We have no idea whether our users are actually using the "Did you mean" results we're serving them. Without quantitative data on this, it's hard to know whether any changes we make will actually serve our users better.
Let's figure out some way of measuring whether users are clicking on the did you mean results we serve them, so that we can perform improvements quickly and measure our success easily.
What I'm looking for to be recorded:
- How many results the user was shown for their query
- Whether the user was shown a did you mean result
- Whether the user clicked on the did you mean result
Statistics that should be computable from this:
- Percentage of queries where the user was shown a "Did you mean" result
- Percentage of queries where, having been shown a "Did you mean" result, the user clicked on that result
- Stakeholder: Search Team, Discovery Department
- Benefit: We'll know how well any changes we make to "Did you mean" results actually work, as well as knowing current effectiveness
- Estimate: To be added