Track two things rather than one. The overall rate matters for cash planning, because returns consume the revenue you already counted. But the rate by product is where the useful information is, because returns are almost never evenly distributed. One or two items typically account for a disproportionate share, and identifying them is usually more valuable than any general improvement effort.
Then find out why, which requires asking. A return reason field with a small number of specific options, rather than a free text box, will separate the categories that need entirely different responses. Wrong size means your sizing information is inadequate. Not as described means your photography or copy is overselling. Damaged means packaging or a carrier problem. Changed mind is the only one that is genuinely a cost of doing business.
The counterintuitive part is that reducing returns to zero is the wrong goal. A very low rate in a category where returns are normal often means your description is so cautious that people who would have bought are not buying. The returns are the price of a description confident enough to convert. What you are optimising is not the rate itself but the combination of conversion and returns together.
Where it becomes urgent is when returns concentrate in one product with a rate far above your others. That is a product problem rather than a policy problem, and the cheapest fix is usually better photography, more specific measurements, or discontinuing the item.
Cost a return properly rather than treating it as a reversed sale, because the arithmetic determines what is worth preventing. You lose the original shipping, pay for the return, spend time processing it, and frequently cannot resell the item at full price. That total is often a substantial multiple of the margin on the sale, which means a modest reduction in returns can be worth more than a comparable increase in orders.
Write the returns policy for the customer rather than for yourself. A clear, generous, easily found policy increases conversion measurably, because uncertainty about what happens if something is wrong is one of the larger obstacles to a first purchase from an unfamiliar business. Restrictive policies buried in a footer reduce returns and reduce sales by more, which is the trade most first year businesses get backwards.