Returns aware decision making in e-commerce

If you’re familiar with the world of digital commerce and returns, it’s relatively easy to visualise the direct impacts and costs of returns across logistics and customer experience.

But what’s often missed or misunderstood is how returns impact the operations and decision making within a merchandising function, and the ramifications this can have for both inventory and profitability.

Parallel causes?

The reasons for this build on what I wrote previously about the impact of returns on product discovery and website algorithms, as I think at least a couple of the causes are shared, in particular the use of gross metrics and the effect of the returns lag.

However, when it comes to the decision making within merchandising I think a couple more factors can exacerbate the problem:

• Misinterpreting “returns maths”
• Not factoring returns costs into product margins?

Let’s take each one in turn…

Gross Metrics

The use of gross values when monitoring product performance through metrics like rate of sale or margin can have a significant impact on how merchandising teams operate.?

Take two similar priced and interchangeable products, one selling 80 per week [pw] and another 100 pw. If you could only restock one, then on this information the one selling 100 pw would be the logical choice.

However, if the return rates for these two products were 12.5% and 30% respectively, then the net sales position ends up the same, at 70 pw for each.?

Gross Sales Rate (1?—?Return Rate %) = Net Sales Rate

• Product A: 80 (1–0.125) = 70
• Product B: 100 * (1–0.3) = 70

The upshot of this is that to achieve the same number of net sales, Product B effectively requires a 25% higher customer demand. One such trade-off like this may not be too noticeable, but if this occurred across a full catalogue of 10,000 products then the required increase in demand and inventory to achieve the same net sales would be quite significant.

A problem with net figures: the returns?lag

The above example probably makes it seem obvious that using net values would be superior to gross ones when making these sorts of decisions. However, as with most things with returns, it isn’t quite as simple as just switching one out for the other.

Depending on the business and sector, often these kinds of decisions need to be made quickly. In such cases, if the product is new then any returns may not yet have occurred as the returns lag will still be underway. If the product is older, then whilst the transient effect of the lag will have ended, it wouldn’t make it a fair comparison with a product which is yet to accumulate any returns.

As I wrote about in the article on product rankings, one way to mitigate this is to assign new products a default returns value. This can be assigned in a variety of ways, but as some responses to that original post noted, if this can be calculated based on product attributes or features (rather than just at the product type level) then this will assist the accuracy with which it can be predicted.

If the lag issue can be resolved, we can then start to use the net value for sales in our decision making, but is this the end of the story?

Profitability?

Returns Costs

The direct operational costs of returns comprise a whole host of factors, from logistics to put-away, customer contacts and so on. However, perhaps the most interesting observation regarding them is that because of their operational nature, for a given organisation they are more or less the same irrespective of the product being returned.

Why is this interesting? Well, it means that the costs of handling the return of a ￡10 baseball cap are the same as for a ￡100 dress, or a ￡50 pair of jeans. If those items have been sold at the same percentage margin, then their absolute margins will vary significantly. As a result, the impact of returns on each product’s profitability can be remarkably different.

Unit contribution

One way to help capture and compare this is through including returns costs within the calculation of unit contribution*.

(*Note: I’m not a finance person, so bear with me if technically this isn’t the right terminology?—?let me know if there’s a better one)

Let’s assume in our example that each returned item costs the business ￡5.

Using this figure with the products described previously, these costs sum to ￡50 for one product and ￡150 for the other. Apportioning these costs to the 70 units that have sold (and stayed sold) in each case, then this equates to a returns cost per successful sale of roughly 70p and ￡2.10 respectively.?

(I realise apportioning the costs of the returns to the non-returned items might seem odd, but think of it as a cost of achieving those positive sales)

Assuming the products are both bought for ￡6 and sold for ￡10 including VAT, they have a net sales value of ￡8.33 and a potential contribution margin (without returns) of ￡2.33.?

If this was the case, then once the cost of returns are factored in, the unit contribution is almost entirely eroded for the latter product, yet the former is still contributing around ￡1.60 for each sold item, a margin of 19%.

As this illustrates, at this low price point, return rates can be absolutely critical to profitability.?

If we now assume the two products are instead sold for ￡100, and scale the buying-in cost accordingly to ￡60 then using the same logic the contribution for each product in this case would be between ￡21.20 and ￡22.60, both around a 25% margin.?

At this higher price point, whilst the absolute costs of the returns are the same, they erode proportionally much less of the gross margin. In practice, what this means is that at higher price points, higher returns can be accepted with products still remaining profitable. Of course, lower returns will always mean greater margins and high return rates are costly in other ways?—?but the breakeven point is much higher.

Thresholds and Decision?Making

An anecdote?—?I was once asked what a good threshold for return rates on products would be to help with decisions on restocking. As you might gather from the examples above, the answer I had to give was “it depends”, as within reason any return rate could be profitable for one product and loss making for another.?

I could have suggested a range of return rates?—?depending on unit costs, sale price and so on, but the result would have been a fairly unfriendly matrix that also made product-to-product comparisons a challenge.?

Instead, adapting unit contribution, such that returns costs are considered a part of the cost of the items sold navigates around this problem of defining a singular threshold return rate. Instead of thinking “this is a high/low returner" it allows merchandisers to think “this product contributes well/poorly/not at all”, using a common margin metric that can be applied across a range of products and prices.

Inventory and?demand

Returns Mathematics

As we’ve seen, it is entirely possible for a product with a high return rate to be profitable, at least so far as unit economics are concerned. What unit contribution doesn’t capture however are the knock on impacts returns cause to inventory planning and required demand.

It can often be misunderstood that the increase in gross sales (and hence demand) required to sell through a product is found be simply adding back the return rate. However, this fails to account for the sales required to sell products returned from previous returns.?

This is perhaps best explained via an example:?

If a product with 100 units has a return rate of 10% then once those 100 units are sold, 10 of them will return. Assuming no wastage, those 10 will then be sold, with 1 of them remaining. Finally, this remaining unit will itself sell, and remain so. In all, the 10% return rate has increased the number of gross sales by 11%.?

This may seem like a small difference and at this level of returns it is, but the mathematics behind it cause it to scale rapidly as return rate increases. At 20% returns, the gross sales increase by 25%, at 30% this requires 43% more sales and by 50% an increase in gross sales of 100% is necessary to sell through.

For those eagle-eyed amongst you, you may have spotted that this follows an equation of: Increase = 1 / (1?—?Return Rate %)

Whilst it may therefore be true that a product returning at 50% provides a positive unit contribution, the additional demand generation required to achieve those sales will itself come at a cost.

Conclusion

Returns are tricky, their causes are often interlaced and their impacts are challenging. However, even without fully understanding why they happen, it is possible to start understanding what and how much they impact, and define ways to make returns aware decisions.