the project is “cool” in tech terms.
But that’s not true. A project doesn’t have to be the most innovative if it carries real value. These qualities are also of interest to users and the media. This is what happened to our chatbot. Yes, it’s simple, but it really improves the efficiency of our business. By sharing our experience, we made it to VentureBeat, TechInAsia, and other industry media. I also became a columnist for Chatbots Magazine.
Realistically, the budget for our simple bot was a hundred times cheaper than that of an AI bot based on machine learning and big data.
The conclusion is simple. If you are thinking about creating a chatbot, decide whether you need sophisticated AI or just a simple, well-built chatbot based on rules.
The audio problem: Learn how new cloud-based API solutions are solving imperfect, frustrating audio in video conferences. Access here