In e-commerce marketing two of the fundamental questions are: how much can be spent to acquire a customer and what advertising is working? Both of these seem to be quite simple but are actually quite complicated.

Most spending in business has an economy of scale: for example the more money you spend on courier the cheaper shipping gets. Marketing has a diseconomy of scale, a good marketing team will choose the best marketing channels and use them to capacity, then the second best and so on. Many of the best channels (such as search marketing) have a very limited volume.

This is a typical graph of Customer Acquisition Cost (CAC) verses volume of advertising:

So to work out how much you can spend to acquire a customer you need to consider i) what the customer will spend over time ii) margin and iii) how long you are willing to wait for profitability.

A second challenge is working out what advertising is working. Here is an example of two adverts:

Advert #1
CAC $17
Advert #2
CAC $20
Sales 2017 $20 $22
Sales 2018 $5 $18
Sales 2019 $2 $14
LTV $27
Loss $3.50
LTV $54
Profit $7.00

In this case simply relying on the CAC would give the wrong answer. This is a very common situation as we often come across advertising where it is more expensive to recruit better quality customers.

The answer to both of these problems is to understand historic and forecast Lifetime Value (LTV) and to focus on that as the key metric. Machine Labs use a range of algorithms and machine learning (when there is sufficient data) to give your forecasts on how well your campaigns are performing.

This allows you to work out what your budget should be and what adverts are working so that your budget is spent effectively.