Startups need to clarify these 6 issues before they can do a good job of data-driven operations!
|
Data analysis is only taken seriously in some particularly large domestic companies, such as BAT. Of course, this is due to their long-term accumulation and their better integration of data and operations . This is my overall feeling after returning to China. Domestic companies' understanding of data itself and the value that data can provide is not as deep as that of the United States, and the difference is quite large. Why do many companies’ data analysis become a formality? The main differences are reflected in three aspects: Value Perception Many companies are in a period of crazy growth, and the decisions made by everyone on the spur of the moment may have generated a lot of value; in this case, it is difficult for them to realize that data decisions can generate greater value than violent growth. Understanding of basic methodology Meaning a core but simple methodology. At present, there is not much knowledge about the basic methodology in China, probably because the domestic development time is relatively short, while the United States has been developing it for decades. Understanding of actual operation methods Domestic front-line employees use data to guide work operations, but have relatively less practical experience in areas such as products, customers, and sales. On the one hand, because of the short development time, on the other hand, the accumulation of data usage concepts is relatively small. However, domestic companies have been rapidly improving this awareness. However, this cognition is a step-by-step, gradual process. In the United States, cognition and methodology have gradually been well unified - technology and business are integrated with data. In China, there is a huge gap between technology and business. Engineers are required to build data systems, but they do not really understand the business side; the business side is not very familiar with the technology either, resulting in many requirements not being able to be directly met using existing technical means. Lack of understanding between each other further exacerbates the slow pace of data usage. Talking in different ways will result in reduced efficiency and inability to realize value. Neither party can benefit from it, and in the end, decisions are made based on feelings rather than real data operations. When many companies start from scratch, they spend a lot of time building a technology platform. First, the technical platform is very complex and requires a variety of different engineering personnel; second, many companies are exploring from scratch, but the data analysis system requires a series of processes and talents, each of which cannot be too weak, so that they can be truly connected. Today, competition in China is too fierce and businesses are developing too fast. We don’t have enough time to produce some good things like BAT and Google . This is also the reason why many companies fail to see the value of data. Many domestic entrepreneurs do not realize the value of data at first; when they realize the value of data, their expectations are often very high. This huge gap also makes it impossible for value to be truly realized, and even makes people question whether "this value can really be realized" and lack patience. What kind of companies need to pay attention to data? Generally speaking, the companies that pay more attention to data in China are those with high average order value and focus on conversion, such as Internet finance, e-commerce , trading platforms, SaaS, and online travel companies. This type of customer has a high average order value and does not rely entirely on traffic, so entrepreneurs will have the motivation to increase conversions. Generally speaking, entrepreneurs will go through 4-5 product and enterprise life cycles. The first stage is called cold start. At this time, the company is in a very early stage, and has not even successfully completed the angel round or A round of financing. For companies at this stage, using big data as a driver is a false proposition because the number of customers is limited and the sample size is insufficient. They need to understand more about the needs of potential customers and "beg" customers to use this product. The second stage - the early growth stage - is when the cold start is almost complete. Experienced entrepreneurs will start to plan some core indicators related to growth, such as daily/monthly activity and retention. The purpose of these metrics is not to measure the current performance of the product, but to provide a comparable benchmark for future growth. And these indicators can tell us when we should do growth. If the product itself has no stickiness, burning money to grow it will not really grow. Because the loss rate exceeds the growth rate. In the past, many money-burning companies were able to succeed because competition was not that fierce and users did not have so many choices. But today, if your product is poor, retention is low, and word of mouth is not good, no matter how much money you spend, you cannot achieve real core natural growth. The third stage is the growth period. At this stage, you can see the huge difference between good startups and ordinary startups - efficiency. Whether it is PR or organizing events, it requires manpower and time costs. How to find the most efficient channel