What is a data closed loop? Let’s talk about data closed loop

Data closed loop can be simply understood as an iterative cycle with data and monitoring. It can be applied in corporate business to help companies improve performance and drive growth. So, what elements does a data closed loop need to have in order to really work? In this article, the author explains the definition and value of data closed loop. Let's take a look.

Introduction

Now is an era where everyone feels that they are being squeezed, and as a person from Northeast China, the word "squeeze" reminds me of a flower roll, which is rolled up in circles and rings. In various companies, one way to keep squeezing in with the industry, parallel teams, and upstream and downstream is to close the data loop.

Today I will try to write about what I understand as a data closed loop and what it can do. I will try to explain an abstract concept through some simple examples.

1. What is a data closed loop?

When I first heard this word, my first reaction was: Is this another newly created word? But as I learned more about the business, I understood its sufficiency and necessity. At the same time, I found an analogy that everyone is familiar with: the wrong question book.

We can recall how the wrong question book came about. In order to help everyone improve their academic performance, the teacher proposed this learning plan. The general process is:

• Teachers teach, students learn;
• Regular examinations, monthly exams, mid-term exams, and final exams;
• Pull up the list, rank the students, and distribute the test papers;
• The teacher explains the test paper and the students sort out their wrong answers;
• The student's own analysis and classification of wrong questions (if the student does it);
• Then continue the cycle.

If you are a poor student, you probably don’t have to sort out your mistakes, but we must learn from the top students. The cycle I just mentioned is first of all a closed loop. Secondly, it contains knowledge points, scores, and rankings, which can be considered a data closed loop. If we include the "parents' meeting", we can even say that the reward and punishment mechanism is also included.

If we make an analogy to a company, the process is roughly as follows:

• The proposal, discussion and development of a demand, which may be a marketing plan or a data dashboard, etc.
• Development and implementation of requirements, followed by testing and launch;
• Perform corresponding data collection and data monitoring;
• Comparative analysis of actual monitored indicators with KPIs, OKRs, historical data, and industry data;
• Everyone should improve their own part and move towards KPI and OKR again;
• Then continue the cycle.

Do they look similar? They are both loops, iterative loops with data and monitoring.

2. What can a closed data loop do?

Just now we used an analogy to explain what a data loop is. Now let’s take a look at what a data loop can do.

In fact, judging from the analogy of the wrong answer book above, the main purpose is to improve grades. If we make an analogy to a company, it means to improve business performance indicators.

But how is this done?

I understand that we need to look at it from three levels: data, monitoring, and connection. Only when these three exist can the closed loop be meaningful.

1. Data

Let’s start with the data. In the past, parking lot barriers were all manually controlled by remote controls, which created a problem: it was difficult to count how many vehicles entered and exited today, which left room for human manipulation. Now, when camera recognition, time statistics, and QR code payment are linked together, all can be digitized, basically eliminating the human problems mentioned above.

2. Monitoring

Let's look at monitoring. If we use the previous example, should the parking lot operate 24 hours a day? If it does, there must be someone at each entrance ready to remotely lift the barrier, so at least two shifts are required; if it does not operate 24 hours a day, what if the parking user wants to go out at night? Therefore, it is necessary to monitor this dynamic at all times to ensure the normal operation of the parking lot.

Let's use the clock-in function of office software to look at this problem. When getting up early, everyone clocks in using the fingerprint machine, work card reader, or face recognition machine at the company gate. The biggest problem is the low concurrency. I wonder if readers have ever experienced the situation of queuing up to clock in in the morning? Especially in autumn and winter, some people's fingers will peel off, and then they will not be recognized, causing everyone in the queue to be late.

What if you clock in via APP? When you open the office software, we can actively monitor whether you are near the company (GPS, company WiFi, etc.). If you are, just clock in and then notify the user; if not, check how long it will take to clock in. You can also give a friendly reminder, which not only avoids the problem of low queue concurrency, but also solves the problem of fingerprint scanning failure.

This is the advantage of monitoring after data is digitized. This is not from the boss's perspective, but just a comparison of the pros and cons of different clock-in modes.

3. Links

Regarding links, there are actually links in the case just mentioned, which are mainly the combination of software and hardware and the links between people. But more links are actually the links between organizations, departments, and people.

