9种常用的数据分析方法

一、漏斗分析法

漏斗法即是漏斗图,有点像倒金字塔,是一个流程化的思考方式,漏斗分析法能够科学反映用户行为状态,以及从起点到终点各阶段用户转化率情况,是一种重要的分析模型。漏斗分析模型已经广泛应用于网站和APP的用户行为分析中,例如流量监控、CRM系统、SEO优化、产品营销和销售、购物转化率等日常数据运营与数据分析工作中。

9种常用的数据分析方法

漏斗模型除了在电商中应用的比较多以外,在落地页、H5等也应用的比较多。我们可以反复优化落地页当中的图片、文案、布局,进一步的提高整体转化率。

9种常用的数据分析方法

这是一个经典的营销漏斗,图像显示了从获取用户到最终转化为购买的整个过程中的一个子环节。相邻链路的转化率是指使用数据指标来量化每一步的性能。因此,整个漏斗模型首先将整个采购过程划分为几个步骤,然后用转化率来衡量每个步骤的表现,最后通过异常数据指标找出有问题的环节,从而解决问题和优化步骤。最终达到提高整体采购转化率的目标。

[En]

This is a classic marketing funnel, and the image shows a sub-link in the whole process from acquiring users to eventually turning into buying. The conversion rate of adjacent links refers to the use of data indicators to quantify the performance of each step. So the whole funnel model first divides the whole purchase process into steps, then uses the conversion rate to measure the performance of each step, and finally finds out the problematic links through abnormal data indicators, so as to solve the problem and optimize the step. finally achieve the goal of improving the overall purchase conversion rate.

事实上,整个漏斗模型的核心思想可以归类为分解和量化。例如,分析电商的转型,我们需要做的是监测每一级的用户转型,找到每一级的优化点。对于不遵循流程的用户,专门绘制他们的转型模型,缩短提升用户体验的路径。

[En]

In fact, the core idea of the whole funnel model can be classified as decomposition and quantification. For example, to analyze the transformation of e-commerce, what we need to do is to monitor the user transformation at each level and find the optimization points at each level. For users who do not follow the process, specifically draw their transformation model to shorten the path to enhance the user experience.

还有经典的黑客增长模型,AARRR模型,指Acquisition、Activation、Retention、Revenue、Referral,即用户获取、用户激活、用户留存、用户收益以及用户传播。这是产品运营中比较常见的一个模型,结合产品本身的特点以及产品的生命周期位置,来关注不同的数据指标,最终制定不同的运营策略。

二、对比分析法

比较分析是指比较两个或两个以上的数据,分析它们的差异,从而揭示这些数据所代表的事物的发展、变化和规律。它能非常直观地看到事物某一方面的变化或差距,并能准确、定量地表达变化或差距。比较分析可以分为静态比较和动态比较。

[En]

Comparative analysis means to compare two or more data and analyze their differences, so as to reveal the development, changes and regularity of things represented by these data. It can very intuitively see the change or gap in a certain aspect of things, and can accurately and quantitatively express the change or gap. Comparative analysis can be divided into two categories: static comparison and dynamic comparison.

我们知道,孤立的数据是没有意义的,除非进行比较,否则没有区别。例如,在时间维度上,同比和环比、增速、固定基数比、与竞争对手的对比、品类、特征和属性的对比等。比较法能找出数据变化的规律,使用频率高,常与其他方法结合使用。

[En]

We know that isolated data are meaningless, and there is no difference until there is a comparison. For example, in the time dimension, year-on-year and month-on-year comparison, growth rate, fixed base ratio, comparison with competitors, comparison between categories, characteristics and attributes, and so on. The comparison method can find the law of data change, is used frequently, and is often used in conjunction with other methods.

9种常用的数据分析方法

如上图AB公司销售额对比,虽然A公司销售额总体上涨且高于B公司,但是B公司的增速迅猛,高于A公司,即使后期增速下降了,最后的销售额还是赶超。

0、对比分析的价值场景

9种常用的数据分析方法

静态比较:不同总体指标同时进行比较,如不同部门、不同地区、不同国家的比较,也称为横向比较,简称横向比。

[En]

Static comparison: the comparison of different overall indicators at the same time, such as the comparison of different departments, different regions and different countries, also known as horizontal comparison, referred to as horizontal ratio.

动态比较:在相同的总体条件下对不同时期的指标值进行比较,也称为纵向比较,简称纵向比率。

[En]

Dynamic comparison: the comparison of index values in different periods under the same overall conditions, also known as vertical comparison, referred to as vertical ratio.

这两种方法既可以单独使用,也可以组合使用。

[En]

These two methods can be used either alone or in combination.

1.时间维度对比

同一指标在不同时间维度的比较,如比率、环比、固定基比率等。

[En]

The comparison of the same index in different time dimensions, such as ratio, ring ratio, fixed base ratio and so on.

同比是与去年同期相比,可以是季节、月、周、日。

[En]

Year-on-year is compared with the same period of last year, which can be season, month, week and day.

