Lei feng's network (search for "Lei feng's network" public attention): author of North Sea, from the calculation, public advertising (ID:Comp_Ad)
The saying goes, "good wine needs no Bush"; they also said, "wine is also deeply afraid of the alley." Then, they said, his wine not fragrant Bush is not deep, just get an abacus weave his magic skills, accountant, data scientists, wine can be sold. This is called data-driven rapid user growth, jargon called "Growth Hacking."
First heard about "Growth Hacking" this word, is in the last year a meeting of big data. (I forget, because now all the conferences are called data conferencing. ) When I got a Pocket full of negative energy, ready to power released a cacophony of comments, suddenly found a former Facebook engineers how to use data analysis to drive user product "Growth Hacking."
In fact, the engineer's speeches are one of the fair, and no advertising material is not a panacea, as also stressed that only products can improve retention rates. But then, this thing was out of shape, the Internet community there beliefs, data analysis is the Midas touch of magic. As you did for many years on the streets before the tabloid publicity: through the clothing acupuncture, can cause increased breast. As a code for so many years to deal with data and livestock, has said nothing on the issue, is the throat. Today we'll talk about, data analysis can really drive the rapid growth of consumer products?
For data and methods in the Jurassic and Cretaceous periods, the Chinese market, emphasized the role of data in General is a meaning of. Look at the data, rather than simple thanks to the wealth of useful, but then again again, data are really not made of money, does not mean that you believe in him and receive eternal life. Saint Cloud, "all number than many" useful in data operations in those scenes, how to make it useful, is the public's highly qualified readers to figure out the problem.
| Data management of the three axes
About using data-driven business growth, there have recently been many books and lectures. In fact, the way there is no mystery here, most data operations, methods can be summed up in the following three axes:
A, building conversion funnels
The so-called conversion funnel, is a step by step how your business users to cheat. Following; a few, you know at a glance:
Advertising: show-> click-> transform
Games: download-> activation-> retirement pay->
Sister: shaking it kissing dating->-> hand->-> bed
Regardless of the business above, can be broken down into a series of stages, after each stage, users are only part preserved. On every aspect of the funnel and accurately record data in order to analyze and optimize all aspects of the adoption rates, data operations infrastructure.
Second, multi-dimensional data report to find the problem
Common pain points in data operations, is an aspect of knowing conversion funnel through rate is low, but can not find the way to improve the. Common solutions, is playing fine data down to each dimension were observed, which often can be found on the product or system problems. If more than one dimension can be flexibly combined observation data into a data cube (data Cube). The funnel and Internet product operation the figure below data it doesn't matter, but the principles are the same.
For example, you find the ad click rate is low, which is found on the Chrome browser hits down the overall statistics, then delve into causes at the Chrome browser, the result is likely to be your Flash advertising materials directly to shield by Chrome.
The multidimensional data reports to locate and find the problem quite effectively, it is actually an efficient debug, is still a "taking" strategies.
Third, using a/b testing to guide product evolution
Is there any data-driven "attack" strategy? There are, of course, develop more products may improve, putting them on the line, who falls in the real data to decide. This a/b testing methods, tend to be all lies in the ideal can optimize the magic of good products, but also "better" one of the foundations of the theory.
Speaking of A/B system, but is not a simple thing. How to build both the accuracy and efficiency of the experimental framework, and write a long article, and we will not go into here.
The three axes above, for running a product is very important. But if you think you have this kind of thinking of data, you can use data analysis to make great products, it might as well wash sleep. Victorias Secret iPhone 5 Case
| Data operation can not solve the problem
Users on how to choose and evaluate a product, different areas have different rules. In simple terms, we can divide the products into rational and emotional products. For example, 3C, is typical of the rational products and apparel merchant, is the perception of relative products. Calculating advertising and recommendation system, although the technology stack have similarities, but the former is the rational products, the latter sensibility is much stronger.
For rational products, due to the objective of a stable and easy to quantify, data is one of the most critical optimization techniques. Advertising products, advertisers use its purpose, in order to obtain higher profits (of course this profit may be long-term or short-term), rather than to spiritual joy or pleasure. Thus, when two advertising platform is very big difference in the input-output ratio, clients will not be taken into account which experience superior, but did not hesitate to choose that makes more money.
But when it comes to sensitive products, it is not so simple. Remember the app just fire up when a wave from the side of the line officers will be diverted over the Internet analysts have Lotus at the mouth, with analysis of why the app is the end product of human social, why people still use QQ is the obstacle of the wheel of history. But last year, about the same analysts, and discuss why after the app transfer to mobile QQ users are having. So what is QQ good or micro-well? So problems at different times, under different users have different answers, and we're never going to such mobile IM products, give a general quantitative targets. Operating in sensitive products, since it is difficult to determine the optimal target is given, data optimization is the ceiling.
So, data operations under what circumstances could experience significant bottleneck? Generally speaking the following aspects.
Direction, product innovation cannot be obtained through data
A few years ago, there was a popular gaming company Zynga. It is said that Zynga's boss does not encourage innovation, adhere to the "-ISM", others copied game ideas, fast operating system with its own set of data ahead of the competition. What kind of data operating system? After all is a lot of a/b testing. Design: grass is green. Product manager says: no, green red data that counts! Then they really will flow into red grass and green grass in two configurations, if the data reflecting the Red meadow high user fees, then into the grass, red, botanists can go to hell! Relying on such a system, prolonged occupation within Zynga Facebook games top three. What happened to it? Answer is very clear – who know Zynga?
Zynga of decline certainly reasons many, but had to said only data on the of products operation thought is pushing one of: you of data test system again mature, also but on can make make grass coated red coated green such of tricks, and game of new mode, and new plot, and new design such of real innovation direction, not you has has data operation system on can has, even also cannot through data system to judge high. Take the iPhone for example, large screen interactive, Multi-touch, store these shine product features, painted by his faith in the products and insight, rather than markets and demand for research results.
