How do I choose between Search and Chatbot?

April 2, 2024

Undeniably, the chatbot is trendy. Since its appearance on the web, many companies have wanted to set one up. According to the very serious British weekly The Economist, 30% have taken the plunge (1) and have already implemented chat in their company in 2018. 

Yes, the chatbot is hype. A small must for all budgets, which can respond to your customers anytime, very quickly. For a clientele in need of personalization, speed - the 21st century customer doesn't want to wait - and who sometimes even prefers to interact with a robot rather than a human as a Facebook study (2) shows, the chatbot seems to be a real must have.

Chatbot: What are we talking about, and how does it work?

The chatbot belongs to the category of "conversational agents". To list the main types, we will remember that chatbot converse with the Internet user in writing, while the voicebot speaks to the user in a generic voice, and that the callbot, finally, is a voicebot that can converse with a person over the phone. The last two are considered more "innovative", according to the Occurrence Institute and the publisher DYDU (3).

The chatbot is now used by 31% of companies, so it is legitimate, at such a level of adoption, to consider it as less innovative. The ordinary chatbot can answer simple questions, and may not look smart enough. At this point, experts might say, "No problem, let's feed them AI!"

Artificial intelligence is a bit of a magic word if you want to titillate the interest of the media, investors, customers. In the last few years, unscrupulous companies have been putting the word "AI" on their chatbot offer, even though they don't actually use AI in their bot (but rather a set of predefined rules) have multiplied.

Of course, there are also companies that play the game. AI can be used for a chatbot, but this process is time consuming, expensive, and requires a lot of data - so much so that only large companies can provide it, which rarely makes it an implementable solution for a small SME.

What's next?

Contrary to appearances, and even though we've just touched on the subject of chatbot coupled with AI, there's still a long way to go to really personalize the user experience.

In 5 years, conversational agents will have become the norm and will offer more advanced functionalities, according to Enterprise Talk (4), for whom chatbots will then have "emotions": they will speak and can incorporate a tone, a personality. We take the liberty of adding quotation marks to "emotions" because chatbots will mimic human emotions. Ethical questions on this subject are numerous and still pending.

According to Stephen Burke (5), executive vice-president at Sciant, a computer consulting firm and software publisher, chatbot is just another user interface. In other words, it is simply a new format for interaction and information delivery to the user.

The problem, precisely in terms of the user interface, is that the chatbot can disappoint, if it is not understood by the user . Often, he doesn't know how to talk to him. The chatbot is expected to be able to go further. He should be able to interact with humans as humans would, and be able to analyze the context in which the user finds himself. For example in a purchasing process, the chatbot will not necessarily direct us to the best product. It won't ask the user why he wants that product and won't be able to offer him a more adequate alternative, as a human salesman who is an expert in his products could do. The chatbot should be able to sell alone, without the intervention of a human, but this is still quite far from happening.

The chatbot will try to propose THE right answer, in a context of dialogue, unlike a search engine which will suggest several answers. It is up to the user to carefully choose the words used when querying with a chatbot in order to maximize his chances of finding the right answer. Otherwise, he risks being misdirected and not being able to go back, because the conversation is generally managed as a conversion funnel with validations at each step. The chatbot, as a conversational agent, tries to mimic a human dialogue, but in reality the technologies used to do so are far from infallible.

The chatbot thus allows a fluid and direct interaction but limited to a perimeter. This perimeter being the knowledge base programmed at the beginning, and generally not very evolutive.

Chatbot VS Search

At Outmind, it is for these reasons that we have chosen to use a search engine instead of a bot. We also asked a lot of questions to our users and prospects on the subject. The results show that chatbots are still very imperfect, and that users don't see them as a real alternative to search engines, even if specific cases exist. When both possibilities are suggested, the most common answer is: "I will still go to Google or my internal search tool first, and enter my search query with the first words I have in mind, rather than opening a chat application and choosing my formulations and hoping that the chatbot will understand them". This type of feedback suggests that although users like the principle of chatbot interaction, it doesn't really meet their need for immediacy and easy access to information .

Based on this observation, the search engine seemed to us to be the most adequate to carry out rigorous and complete queries. Indeed, it allows to display an exhaustive list of results and to classify them according to their relevance. The AI allows, among other things, to improve the relevance ranking as it is used, and the user can use various parameters to refine his search as he wishes. 

A bot, on the other hand, would have tried to imitate the discussion with a human by proposing only one or very few results.

The results offered by a search engine will analyze all the information and will be updated in real time, where a bot is only limited to what is initially scheduled, it will usually need to be updated.

Using the latest indexing and machine learning technologies, we have created an intelligent search engine, with a carefully designed user experience, to facilitate your searches and better capitalize on existing knowledge. If you wish to test Outmind or discover it during a demonstration, you can directly request a free access.

References :

(1) The Economist,

(2) Facebook,

(3) Occurence Institute and DYDU,

(4) Enterprise Talk,

(5) Stephen Burke,