Gartner study: developments and best practices in chatbot projects
DISCLAIMER: this article is largely a transcription in French of an article published on the blog of Gartner, an American consulting firm that publishes the famous “Gartner magic quadrant” every year, which compares software solutions in many sectors. The article in question is “Chatbots Will Appeal to Modern Workers”.
According to Gartner, chatbots continue to be a hot topic and vendors are starting to mature, primarily due to improved natural language capabilities based on artificial intelligence.
In the 2019 Gartner CIO Survey, CIOs surveyed see chatbots as the top application of artificial intelligence in their companies. The growing interest in this solution is driven by the needs of customer support, knowledge management and internal user support.
So far, nothing very surprising from our point of view 😉
Why is the use of chatbots becoming the norm?
The power of the chatbot tools available on the market today allow to integrate in a fluid and intuitive interface actions that are more laborious in a traditional way, as illustrated in the following graph from the Gartner study:
The study anticipates that by 2022, “70% of ‘white collar’ workers will interact with conversational platforms on a daily basis.”
One of the reasons for this growth would be the mechanical increase in the share of millennials among employees .
Being late 2022 as I write this, we have been looking to see if this prediction is proving valid or not. We did not find this exact statistic on Gartner’s website, however a study dated September 2022 indicates that 54% of organizations surveyed use at least one form of chatbot.
In addition, 79% of CIOs anticipate an increase in investment in so-called “self-service” systems (such as, typically, chatbots…) by 2024.
The adoption of chatbots within companies is already well underway and this trend should consolidate rapidly in the coming years to become the norm.
What are the best practices for a chatbot project?
According to Gartner, the market is occupied by nearly 2,000 chatbot vendors, most of which are not equipped to properly deliver or maintain professional chatbots. The company therefore recommends the following practices to deploy a sound conversational platform strategy.
1 - Avoid chatbot solutions that do not provide sufficient performance
It seems obvious, but not all chatbot editors match the requirement level of a medium or large company. Many editors are aimed at webmasters, small structures or are dedicated to a very specific use case such as basic customer support on a website.
If you want to develop an internal chatbot, there is no secret: in general, you need to allocate a minimum budget to work with a serious editor.
2 - Reduce the risk of project failure by using a specialized supplier
Gartner specifies that you should only create your own chatbot if the company already has specific machine learning and data science skills in-house (yes, this is quite rare…). In all other cases, it is better to go through a specialized actor.
We agree 100% with this statement. You’d rather pay 200k in development costs for a disappointing home-made chatbot (since, sorry to say it like this, but this is not your job… nor that of general consulting agencies) or pay a few thousand euros for the license of a solution whose core business is to make a POC and see if it is worth going further?
3 - Secure the resources needed to maintain the chatbot
Most chatbots require continuous improvement: this ensures that the topics mastered by the bot always remain aligned with user needs.
Indeed, a chatbot is a living structure that evolves with its audience. Another aspect I would like to add here to underline the importance of appointing a chatbot manager: as we often repeat (see for example this excellent article by Sacha Lemonnier, part 2), the best approach to create a chatbot is to start with knowledge bricks or simple actions that bring in themselves a strong added value (think of it as the Pareto principle applied to chatbot project management). Afterwards, different bricks can be added step by step.
4/ Setting the stage for Voice Bot adoption
Although there are still few chatbots using voice, Gartner recommends preparing for this eventuality, noting that the demand for this type of bot is growing.
On this recommendation, we are not aligned. However, Vizir is quite capable of creating a “voice bot” (our CEO does the demonstration in this video). The back-end of the bot will be the same, the only difference will be the means of communication used to interact with it.
Nevertheless, it seems to us that it is still too early to deploy voice bots on a massive scale in an efficient wayThis is because voice recognition technologies still lack fluidity in our eyes (difficulty with proper nouns, with accents, etc.).
5/ Think about the tone, emotions, personality of the bot and other " soft features
These aspects are important for the success of the chatbot project by conditioning the level of user adherence and improving the level of error acceptance. The way the chatbot communicates should reflect your company’s values.
This time, I totally agree! This is something we often stress: beyond the technical aspects that often crystallize most of the attention of your teams, the “communication” aspects are essential for the acceptance of the chatbot as a new channel. We systematically recommend working on the image of the chatbot by giving it an identity (a “little name”, a custom-made visual, etc.) and by homogenizing the tone used by the bot.
This good practice has been particularly well followed by many Vizir customers, such as the Adapei association, which has given a friendly personality to its bot “Rosie” (you can see the Adapei case study here) or by the group Bolton with “Tom the Bot” (Bolton case study to be found here).