during growth? I think this is the core competitiveness among startups. If you don't do data-driven things and just rely on intuition, it may work once or twice, but no one can go to a casino and win ten thousand times in a row. Therefore, intuition needs to be combined with data so that companies can quickly optimize various channels to improve conversion efficiency per unit time. By continuously improving and superimposing the conversion efficiency per unit time, it can become the core competitiveness of the enterprise. A company that is not data-driven, and a company that is data-driven. Assuming the same operating strategies, similar capital reserves, and the same customers, the company that can learn quickly from data will definitely win. The fourth stage is the monetization period. Business monetization requires a very large user base. For general Internet products, a small number of highly active users with good experiences will be converted into paying users. It is like a funnel that continuously screens, and the efficiency of operations is what matters here. For example, the conversion funnel of e-commerce users is generally: visit - register - search - browse - add to shopping cart - payment, or return in the future. This is a very, very long funnel. To truly implement data-driven operations, we must continuously track every link in the funnel. Why? It is difficult to achieve growth because it cannot be measured. A good company, especially one that plans to generate revenue in the future, must pay attention to the conversion efficiency of each department and each link. This conversion efficiency can be achieved through marketing methods, product improvement methods, and even customer operation methods. A small improvement in each link together will result in an exponential improvement. It is difficult for people who have not worked in data-driven operations to understand how huge this kind of multiplication will be. For example, when we were doing data-driven conversions at LinkedIn in the past, if we wanted to push a certain EDM to 100,000 people, the conversion rate based on a snap decision was 0.01%. However, after the data-driven department made a simple data model and pushed the same message, the conversion rate increased to 0.3%, which is much higher. If you do this every week, the conversion effect will be very significant. Each industry has its own KPIs. For example, in the SaaS industry, whether user registration can be successful is such a simple question, but many companies may overlook it. After the user successfully registers, have you identified your core product features, and has the user used your core features? Which core product features keep users coming back? Which features don’t? These should all be recorded in product analysis, but how to analyze if there is no data? How to measure it? Many American companies have summarized these things and have been using them for more than ten years. Many domestic companies can imitate and learn from these experiences. There is no need to go through them again blindly, which would be a waste of time and resources. Another point is that enterprises should be operationalized. What concept? That is to say, data analysis is not a campaign, but a daily matter - we are looking at these things every day, every week, every month, and every quarter. Continuous tuning, learning and promotion is a very important process. But developing habits is quite painful, because many entrepreneurs are very busy and have no time to look at those things. What common myths do Chinese companies have about data? I think domestic companies have two extreme understandings of data analysis: one believes that it is pure technology, while the other is more superstitious, believing that as long as they engage in big data, they will become high-end companies. I think there are certain misunderstandings in both approaches. At its core, I think whether what you have made is valuable and effective? The most direct way to measure is by effect. Some other companies want to build their own platforms and set up large teams, but their efficiency and output are relatively low. I advise everyone to be cautious about this. As the ecosystem continues to develop, many tools are now very useful, and you have to learn how to use them. These are some great aids for entrepreneurs to succeed - you can't say that because you know how to use the tools, you will succeed in starting a business; but good entrepreneurs will definitely be able to use these various tools to achieve their goals. What does good data analysis look like? Good data analysis can benefit everyone in the company. It is not a privilege, nor is it only for one or two people in the company, but it can directly benefit all operating departments in the company, especially those fighting on the front line. Normally, we only talk about strategy and general direction, and only show it to the CEO, VP or operations - this is not enough. It needs to be given to frontline employees so that they can use it. I think this is a big difference between a data-driven enterprise and a non-data-driven enterprise. Efficiency improvement means improvement for everyone, not just for one or two people. If a company wants to build a complete data analysis mechanism, it should start with the business. All data analysis operations or data systems should start with the business and the customers. This