Crack » Manage » E-commerce supply chain: is it better make or buy?
Creative Commons Licence
Read time: 12 mins

E-commerce supply chain: is it better make or buy?

Imagine you are planning to launch a traditional e-commerce website to sell exclusively your own products which can be found just on your online shop.

You’d love to understand the best way to set up the operations. Is it managing warehouse and deliveries on your own or go with a partner?  

Let’s go to investigate it!

Our goal

In traditional e-commerce, the operations play a fundamental role both for the project profitability and the customer satisfaction.

We are going here to compare two different models for managing the e-commerce supply chain in terms of end-product warehouse and customer deliveries.

The final goal is to highlight the e-commerce size, in terms of customer orders per month, such to go for one model rather than the other.

The first operating model consists in managing the end-product warehouse in-house, hence running a physical space where products are stocked, and assigning the customer deliveries to a national courier service.

The secondo operating model consists in exploiting Amazon Logistics Services to stock end-products and deliver them to customers. This can be done through the Multi-Channel Fulfillment, which offers supply chain services to sellers selling through proprietary channels.

Business data

Let’s assume some business data.

The average sales ticket is €30 and the Cost of Goods Sold (COGS) accounts 50% the sales ticket.

For every purchase, a quota of the transaction will be left to the payment processor. For every transaction, we assumed a fixed cost of €0,25 plus a variable cost accounting 1,4% of the transferred amount. This is, actually, the Stripe fee for European cards.

Moreover, to go further on the investigation, we should assume the weight and dimensions of the average product, for estimating the delivery costs, the warehouse space needed and the related costs.

So, let’s assume the average product weights 0,9kg and is 30x30x20 cm big.

At last, we are going to conform the analysis to the following business sizes, expressed in terms of customer orders per month: 50, 100, 200, 500, 1.000, 1.500 e 2.000.

The latter consideration implies that we’re taking a picture at the business in a certain point in time and considering it as stable. For each of the sizes, we’ll define the most convenient operating model.

And, as a consequence, we’ll discover which is the business size which marks the line between one operating model and the other.

The underlining assumption of the analysis is that the COGS is not changing along the business the size. This is not true, of course, but it has the same effects on both the operating models, hence it is not affecting the analysis.


Originally, the financial model was developed for the Italian market. Hence, some of the costs considered in the model refers to such market.

For example, VAT is considered at 22% rate. Moreover, warehouse and shipping costs for both operating models refer to the Italian scenario.

Lastly, the customers are expected to be all in Italy and hence the shipping costs of both operating models refer to the domestic option.   

In-house: variable direct costs

The variable direct costs in the in-house scenario are the payment processing fee and the cost of goods sold. These two cost accounts apply also to the operating model (the Amazon Multi-channel Fulfillment one).

Then, we have customer shipping costs as agreed with the courier service and the cost of the shipping box. These two cost accounts are volume dependant and do not apply to the Amazon Multi-channel Fulfillment scenario.

Which are the figures considered? Here they are.


In-house: warehouse space costs

When it comes to the warehouse, the first cost we should evaluate is the one related to the space needed. The higher the business size, the greater is the space we need. But, also, the space we need depends on the size of the average product.

Once we define the space needed, we can estimate the monthly rental cost and the costs for utilities (e.g: electricity).

In-house: warehouse workforce costs

The warehouse has to be operated, for such reason we need a dedicated person. The latter might be employed part-time (4 hours /day) of full-time (8 hours /day), depending on the business size.

The part-time employee costs to the company €1.000 /month, while the fulltime employee costs €2.000 /month.

These two costs much depend on the country we are operating our business from, because two factors affect the figures: the average salary and the employment tax rate.

To estimate how many people we need at the shop floor and the working hours, we need to estimate the time to process an order. The latter, multiplied by the number of daily orders, will provide us with the hours needed to operate the warehouse.

To keep the analysis smooth, we assume that the monthly orders are evenly distributed along the working day.

The time to process the order is instead assumed to be increasing with the business complexity.

This allocation model might be the more efficient ex-ante, but we can not say the same when used in the daily operations because it assumes a flat workload.

In-house: total cost

The costs considered so far contribute heavily to the total cost of the supply chain in the in-house scenario, but it’s not all.

In our model we have also added the costs for returns and cancellations.

The returns cost is equal to the shipping cost to have the product back, considered raising for the 3% of the orders. In this case, we are going to refund the customer and be refunded ourselves by the producer because of the faulty product, so there’s no other cost apart from the shipping.

The cancellations cost was estimated to be the 1% of the orders. Even in this case the only cost we have is the cost for having the product back, because it should get back in perfect conditions to be sold to another customer.

Amazon Multichannel Fulfillment costs

To describe the Amazon costs in a financial model, there’s not much creativity to apply. As above, we have considered the case of an Italian company shipping to Italian customers.

The program fares are based on a fixed component of €4,5 plus a variable component for every 100 grams, accounting €0,1 up to 1 kg and €0,2 above.

Moreover, for every unit sent to the warehouse, there’s a €1,8 fee to have the product stocked and packaged in a standard Amazon box 45x34x26cm, in which our average product perfectly fits.

Then, Amazon also applies a fare for the average monthly stock priced per cubic meter.

From January to September, the average stock has a cost of €26 /m3, while from October to December the cost raises up to €36 /m3. The average between these two values is €28,5 /m3 and we used this value in our model.

The figure shall be multiplied by the average monthly stock to get the cost we have to pay. So, we assume there’s a sort of safety stock every month amounting to the 20% of the monthly orders.

This 20% SKUs are, for each month, different from the SKUs in stock the previous month. Why? Because otherwise we should pay an additional fee for long-term stocks.

This is a clear simplification of the model.  

Returns and cancellations are manged by a courier service even in this case. The courier will bring the products returned to the producer or the E-commerce Manager. The costs are hence the same of the “in-house” scenario.

Profitability, compared

Comparing the profitability of the scenarios considered in the analysis, we can draw some conclusions.

The variable costs lead the “in-house” scenario to be more profitable, because we can get better shipping and packaging costs outside the Amazon world.

When we introduce the warehouse fixed costs, we need to reach a certain business size before the “in-house” scenario turs to be competitive.

According to the financial model, the line is set to 390 customer orders per month. Above the threshold, it’s more convenient to manage the supply chain “in-house”. Below, it’s better to go for Amazon Multi-channel Fulfillment.

There’s no sensitivity to the average sales ticket or the COGS because they apply at the same magnitude to both scenarios.

My two cents

We should carefully consider that the model does not feature, in the Amazon scenario, the long-term costs of unsold items in stock. This cost account may affect heavily the profitability, especially when you’re just starting and sales are uncertain more than ever.

In this case, you may schedule aggressive sell-out campaign in order to avoid long-term stock overhead.

On the other side, in the “in-house” scenario we have considered the presence of a dedicated workforce to manage the warehouse even with a business size of 50 customer orders per month. At this size, the E-commerce Manager or the entrepreneur may operate the warehouse themselves.

What would I do? I would manage the supply chain myself in order to gain more control over the customer experience.