The principle of slicing or dimensioning a metric is to take a metric value (for example, Sales Actual) and break it down by a dimension object, for example by Promotion, by Country, by Category. At the SQL level this means employing a GROUP BY clause. To define what to "slice" on the universe needs modifying to include dimension objects. For example, you have a dimension called Promotion. Information about the promotion name is stored in the promotion table so this is loaded and joined to the fact on promotion_type. Standard dimension objects are then created for promotion_id and promo_description against the dimension table.
When refreshed the metric can be viewed as normal in the interactive analytic. The user now gets a link under the legend that allows the value shown in the chart to be sliced. When selected the Sales Actual value is split into the distinct available values for promotion.