Fermi estimates in the application of risk control scenarios

In the construction of risk control, some tools need to be used to help business decisions. What is commonly used here is the application of Fermi estimates in risk control scenarios. In this article, how much pancakes can be sold in Beijing Hutong as an example, analyzing how to apply Femy to estimate, in order to help everyone improve their self -data thinking.

Usually the new business enters a period of rapid development, and business students are more concerned about direct business indicators such as new users and business growth.

Health indicators such as the number of orders and orders in the scene are relatively small.

When I looked back, I found that the platform had been picked up a lot of marketing funds, and began to establish risk control governance capabilities.

When to start construction of risk control, some tools need to help business make decisions.

1. Fermi estimation

On Baidu explained: Fermi estimates that the estimation problem of solving the unknown results is to disassemble the complex problems into a small and knowing result part.

Straight, it is to split complex problems into one -to -be complicated small problem.

At the same time, the logic of the solution of small problems into familiar business factors and computing logic.

For substituting the known business factors, you can calculate the estimated value of the problem.

Speaking of a very round -out, students who want to get started can search: how many piano tunners in Chicago have.

Basically, you know how to use it.

2. How many pancakes can be sold in Beijing Hutong in one day

After understanding the basic usage, use a exercise to consolidate: How many pancakes can Beijing Hutong sell in one day?

Solution idea:

Starting from the demand side:

Number of pancakes = Number of purchases* Number of purchases; Number of purchases = Hutong mouth coverage* Purchase rate is mainly concentrated in the central area of ​​the hutong, the number of people covering the Hutong port = the number of people available in the central area of ​​Beijing The number of people can be purchased = the number of people in the central area of ​​Beijing* (1 -infant ratio) According to the 21 years of Gaode statistics, the total population of Beijing is 2188W, the number of hutongs is 1004, the number of roads is 14,848, and the population of Beijing's central area accounts for 50%. Toddlers account for 20%; usually a person eats pancakes per week, it will not exceed 7 times, and at least one will not exceed once. On average, eat 3 times a week 7 days, the purchase rate = 3 /7 = 43%pancakes Seasonal interference is very low, but the time period interference is serious, usually concentrated in breakfast, and one is the main. So the purchase quantity = 1;

The above data can be obtained:

The number of people can buy = 2188W * 50% * (1 - 20%) = 875W; the number of people covering the alley = 875W * (1004 /14848) / 1004 = 569; the number of buyers = 569 * 43% = 244; * 1 = 244;

Starting from the supply side:

Number of pancakes = the main working time of stalls / time of time for pancakes is concentrated at 6:00-9:00 in the morning, 3 hours = 10800 seconds; the production time of each pancake is about 30 seconds;

The above data can be concluded: the number of pancakes = 10800 seconds / 30 seconds = 360;

Pancake business is obviously a business that is greater than in need, so it is mainly driven by demand side. The overall order will be close to the demand side magnitude.

3. How much does the business scenario exist in the business scenario

At this point, we can borrow Fermi estimation to estimate how many orders in the trading scenario exist in the trading scenario to assist the business for decision -making.

Solution idea:

Starting from the demand side:

Number of order orders = number of users to brush orders* Households all swipe orders. Single number = Veteran active number* (1 -normal loss rate of veterans) Amber number of households = (Total order number -normal order number)/ Number of orders for the number of users in the number of users = estimated order number* (1 + estimated floating floating floating fluctuations Ratio) Estimated order number = Estimated number of users to complete the number of users* Average number of orders from the supply side:

Number of orders for ordering = Total order number * Actual discount subsidy poor value Disultation subsidy difference = actual preferential rate -Estimated preferential rate. Is this a demand -side drive scenario? Or the supply side drive scene?

Fourth, summary

Here, you cultivate data analysis ideas.

In actual business scenarios, many problems are often encountered: such as the common "DAU decline problem", "payment conversion rate decline" and so on.

In fact, each answer is not the only one. What is important is the logic and data reasoning of thinking.

Even though there are some things that cannot be determined, you can gradually approach the results by analyzing disassembly. In the face of uncertainty, we should not stagnate, but we must know how to grasp value information around us and make rapid decisions.

Finally, leave a exercise: How many taxi drivers are there in China?

This article is published by @本 本 本 本 本 is a product manager. Reprinting is prohibited without permission.

The question map is from UNSPLASH, based on the CC0 protocol.

The point of view of this article only represents the author himself, and everyone is the product manager platform that only provides information storage space services.

- END -