Data thinking the WHAT WHY thinking skills of data analysis HOW how to exercise analytical skills in business time three core thinking structured pyramid way of thinking central argument select the top of the pyramid, it can be assumed that the problem is predicted, the reason structure dismantling from top to bottom, the layers of the core point component dismantling argument, or as a causal dependence between the upper and lower MECE independent, completely exhausted. Avoid overlap and duplication between the arguments, sub-arguments to try to improve verify whether the core of the arguments or sub-arguments, should be quantifiable, with the data speak. They must be verifiable eg split this layer by layer under the net, can make use of mind mapping, xmind and other sales inside the consumer area region A sales price sales area B Time External Market competition in the market capacity of the policy changes will be arguments and summarized the arguments and progressive dismantling of the arguments developed and complemented Cons: structured analysis is thinking, but it is not enough data, but it is inevitable shortcomings divergence formulation everything quantify structural Jieke (can be discount + - * / ) the smallest indivisible eg sales made of? - sales volume and unit price multiplied by the profit made of? - Sales revenue and costs are subtracted sales area A and sales per capita sales number of people buying new customers and old customers the price of the original price of the rest of the business of with a structured and formulaic dismantling, in the final analysis the arguments get, very often, is a phenomenon. Data is some reflect the results, but do not represent reason. Only understand the business, in order to establish business data model Summary: The three core thinking is a structured guidance, practical application tools should help thinking skills, to achieve powerful chain of effects. Data analysis thinking skills quadrants X axis, Y axis is completely customizable. eg: customer value, customer churn. RFM quadrant method is a policy-driven thinking. Using a wide range, strategic analysis, product analysis, market analysis, customer management, user management, can be intuitive, clear, artificial data division, the division's results can be directly applied to the policy. Comparative Method good data index, or the ratio of a certain proportion. Good data analysis, will be used in comparison. Solitary number does not permit competitors contrast comparison category features and attributes contrast time contrast-year comparison transformation before and after contrast change ... contrast method is a way of thinking data mining law. antitheses can find a lot of data to see the law, which can be combined with any thinking skills, such as multi-dimensional contrast, contrast quadrant. More contrast is a habit, is a dead end data analysis, a qualified analysis, be sure to use the N-th comparison. Law funnel funnel method is a process-oriented way of thinking. We did not use a single funnel analysis, conversion of 20 % , but what can explain it? It is to be combined with other analytical thinking, such as multi-dimensional, contrast, to illustrate the problem. Twenty-eight law sustained attention TopN data. Although many indicators, but some indicators are often more valuable, Pareto rule not only able to analyze data as well as data management assumptions law many times, when the data analysis can be no explicit reference data, such as entering a new market, open up a company kind of product, the boss let you forecast sales a year later. After eg commodity price increases, earnings will not change? After assuming that commodity price increases, sales will fall. The problem is that the number of sales fell? First, it assumes that there will be no change in traffic flow and channel marketing positively correlated with commodity prices affect the conversion rate. So now to determine the volatility of the conversion rate. Found conversion usual, say 5 % . After the conversion price changes in estimates. Assuming that all types of users for different price sensitivity, so users into loyal type, ordinary type, wool type. Loyalty, conversion rates change is very low, almost no wool transformation. . These assumptions can be made through experience, the final summary. Hypothesis-driven thinking is a way of thinking inspiration. Index (indicators) multidimensional method Analysis of indicators classic business model did not move, advance indicators. If you can not measure it, you can not grow it. Marketing indicator product performance indicators of user behavior metrics e-commerce indicators flow indicators on how to establish business analysis framework from the perspective of indicators from the business point of view from the perspective of the process of marketing model potential customer conversion opportunities for customer conversion rate of new customer conversion fee ... AARRR model user behavior model of e-commerce model