Let's look at the clock-in issue again. If an employee is frequently late recently and is always a few minutes late, as the department HR or his +1, should you take the initiative to inquire why, whether he has any family or physical difficulties that he needs help to overcome, or whether he has been working too late recently and needs to increase or adjust his work content? And if a person is late for a long time, then shouldn't HR consider preparing to recruit people?

If we look at the food delivery service that everyone often uses, it involves restaurants, riders, and customers, and the status of information flows among the three. From the customer browsing the menu and paying for the order, to the restaurant accepting the order and preparing the meal, to the rider picking up the meal, transporting and delivering it, the entire link connects the three parties and also supports real-time online interaction between the three parties. Such efficiency or user perception level is definitely much better than placing an order and waiting foolishly.

So if we combine the three points we mentioned above, we will find that those that are not digitized need to be digitized; with digitization, we can set various monitoring indicators, and monitor whether they are high or low, and provide them to the corresponding person in charge in a timely manner; at the same time, information from different links can flow in the entire business channel, so that everyone can know who is doing what in real time, and on this basis, the most important result is born: iteration.

4. Iteration

One of the strong dependencies for iteration is *industry knowledge*. The examples above are basically some examples that we often see in our daily lives, and the data closed loops in some industries may not be in our daily lives at all. For example, manufacturing, clinical trials, etc. Therefore, real iteration is inseparable from business knowledge. But in order to better understand this matter, we still need to find an everyday example. Let's take takeout as an example.

In the link we just talked about, if we want to shorten the time from ordering to delivery, there are many nodes that can be optimized. We will analyze the points that can be optimized in this closed loop from the perspective of restaurant cooking and rider transportation.

Restaurant food preparation: The eating habits or methods of a single person may fluctuate, but the eating habits of a group of people in an office building or park follow a general trend. So if there is a pattern in ordering food within a 5km radius, especially when there are discounts, this can in turn affect the progress and quantity of food preparation in the restaurant. If there is an error or a large lack of understanding, the process can be optimized: order food in advance. Notify everyone at 10 o'clock to order food in advance and eat at noon. These are iterative methods and directions.

Meal delivery by riders: This is actually similar to navigation, except that the difference is the global optimal solution for multiple transit points. A rider delivers 5 items at a time, and may pick up 2 items along the way, which means 7 items. How should this process be carried out? Or if it is found that the overall time will exceed, the rider can let other riders pick up the 2 items. These are all points that can be continuously optimized.

I even think that the best iteration of food delivery is "delay insurance". Since we have monitored every link and tried various methods, users still feel that the delivery is slow, then OK! I will use delay insurance to make you less disgusted with being late.

I originally expected it to arrive at 11:00. If I didn't arrive, the customer would complain. But now I tell you that if it doesn't arrive at 11:15, the freight insurance will be paid, so the customer will probably want to wait. However, the system believes that the probability of arriving at 11:00 is 80%, and the probability of arriving at 11:15 is 99.99% (personal guess, no empirical evidence).

3. Does data closure only solve bad cases?

The data closed loop is a system, and it is even more complex for some companies with complex businesses and a large number of people. A system generally includes positive and negative cycles;

All the situations just mentioned are based on the negative cycle, that is, using negative examples to make the system more stable. We can also look at how positive examples are reflected and used in the data closed loop.

Now many offline outlets of new energy vehicle companies are directly operated by the companies, and each salesperson is equipped with corresponding marketing software (maybe a corporate WeChat). In the process of regular statistics, we actually find that some stores and some store employees have sales performance far ahead of their peers. This raises a question: What did this far-ahead person or store do right?

We can let him talk about what he did, or use the data that the system can analyze to see the difference between him and others, and then rationally analyze whether there is a logical causal relationship in the middle. If so, then it should be deposited into the system, or deposited into training, or deposited into assessment.

For example, he will ask customers a second time on the third day after they arrive at the store. This has a higher conversion rate than the seventh day, so he should change it to three days.

For example, he asked customers to register an APP after they arrived at the store, get a small gift first, and then they can pay more attention to the price and configuration information of the vehicle on the car company's APP later, so other people will also do it;

For example, if this store has a children's playground, parents can bring their children there and the children can play for a while longer, and the parents will have more time to interact with the salesperson. Other stores should also do the same.

All of the above are positive cycles in the data closed loop.

4. Final Thoughts

Both the top students and the poor students in the company may have used data closed loops, but most of the poor students in schools do not use this thing. This is a big difference between companies and schools, so everyone should be cautious when actually implementing data closed loops in the company.

Author: Dai Chenglong