data analysis system should not only solve one or two very narrow problems, but also need to have a system and a big picture. Then, in fact, the most difficult part of data analysis is data collection and data organization. This process is the most time-consuming, probably because the initial plan was not thorough enough. Therefore, we should pay great attention to data collection and data organization in a planned manner. In the end, data analysis cannot just stay on the basis of reporting, the value is still not enough. Ultimately, once those numbers come out, it’s right and effective to tell others what they should do. There is a lot of profound knowledge involved here, and it requires strong operational capabilities. Therefore, an enterprise must have a broad perspective and also focus on feasibility. I suggest that if general enterprises want to build their own, they should first break through from a single point, find a transformation point, see the value, and then learn the method for the next practice through this practice. This is also a learning process. Don’t start by building a huge system or combining 50 data circles together in an attempt to build a data science framework. I think if you do this generally, unless you have a lot of resources, you will definitely fail. How to break the vicious circle of data not being fully utilized? In the past few months, we have been dealing with customers and found that some companies use our products very well, while others are just so-so. Usually, companies that have someone in-house who is responsible for data use it very well; some companies do not have a core person to follow up on this matter, and their performance is average. Therefore, there must be at least one person in the operations department who has a certain understanding of data analysis. It's like if we move a set of advanced surgical equipment into the company, it will be useless if no one knows how to operate it. I think the best way to acquire knowledge is through practical operation. The prerequisite for actual operation is to have someone who knows a little bit and can guide you to do it a few times. Then start to apply and learn. This is the fastest and most effective way to acquire data analysis knowledge. I don't think just reading books or textbooks, or looking at some external big data guidance books, can have this effect. With this person, we can get methodological support from people who understand this area and the company's products, and this learning mechanism will be established. This is quite important. Otherwise, even if the system is powerful, it cannot be fully utilized if no one knows how to operate it. FAQs for startups Especially early-stage companies, the things they focus on are very standardized. For example, they want to know about new users, retained users, strong channels, which product features new users use, etc. Every company has different strengths and weaknesses. Let me give you two examples: For example, there is a client who is a SaaS company. They do a lot of offline activities and then direct traffic online. But he had never observed his own registration conversion. As a result, the volume came, but the conversion rate was very low, and the actual registrations were still very low. Later, through simple optimization of the registration process, the conversion was increased three or four times. For example, for e-commerce clients, we used to look at how many people came to trade, how much the transaction amount was, and then the weekly/monthly/quarterly growth rate. But those who come in early are generally core users, and the growth rate is faster. And then it soon entered a plateau period. Why? This is because many things in it are not detailed enough and a lot of them are diluted. For example, in auctions, many users come and it seems to be prosperous, but if there are too many categories, the number of bidders for each product will decrease. Once the quantity decreases, the price will decrease. After the decrease, the GMV and growth rate will decrease. In this case, it needs to consider focusing more users on relatively few products, thereby increasing the average order value. The advantage of doing this is that sellers can increase their sales and be more willing to sell on your platform; buyers will also feel that they have purchased scarce items. Retention is the most core and most pressing issue that a startup needs to solve if it wants to succeed. With the retention rate , you basically have the growth rate. The core users attracted in the early stage generally have a higher retention rate; the users attracted in the later stage have relatively low stickiness. The more successful Internet products usually focus on core users in the early stages, meet their needs, and then continue to spread downwards. Therefore, retention should still receive more attention. At the same time, retained users also need to be decomposed. The users who remain are some new users and some old users. It seems that they are measured by the same time, but in fact they are different. Many startups sometimes don’t break it down into categories: for example, among retained users, how many are new users and how many are old users; what is the retention rate for old users and what is the retention rate for new users? Facebook divides users into seven levels. What do these seven levels mean? That is to say, the activity level of