Just starting out and want to boost your yields 10-20%? Triple rinse.
As you set up, work up, and purify your reactions/products rinse every surface that one your compounds touches times with a compatible solvent, with the rinses transferred to the next flask. Compounds dissolved in solvent are especially prone to leaving a thin residue behind on most surfaces, steadily lowering yields as the product moves from one container to the next. It’s hard to believe that these residues can add up so quickly, so to provide a bit of evidence I ran a test case with KMnO4 and water in a 4 mL vial, showing just how much KMnO4 stays behind in the flask .
Three is the optimal number of washes for well behaved compounds, and because of basic statistics you’ll see the number three crops up in quite a few unrelated techniques (liquid/liquid extractions, solvent degassing, and contaminant removal via co-distillation, etc). With good solubility in the wash solvent each wash transfers a bit more than 90% of the remaining residue to the next flask . After two rinses there is then less than 1% of the material left on the original glass/paper/surface, and after three less than 0.01%, a generally negligible amount.
Getting into the habit of washing everything from weigh boats to test tubes takes a little practice. But you aren’t going to find a simpler way to improve the the yield of almost any reaction.
 For the sticklers: A spatula tip quantity of KMnO4 was added to the vial, then dissolved in ~2 mL ddH2O. The water was drawn out with pipette and a fresh aliquot added. The vial was then shaken vigourously and the rinse withdrawn with the same pipette. The vial was brand new, with the water sheeting off the walls. As a result most of the residual KMnO4 is from small droplets left on the upper portion of the vial and the cap.
 This number will vary wildly depending on the size of the surface being washed, the speed at which the compound dissolves in the solvent, and the contact time between compound and solvent. Anecdotal evidence suggests that 90%+ efficiency isn’t unreasonable in most cases, and since the math works in excess of 80% I’m ok with a bit of hand-waving.