What makes a good recommendation?
Recommendations are what we think we should do with the knowledge.
The practical result of what came before.
A recommendation looks complicated with 5 different parts, but it's really quite simple and not all parts are always necessary.
So we have…
- Move the invoice history to the profile settings [Action]
- from the footer [Context]
- so that customers [Audience]
- can find them better [Benefit],
- which will reduce calls to the contact centre [Measurement].
We want these to be short enough to be consumed in a few seconds and contain enough context to be understood in isolation.
Recommendation taxonomy:
As well as context we want to give status and prioritisation.
Some example statuses include:
- Unactioned if it's waiting for a decision to be made.
- If they don't have confidence to make a decisions they might require More evidence.
- With that evidence they might to Reject the recommendation.
- Or they might Approve it.
- This will send it to the blacklog to be Completed.
- And once it is it will need Retesting.
- A recommendation might also be an Hypothesis that we are looking for evidence to support or disprove.