Tool
The ground you have not mapped
Each mark is a concrete way this breaks in production rather than in the demo. Move how many agents your team has shipped and the light moves with it. The ones left in the dark are listed below by name.
15 of 18 are still unexplored for you. They are not categories, they are cases with names. These are the ones on the dark side at your number:
- they send a two minute voice note instead of writing
- they send a photo of the receipt instead of the order number
- they answer yes to a question that offered two options
- the regional holiday that is in no calendar the system knows about
- the same customer writes from two different numbers
- the payment came in twice and the system shows one
- the question arrives in two languages mixed into the same sentence
- they forward an old message from the agent and start the conversation from there
- and 7 more.
The model, in the open
The catalogue comes from our own implementations, not from a study. The number behind each case is the implementation by which we had already seen it at least once, which is an order we observed rather than a measured probability. That is why it can be anticipated, and why it is declared here.
How to read it
- Set the control to zero. That is where a team that has never put an agent in production starts, and the whole terrain goes dark.
- Read the cases, not the counts. Each one is a conversation that went badly before anybody knew that case existed.
- The first three belong to the first implementation and read as obvious here. None of them was in the scope document of that first implementation.
This object is the interactive half of an article: Why having your own IT team doesn't replace experience with agents