At the moment the facets are static in the sense that the page state doesn't affect the choices displayed.
An alternative would be to have the facet control show only those options that would continue to yield observations given the current settings on other facets.
In general this should prevent you from reaching no results.
This can also provide a signal as to the restrictiveness of your choices - i.e. how the criteria interact - and so help you to prioritise. This is only really possible when the range of choices is already small so you can see what's going on. Given the broad range of e.g. product codelists, it would be easy to miss that a previous selection had reduced the range. You mightn't realise that you could expect more possibilities by resetting that selection first (and re-ordering the process). We might need to have harmonised codes #81 before we attempt this.
An alternate approach might be to provide this signal with a suggestion on the code-choices as to how many observations would be available (in total) - seeing zero would be akin to that option having been hidden.
At the moment the facets are static in the sense that the page state doesn't affect the choices displayed.
An alternative would be to have the facet control show only those options that would continue to yield observations given the current settings on other facets.
In general this should prevent you from reaching no results.
This can also provide a signal as to the restrictiveness of your choices - i.e. how the criteria interact - and so help you to prioritise. This is only really possible when the range of choices is already small so you can see what's going on. Given the broad range of e.g. product codelists, it would be easy to miss that a previous selection had reduced the range. You mightn't realise that you could expect more possibilities by resetting that selection first (and re-ordering the process). We might need to have harmonised codes #81 before we attempt this.
An alternate approach might be to provide this signal with a suggestion on the code-choices as to how many observations would be available (in total) - seeing zero would be akin to that option having been hidden.