A service dashboard can provide a clear number while leaving the client and provider with different ideas about what it means. Before using a self-service measure to assess outsourced support, agree which events it counts and which customer outcome the team wants to understand.
The Genesys Cloud release notes dated November 16 describe a Self-served metric in Bot Optimization dashboards. Its stated definition combines contained and abandoned conversations that did not require a human agent and had no errors. Availability depends on the relevant region, licensing and release schedule. This is the vendor’s definition of a measure; it does not establish that every counted conversation achieved the customer’s goal. We have not tested a particular organisation’s dashboard.
Begin with the question behind the number
The following is our original client–provider analysis. The examples are invented and do not describe measured Genesys results or a Myaa customer’s performance.
Imagine a client asking whether customers can obtain delivery information without speaking to an agent. The provider shows a self-service figure. Before interpreting it, both sides need to identify the question the figure actually answers.
One question concerns whether a conversation reached a human agent. Another concerns whether the customer obtained the information they needed. A client may care about both, but one should not silently become a substitute for the other.
Write the intended question in the service review. That gives the team a reason for collecting the measure and a boundary for interpreting it. It also makes it easier to notice when a useful operational count is being presented as evidence for a broader conclusion.
Keep different customer journeys visible
Consider two hypothetical conversations. In one, the customer receives the expected delivery information and leaves. In the other, the customer leaves before the team can tell whether the information was obtained. Neither example involves a human agent, but the information available about the customer’s outcome differs.
These invented cases do not establish how any particular system would classify them. They illustrate why a client and provider may need more detail than a combined label supplies. Ask the reporting owner what can actually be examined and which distinctions the available records support.
If the team cannot determine why a customer left, preserve that uncertainty. A missing account of the outcome should not automatically become either a successful resolution or a failed interaction. A useful review explains what is known and what remains unobserved.
Agree how to investigate the outcome
The provider could propose examining suitable examples from the work, while the client explains what would count as obtaining the requested information. The approach needs to fit the service, the available records and the decision being considered.
In the delivery example, a review might ask whether the information supplied addressed the customer’s actual request. A general explanation of delivery times and an answer about a particular order would not necessarily serve the same purpose. That is an operational question for the review, rather than a result claimed here.
Be clear about the limits of examples selected for discussion. A few interesting conversations can help the team formulate a better question. They do not, by themselves, establish how often each outcome occurs across the whole service.
Attach the interpretation to the decision
If the client is considering changing staffing or expanding automation, explain which evidence supports that decision. The dashboard measure may describe part of the operation while other information is needed about the customer journey, exceptions or work that follows.
Keep the metric’s definition with the reporting period and note any relevant change in how the service is measured. The people comparing reports should be able to tell whether they are examining the same quantity. Ask the reporting owner to check the actual system before making claims about its calculations.
A useful service review can hold two ideas together: an operational measure can be valuable, and its label has a specific scope. Agreement about that scope helps the client and provider discuss the work honestly and choose what additional evidence they need before drawing a conclusion about customer success.
AI-generated conceptual photograph of a fictional service review; the blank cards do not represent measured customer outcomes.