环比是与上一时间段进行比较(也有与下一时间段的比较,也称为事后比较),如本月与上月、本周与上周

[En]

Ring comparison is compared with the previous time period (there is also a comparison with the next time period, also known as post-comparison), such as this month and last month, this week and last week

固定基数比率与特定时间段进行比较,例如2013年和2013年1月的月度销售额。

[En]

The fixed base ratio is compared with a specified period of time, such as monthly sales in 2013 and January 2013.

9种常用的数据分析方法

上图显示的是月度销售额、相同时间范围(全部月度汇总)、相同指数、相同指数含义的对比,整个企业信息的表现,整体性质是可比的。

[En]

The above picture shows the comparison of monthly sales, the same time range (all monthly summary), the same index, the same meaning of the index, the performance of the whole enterprise information, the overall nature is comparable.

2.空间对比

不同空间数据的比较,如北中国和南中国,北京和上海,上海古北店和成都春熙路店。同类空间的比较对象必须是外形相似的,而先进的空间必须与形式相同的优秀空间进行比较,与扩展后的空间的比较,如北京与全国、北京王府井门店与全北京的数据比较,与竞争对手的比较也包括在这份名单中。

[En]

Comparison of different spatial data, such as North China and South China, Beijing and Shanghai, Shanghai Gubei Store and Chengdu Chunxi Road Store. The comparison object of the similar space must be similar in shape, while the advanced space must be compared with the excellent space in the same form, and the comparison with the expanded space, such as the data comparison between Beijing and the whole country, Beijing Wangfujing store and the whole Beijing, and the comparison with competitors is also included in this list.

9种常用的数据分析方法

上图为2018年全年各销售小组销售额对比,其对比的时间范围一致、指标一致、指标含义一致、维度为各个销售小组,具有相同性质。

3.计划对比

与计划标准的比对是销售跟踪的一个非常重要的部分。所有的绩效考核都是计划标准,例如实际实现的销售额与销售计划中达到的金额之间的比较,以查看销售是否完成了原计划,如果没有,原因是什么?

[En]

The comparison with the plan standard is a very important part of sales tracking. all performance reviews are plan criteria, such as the comparison between the actual amount of sales achieved and the amount reached in the sales plan, to see whether the sales have fulfilled the original plan, and if not, what’s the reason?

9种常用的数据分析方法

4.与经验值或理论值对比

经验标准是在大量实践中总结出来的价值,理论标准是从理论中推断出来的价值,平均值是一定空间或时间的平均值。

[En]

The empirical standard is the value summed up in a large number of practice, the theoretical standard is the value inferred from the theory, and the average is the average of a certain space or time.

例如,一个项目的费率:只有一个项目占所有销售收入的比例。参考值小于40%。如果数据超过40%,您需要考虑如何调整策略来帮助客户进行关联购买。如果参考值小于40%,则为理论值。

[En]

For example, the rate of one item: only one item accounts for the proportion of all sales receipts. The reference value is less than 40%. If the data exceeds 40%, you need to consider how to adjust the policy to help customers make associated purchases. If the reference value is less than 40%, it is a theoretical value.

9种常用的数据分析方法

在比较分析中,可以单独使用总指标、相对指标或平均指标,也可以将它们结合起来进行比较。比较的结果可以用相对数字来表示,如百分比、倍数等指标。

[En]

In comparative analysis, total indicators, relative indicators or average indicators can be used alone, or they can be combined to compare. The results of the comparison can be expressed by relative numbers, such as percentages, multiples and other indicators.

在使用对比分析法时,需要先注意以下几个方面

①指数的口径范围、计算方法和测量单位必须相同,即应采用相同的单位或标准进行测量。

[En]

The caliber range, calculation method and measurement unit of ① index must be the same, that is, it should be measured by the same unit or standard.

②对比的对象要有可比性

③比较的指标类型必须一致。无论是绝对指标、相对指标、平均指标,还是其他不同类型的指标,在比较时,双方都必须统一。

[En]

The types of indicators compared by ③ must be consistent. Whether absolute indicators, relative indicators, average indicators, or other different types of indicators, when comparing, the two sides must be unified.

三、聚类分析

聚类分析是一种探索性的数据分析方法。通常,为了更好地理解研究对象,我们使用聚类分析对看似无序的对象进行分组和分类。聚类结果要求组内目标相似度高,组间相似度低。在用户研究中,利用聚类分析可以解决许多问题,如网站信息分类、网页点击行为相关性、用户分类等。其中,用户分类是最常见的情况。

[En]

Cluster analysis is an exploratory data analysis method. Usually, we use cluster analysis to group and classify seemingly disordered objects in order to better understand the research objects. The clustering results require that the similarity of objects within the group is high and that between groups is low. In user research, many problems can be solved with the help of cluster analysis, such as website information classification, web page click behavior relevance, user classification and so on. Among them, user classification is the most common situation.

