How we implement chatbots: analysis/design
- We always start with the business problem. We bring in one of our specialists early on to understand the customer journey and the type of conversation that the chatbot needs to go through. This process can reveal complexities in the requirement that the client isn’t aware of – for example, the need to access multiple systems or services.
- It’s important to understand the requirement from the perspective not just of the client but also of the client’s customer or employee who will be using the chatbot. We need to really understand what services that user needs.
- We then make sure we understand the integration requirement: what relevant solutions are already available to meet the needs identified, and whether any additional ones need to be built to achieve what the client wants.
- If additional solutions are required, we can often build them quickly. For example, with one client we were able to have the new chatbot access an existing spreadsheet via a Google API.
- We engage as early as possible with the in-house IT department. That way, we find out whether there are any constraints. For example, there may be security obstacles to making an inhouse database or spreadsheet available externally via a bot. This type of issue can usually be resolved but it should be tackled early on.
When we go to meet with a client we know that any chatbot solution needs to be tailored to meet the needs of the business. It’s vital to first understand the requirement after which we can propose a bespoke solution that can effectively address any pain points discussed previously.