Even when small short-term improvements, total reliance on an a/b test also does not work. We are able to pick out the data of a good plan, but if you only X/Y/Z option, then how do you measure a also can't get out. Someone asks: If all listed all the possible choice of products, data to choose? Operation is not playing mahjong, cards that combine the total number of children, will truly have the potential products need to have systematic and innovative thinking, product manager, coffee and others go to the bathroom on time trying, it is possible to find. Besides, even products like tiles, so clearly, when the product direction and becomes a lot, due to "dimension curse" exists, we can't accumulate sufficient statistical data to make decisions within a limited time.
Here actually have when it comes to the user product is truly the most important engine of growth, that is, product manager of insight and creativity.
Second, long-term user feedback it is difficult to judge
The Facebook engineer in his speech spoke of a vivid example: Facebook for years insisted on rigorous a/b testing framework decision if a new feature is been used in online systems. However,; what is it? He said Facebook actually PC version of Home for three years without any significant upgrades. In fact, the result is thought-provoking in itself: is Facebook becomes the end product of human social networks, and has been optimized to a global optimum and cannot be improved?
Conclusion clearly is not the case, I think even in that three-year abandoned many of the revised programmes, not necessarily manslaughter version: current a/b testing framework, data for only a relatively short period, and long-term trends and conclusions, if there is no faith in the product adheres to, the day it is hard to wait to come to fruition.
I heard about it from the traditional print journal rule: generally, the magazine for the first few months after the revision, all stink. And then a few months later, the new version may lead to significant increases in circulation. It can also be used as a case of short-term data deficiencies.
Three, game scene cannot use data decision
Game sex scenes are very common in the Internet products. For example, you might know, Explore and Exploit problem for Internet products, is part of stochastic-flow to explore unknown space, another part of the flow according to the statistically optimal decisions. Friends familiar with this area knows that if we have two E&E strategy is through a/b testing to determine their advantages and disadvantages. As to why, it is recommended that you settle E&E background knowledge on their own, when one questions the right now!
In addition to E&E, still many other game scenarios, such as advertising in mechanism design problem (that is, the rules of the bidding market how to formulate); another example has some social gaming strategy, are in principle not through simple flow can be effective a/b test.
Of course, understanding the game questions, needs deep product insight and macro-thinking skills, there are many product managers adhere to the rough hard, rather pretend that this problem does not exist.
| What do fast user growth?
Data this cashier, while essential, is not the anchor pin for user growth. It can really drive a product's users is growing rapidly, and by what methods? From the perspective of successful products in the history, here are some ideas worthy of note:
First, make a truly outstanding products
To put it bluntly, rapid user growth, the core issue is to do everything possible to make good wine. Product is good is really good, this is fundamental to all products and operations should be pursued. As mentioned above some of the data issues, excellent product manager's intuition and judgment, not findings consistent with short-term data. On the fork, adhere to the principle of product first, data feeds to the second sex, is more reasonable in the long term.
In particular, really good products generally are not based on the analysis of stock market came up, so there is no ready statistics can support. Pokeman recent fires Go, rapid user growth staggering, but I don't think it has established a comprehensive data operating system, even if there are not key to its growth.
Fast user growth in almost all of the great products, all from "good products" within the power, much like Google, Facebook, iPhone, such as taxi and broadcast applications.
Second, find a strategic promotion channels
What is strategic marketing channels? Simply put, are "cheap, full" channel. Many Internet users of the fast-growing, often in foot work. Victorias Secret Case
For example, the rise of cross-border e-commerce, and early they strategically a lot purchased from the Facebook traffic has a lot to do with it. And to this day. Facebook advertising prices have been pushed up a lot, it the opportunity isn't there any more.
In China, many of the fast-growing mobile Internet products non-BAT, most rely on pre-installed channels of this strategy. Manufacturers have pre-installed, operators have pre-installed, solution providers also have preinstalled. Pre-installed in the year cost of one-fifth positive spending today's products, today has grown into a big company.
Undervalued strategic channels keep a keen sense of smell, the rapid growth of the product is essential.
Third, using viral tools
Viral distribution, there are two classes, one is infected and one is a pyramid scheme.
So-called infection was induced and bundled installation through a already has a large installed capacity of products, by means of coercion or even secret operations, bringing another product to the user's Terminal is often more economical way. Even in the ordinary channels of advertising to do good results, for example: cleaning or security products on the move, often directly in advertising intimidating user "your phone memory low" or "your cell phone can be risky", users feel a daisy a tight, was installed.
Pyramid schemes are all familiar. After social networks, MLM-style promotional programme is more convenient. All kinds of variety group fight mode, and forwarded the draw mode, although it cannot be said to be pyramid schemes in the strict sense, the essence of which is to mobilize the masses the masses, this way if used well, will do wonders, who with who knows what.
Four, branding the users cognitive
Community awareness of the importance of this is lacking. Products stand to keep GE Uncle lying like that attitude can also earn money, core user awareness is critical.
In the Internet sector, people often direct communication to Minister creed, insufficient emphasis in brand-building. However, the recent success of companies like oppo, began re-examining the role of branding and user awareness. In my opinion, successful brand building user growth, may be fast and healthy.
Talk so much, it doesn't mean that data management is important, but rather to achieve rapid growth in product, in fact, there are more important things. All of today's Internet practitioners, to master the correct data management methodology, and data do not blindly worship teach religious. A popular and successful products, although the data operations indispensable, but to focus on core "wine" and "moved to the alley" on the two key points.
Lei Feng network Note: please contact the authorized and keep the source and author, no deletion of content.
No comments:
Post a Comment