# Making decisions by using your business data

### Introduction

I have been working with startups in Russia for already 4 years now. I help them to make decisions based on their business data. Now I want to speak about how I do this and what tools and methods I use.

First of all I use unit economics and Goldratt’s Theory of Constraints. Let’s take as an example a small business selling by subscription boxes with educational material for babies. We have the following data:

Price $23, Educational material $10, Box & Delivery $5, Customer Acquisition Cost $5 and Profit from one sold box $3. On the average each customer buys 2 boxes per year.

This is business data provided to VC by the founder. Let’s see, how much money we can get from this business model.

Average Revenue Per Customer (**ARPC**) = ($23 — ($10 + $5)) x 2 = $16

Customer Acquisition Cost (**CAC**) = $5

Return of Marketing Investments (**ROMI**) more than 300%!

What decisions can we make considering this data? The answer is **no decisions at all**! The thing is CAC is not an actionable metric. It depends on two other metrics: User Acquisition (**UA**) and Conversion from User to Customer (**С**).

According to the business data, C is equal to 4.33%. Now we can recalculate everything with actionable metrics UA & C:

Average Revenue Per User (**ARPU**) = ARPC x C = $0.69, that means every user gives us $0.69 of revenue.

Cost Per Acquisition (**CPA**) = CAC x C = $0.22, that means we pay for every user $0.22.

Our ROMI doesn’t change and is equal to ARPU / CPA = 313%, that means more than 300%!

So what do these numbers mean to us? That business is ready to expand! We should put money into marketing and get our profit. Every user visiting our product gives us $0.47, no matter whether he buys something or not.

**Sounds cool. But it is totally bullshit! Why?**

Let’s have another look on our business data:

C=4.33%, CPA=$0.22, and APC=2 (Average Payment Count).

What does it mean? First of all, it means that our business team must have some skills to get such metric values. And we must find out whether their skills are enough to get such metric values or not. Our business team used wrong data for Conversion. The thing is we have two big groups of users:

- users who have never bought our product;
- users who bought our product.

These groups act differently. The first one buys our promise that our product will satisfy their needs, the second one buys our product because it actually does this.

If we put our users in two groups, we will get 1.57% conversion for the first purchase.

The second problem is APC=2. How can we calculate the value of this metric? We need to divide all transactions by the number of customers. Thus we’ll get 1.4. But our business team somehow got 2. That’s the problem of rounding of metrics. Let’s see what will happen to our economy after using the correct data:

ARPC = ($23 — ($10 + $5)) x 1.4 = $11.2

ARPU = ARPC x 1,57% = $0.18

But CPA is still equal to $0.22 (we still attract 15 000 users to our landing page). Our ROMI drops down to 82%, so our economy shows negative growth. It means that our first decision about business expansion was wrong, because we would only expand our losses.

In addition to this, let’s look at the CPA. Can we change the value of this metric so that it will bring us some profit? Our CPA is equal to $0.22, so it is very cheap. But what will happen if we change it to $0.17? Let’s calculate: ARPU = $0.18, CPA=$0.17 and Contribution Margin (**CM**) is equal to CM = UA x (ARPU — CPA) = 15 000 x ($0.18-$0.17) = $150. Bingo, our economy shows positive growth. But how many users we need to make $100,000? The answer is 10,000,000! And we must also have some marketing skills to get ten million users.

If we look closely at our marketing, we’ll find that most of the customers we obtain from channels like Direct type in, Google Search etc. It means that our business team doesn’t have any skills to attract users by marketing channels, especially ten millions.

So we come to the conclusion that our business team is not ready to expand their business. Their economy shows negative growth, they don’t have skills in marketing and finally, CPA is not an actionable metric. Even if we make it equal to $0.17, we still can’t get ten million of users.

### Unit economics

All businesses are like conveyors. We take users and turn them into profit. At first we use marketing to attract users. All of them come through our business (black box) and some become our customers — who bring us profit.

Contribution margin (CM) — is the money we get from selling our goods or services. CM is equal to Users Acquisition multiplied on the revenue from each user (ARPU — CPA):

CM = UA x (ARPU — CPA)

We remember that ARPU = ARPC x C1, where C1 is conversion to the first sale. And now we get a good formula describing our business:

CM = UA x (ARPC x C1 — CPA)

Different monetization models need different ARPC formula. So every business needs its own formula.

### Conveyor for our business

If we put our formula in the table, where all metrics have their own cell, we’ll get a conveyor that make money from people.

On the left we see people who are interested in our product. They come through some machines like C1, AvPrice, COGS, 1sCOGS, APC, CPA and turn into our money — CM.

If we have conveyor we can use any optimization method for conveyor, for example, Goldratt’s Theory of Constraints. The method, in short, consists of four steps:

- Find the bottleneck of our business model — the place where we lose money. The idea is that by small change of this place (our machine) we’ll get big change of CM.
- Focus the work of the entire conveyor on this metric.
- Expand the bottleneck.
- Return to the first step.

In our example, we change C1 from 4.33% to 1.57%. Let’s see how this change affects our CM.

We changed C1 from 4.33% to 1.57% that means it decreased by 3 times. And our CM changed dramatically — it decreased by 15 times! So this metric deeply affects our CM.

Taking this effect into consideration, we can calculate what metrics’ value we need to achieve the needed CM value. For example, we need to get CM=100,000

As you can see, not all alterations give us the desired result. For example, we can’t reach the needed CM value by changing 1sCOGS and CPA. This method only shows us where is the bottleneck but doesn’t give any ready solutions. How difficult will it be to change conversion from 4.33% to 43.04%? Maybe we should just change conversion from 4.33% to 4.67% and then change Av.Price from $23 to $35?

We should slowly change the value of metrics and check the bottleneck after each alteration. We also have to remember that with each metric alteration it becomes more difficult to make any further alterations.

So, as you can see, we can get the needed CM value by small alterations of different metrics.

### Conclusion

Using business data we can find the bottleneck and work out the metrics value we need to get the desired Contribution Margin. That means we can focus on specific business processes that can improve these metrics and help to expand your business.