This user's questions cut straight to the heart of what separates a functional spreadsheet from a truly empowered data model. They are not asking about a quick fix. They are asking about best practice, and that instinct is exactly right. The answer to both problems is the same: build the logic into your DAX measures, not the Pivot Table filters.
For the year-filtering issue, the temptation to use a manual Pivot Table filter is understandable. It is fast, it is visual, and it works for a single report. But it creates a silent dependency. Anyone who opens your file afterward has to know that a filter is sitting there, invisible, controlling the output. One accidental drag, one refresh, and the constraint disappears. A DAX measure using `CALCULATE` and `FILTER` around your `Date` table is explicit, self-documenting, and survives any change to the Pivot Table layout. Write one measure called `Sales 2020-2021` that wraps your base calculation with `CALCULATE( [Total Sales], FILTER( 'Date', 'Date'[Year] IN {2020, 2021} ) )`. That single measure becomes the source of truth. Every chart, every slicer, every subsequent measure inherits that constraint without you having to remember to set it.
The cross-table calculation follows the same principle. The user is right to feel stuck, multiplication across related tables is a common hurdle because DAX does not work the way Excel cells do. In Excel, you fetch a price cell and multiply. In DAX, you let the relationship do the heavy lifting. The correct measure for Total Sales is `SUMX( 'Sales', 'Sales'[Units Sold] * RELATED( 'Products'[Unit Price] ) )`. `SUMX` iterates over each row in the Sales table, and `RELATED` pulls the matching price from the Products table through the existing `ProductID` relationship. No `LOOKUP`, no `INDEX`, no manual mapping. The relationship IS the bridge.
The deeper lesson here is about ownership. A Pivot Table filter is temporary. A DAX measure is permanent logic that travels with your model. Every time you choose a measure over a manual filter, you are reducing the hidden complexity that will trip up your future self, or the colleague who inherits your file. Start with these two measures. They are not advanced. They are foundational. And once you write them, you will stop fighting your data and start directing it.