each user is different this week. Some people come for seven days, some for six days, five days, four days, three days, or two days. It breaks down the daily user activity into very detailed categories. Then, it is further split into new users and old users based on this dimension. After the breakdown, you can operate for each different type of user . For example, it can analyze which functions are used by users who use it more than five days a week. My suggestion is that in the early stages of a product, you should focus on product retention before adding new products. This way, the founder’s energy will be more focused. Because if you try to attract new customers and retain existing customers at the same time, you'll be dividing your efforts into two parts and you won't be able to take care of both. Having high retention will also help in attracting new users. You can find channels to acquire high-retention users and then replicate the operations continuously. The second point is that once you have good retention, you can expand quickly. Because after the expansion, users will stay and your growth rate will accelerate. There is actually a very ready-made methodology here, because if you don’t find good retention, the business you are running is a money-burning business. If the financing environment is poor, then the business is likely to fail. But if your user stickiness is very high, your operating costs will be very low, so the founder can manage the entire resource allocation. I think this basic way of thinking is necessary after the product is cold-started; during the growth phase, extreme focus is required. In the early days, we relied on intuition, and in the later days, we relied on science. I think the earlier you lay the data groundwork, the more beneficial it will be for a company. It is a process of continuous iteration and accumulation. However, don’t put the cart before the horse and do AB testing right after you launch your app. It’s unnecessary because you haven’t accumulated enough users and the data analyzed is not representative. Finally, I will briefly summarize the five stages of data analysis: ➤The first stage is when there is nothing; ➤The second stage requires the company to be able to trace back history: know what is happening with its products, which is the most basic and original stage; ➤The third stage, people who work on products, operations, and marketing internally need to ask why: This stage is prediction, that is, predicting what a certain group of people will do next, so that products can be developed in a targeted and better manner; ➤The fourth stage is to have a solution: that is, I predict that this group of people will do this, then I will give it a better solution, so that it can have better conversion and retention, and bring better new customer acquisition effects; ➤The fifth stage is optimization, how can diversified product lines find the best balance point: there is a balance point in all aspects of price, marketing, product design, and sales. This balance point is the point where the entrepreneur’s interests are maximized, and it is also the point where users like this product the most. These five stages require time and continuous accumulation. Don't skip, as skipping often leads to failure. Start from the basics.
|
<<: Baidu bidding data analysis skills, what aspects should be considered in bidding data analysis?
>>: 27 provinces have resumed passenger routes! Which 27 provinces are they? Attached is the list!
Recommend
Disassembling 6 ways to play red envelopes in event marketing
Red envelopes are a value-added service provided ...
7 Christmas marketing cases worth seeing, take them and thank you!
Christmas is here, and it’s time to have fun agai...
How to open a store through WeChat Mini Program? What are the steps?
When smartphones appeared, few people thought tha...
How to play with your own nipples to feel good? How to play with nipples comfortably?
How to pinch and how to make it feel good? ? Plea...
How can a startup team quickly acquire seed users and detonate a user nuclear reactor?
After the product is launched, someone is needed ...
Shen Yi, "Reading to Lead the Trend: Five Thousand Years of Poetry"
Shen Yi's "Reading to Lead the Trend: Fi...
Disassembling Xiaohongshu’s home notes and summarizing 3 types of hot-selling items!
No matter your age, you must have imagined what y...
Gaocheng SEO training: How to correctly select keywords for online promotion and marketing?
How to correctly select keywords for online promo...
How to complain about false sales behavior in WeChat Mini Programs? Can I file a complaint about after-sales service for items purchased through the Mini Program?
As many businesses have opened their own WeChat m...
How much does it cost to develop a Panjin toy mini program? What is the quotation for Panjin Toy Mini Program development?
More and more businesses are paying attention to ...
Soul product analysis!
In China, stranger social apps are a mysterious e...
Baidu video promotion, vertical video information flow advertising style!
Information flow advertising style - vertical vid...
Xiaohongshu Food Popularity Article Methodology
The article starts with analyzing the hot food ar...
Facebook Advertising Guide!
If your issue is that your ads aren't serving...
New Media Matrix - How to distribute one article on multiple platforms most effectively!
" Wanda 's new media matrix ", &quo...