I would have the pleasure of taking part in the morning “Rencontre Technos” organized by the Lyon Association CPRC (Club of Professionals in Customer Relations) on June 27th in Lyon.
I’m going to answer a few questions about the impact of bots (callbots, chatbots, mailbots, etc.) on customer relationships.
Here is a summary of those questions.
In your opinion, what are the main advantages of chatbots, callbots and virtual assistants in customer service management?
- They come on top of the bot. Like a “new colleague” who will handle some of the low value-added “push button” requests.
- The chatbot was quite far from the CRCs as it was a web usage that had little impact on agent activity. With the Callbot we are really at the heart of the business.
- Now, thanks to its transcription and analysis capabilities, we can also speak of an advanced agent with call reports, automatic and personalized response to emails, notifications, etc.
In which specific use cases do you think chatbots and virtual assistants can be particularly effective in customer service?
- There are scenarios in which the use of a callbot or chatbot is obvious, namely when there are seasonal call peaks. In this case, the bot can handle some of the low-value requests.
- The method of setting up a bot starts with analyzing the types of requests to the CRC at the right granularity (e.g. passwords <=> Windows passwords).
- Then determine which ones can be automated (generally these are requests that are currently easy to process).
- A minimum of 50 conversations being processed in self-care by a chatbot and 100 automated calls for a callbot are required.
- This is followed by an analysis of the technical feasibility
- This is usually done in advance of the project to demonstrate an ROI
How have you integrated chatbots and virtual assistants into your overall customer service strategy?
- It is an activation option, but more and more it is a native request, especially via the first stage via a natural language SVI that proposes an FAQ.
What were the key steps in this integration?
- After analyzing the ROI, it’s really time to get down to business operations. 1 to 3 months for a chatbot and 3 to 6 months for a callbot.
- For example, for Toy King, we set up the bot in 2 months. It’s an FAQ bot but with 80+ intents.
- There is always a phase of modeling, continuous improvement of conversation design, customer service testing, controlled rollout, optimization and follow-up
How do you measure the effectiveness of chatbots and virtual assistants in your customer service?
- Callbots: To know the number of calls avoided, a very simple solution is to check the number of calls before and after a few weeks. For example, for a client in the energy sector, we implemented SMS for high call volume intentions and over night over 3% of calls were avoided on top of the calls already avoided. through the voicebot.
- Above all, however, it should be analyzed whether people call back after they have spoken to the bot. Depending on the application, this can be 30, 40, 50 or 60%.
- Satisfaction measurement is more complex to measure because the customer cannot be systematically asked after each response to a query whether they are satisfied with the response.
- With chatbots, this is done by measuring post-inquiry satisfaction when a customer clicks “satisfied.”
What performance indicators do you use?
- For a client in the energy sector, we have a dashboard with almost 20 indicators, because he wants a close monitoring of requests, performance of services, etc.
- But in general it’s the number of calls handled, rate of self care/callbacks, satisfaction with whether or not calls are forwarded…
How do you ensure a smooth transition between chatbot/virtual assistant interactions and human agent interactions?
- It’s important not to frustrate the customer. That’s two calls at most before you contact an agent.
- Then you need to tell the agent the whole conversation and the reason for the call…
- After all, it has to be fast and determine the urgent/non-urgent reasons.
What challenges have you encountered when integrating chatbots/callbots and virtual assistants into your customer service? How did you overcome them?
- With chatbots and callbots, it is the completeness of the response. The answers are too often general and not specific enough… And so the customer is frustrated.
- For callbots, the voice output needs to be optimized to recognize technical terms, product names, numbers, etc. There is real optimization work to be done.
How do you rate the personalization and adaptability of chatbots and virtual assistants to provide a quality customer experience?
- It is important to analyze the bot’s responses and follow up in the first few months to spot the responses that are not satisfactory and not clear…
- Next comes intent recognition, which is a key task in analyzing NLP recognition.
What are the top things to consider when choosing a chatbot/virtual assistant platform or vendor for your business?
- Integration with the contact center solution as the bot will never process 100% of the requests and therefore the transition needs to be smooth.
- Support because you can easily set up a chatbot with a low-cost solution, but it inevitably reacts sideways and leads to customer dissatisfaction, which ends up being a false economy as you have disgruntled customers and agents doing this with even more dissatisfied customers too have fight
What are your predictions for the future of chatbots and virtual assistants in customer service?
- ChatGPT is a real game changer and allows you to take a big step forward in automating customer service actions.
- Both for the advanced agent (personalized auto-responses, call summary, etc.) and for a better understanding of requests.
- On the other hand, the technologies are still immature, especially at GDPR level, but Microsoft Azure Open AI is in place, as are alternative solutions using GPT AI. So we need to show interest in it as soon as possible, because it will bring a real benefit to the customer services implementing these “second generation” AIs.
What developments or trends do you expect?
- ChatGPT allows for more efficient parsing of fonts and hence future advances in mailbots. For example, anything that is a response template can be overridden.
- The democratization of STT (speech recognition) will allow CRMs, contact center solutions, etc. to be enriched with the customer’s voice.