常见的聚类方法有不少,比如K均值(K-Means),谱聚类(Spectral Clustering),层次聚类(Hierarchical Clustering)。以最为常见的K-means为例,如下图:

9种常用的数据分析方法

可以发现,数据被分到红蓝绿三个不同的簇(cluster)中,每个簇应有其特有的性质。显然,聚类分析是一种无监督学习,是在缺乏标签的前提下的一种分类模型。当我们对数据进行聚类后并得到簇后,一般会单独对每个簇进行深入分析,从而得到更加细致的结果。

四、路径分析

用户路径分析跟踪用户从开始事件到结束事件的行为路径,即监控用户流量,可以用来衡量网站优化或营销推广的效果,以及了解用户的行为偏好。其最终目的是实现商业目标,引导用户更高效地完成产品的最优路径,最终促使用户付费。如何分析用户行为路径?

[En]

User path analysis tracks the behavior path of users from a start event to the end event, that is, monitoring the flow of users, which can be used to measure the effect of website optimization or marketing promotion, as well as to understand users’ behavior preferences. its ultimate goal is to achieve business goals, guide users to complete the optimal path of the product more efficiently, and finally urge users to pay. How to analyze the path of user behavior?

(1)计算用户使用网站或APP时的每个第一步,然后依次计算每一步的流向和转化,通过数据,真实地再现用户从打开APP到离开的整个过程。

(2)查看用户在使用产品时的路径分布情况。

例如:访问电商产品首页后,搜索的用户占多大比例,访问分类页面的用户占多大比例,直接访问产品详情页面的用户占多大比例。

[En]

For example: after visiting the home page of an e-commerce product, what percentage of users searched, what percentage of users visited the classification page, and what percentage of users visited the product details page directly.

(3)进行路径优化分析。

例如,哪条路径是用户访问最多的,在哪一步,用户最有可能失败。

[En]

For example, which path is most visited by users, and at which step, users are most likely to lose.

(4)通过路径识别用户行为特征。

例如:分析用户离开后是有目标的还是漫无目的的浏览。

[En]

For example: analyze whether users are goal-oriented or aimless browsing after leaving.

(5)对用户进行细分。

通常按照APP的使用目的来对用户进行分类。如汽车APP的用户可以细分为关注型、意向型、购买型用户,并对每类用户进行不同访问任务的路径分析,比如意向型的用户,他进行不同车型的比较都有哪些路径,存在什么问题。还有一种方法是利用算法,基于用户所有访问路径进行聚类分析,依据访问路径的相似性对用户进行分类,再对每类用户进行分析。

以电商为例,买家从登录网站/APP到支付成功要经过首页浏览、搜索商品、加入购物车、提交订单、支付订单等过程。而在用户真实的选购过程是一个交缠反复的过程,例如提交订单后,用户可能会返回首页继续搜索商品,也可能去取消订单,每一个路径背后都有不同的动机。与其他分析模型配合进行深入分析后,能为找到快速用户动机,从而引领用户走向最优路径或者期望中的路径。

用户行为路径图示例:

9种常用的数据分析方法

五、帕累托分析

帕累托规则是从经典的28规则衍生而来的。例如,在个人财富方面,可以说世界上20%的人控制着80%的财富。在数据分析中,可以理解,20%的数据产生了80%的效果,需要围绕这20%的数据进行挖掘。在使用2008规则时往往与排名有关,前20%被视为有效数据。28-8方法是专注于关键分析,适用于任何行业。找准重点,找准它的特点,再想想怎么把剩下的80%转化成这20%,提高效果。

[En]

The Pareto rule is derived from the classical 28 rule. For example, in terms of personal wealth, it can be said that 20% of the people in the world control 80% of the wealth. In data analysis, it can be understood that 20% of the data produces 80% of the effect and needs to be mined around this 20% of the data. It often has something to do with ranking when using the 2008 rule, and the top 20% is regarded as valid data. The 28-8 method is to focus on key analysis and is applicable to any industry. Find the focus, find its characteristics, and then think about how to convert the remaining 80% to this 20% to improve the effect.

一般地,会用在产品分类上,去测量并构建ABC模型。比如某零售企业有500个SKU以及这些SKU对应的销售额,那么哪些SKU是重要的呢,这就是在业务运营中分清主次的问题。

常见的做法是将产品SKU作为维度,并将对应的销售额作为基础度量指标,将这些销售额指标从大到小排列,并计算截止当前产品SKU的销售额累计合计占总销售额的百分比。

百分比在 70%(含)以内,划分为 A 类。百分比在 70~90%(含)以内,划分为 B 类。百分比在 90~100%(含)以内,划分为 C 类。以上百分比也可以根据自己的实际情况调整。

ABC分析模型,不光可以用来划分产品和销售额,还可以划分客户及客户交易额等。比如给企业贡献80%利润的客户是哪些,占比多少。假设有20%,那么在资源有限的情况下,就知道要重点维护这20%类客户。

9种常用的数据分析方法

六、公式拆解

所谓公式解析法,就是针对某一指标,用公式对指标的影响因素进行分解。

[En]

The so-called formula disassembly method is aimed at a certain index, using the formula to decompose the influencing factors of the index.

下图分析了一款产品销量不高的原因,用公式法进行分解:

[En]

The reason for the low sales of a product is analyzed in the following figure, which is decomposed by the formula method:

9种常用的数据分析方法

七、A/Btest

A/Btest,是将Web或App界面或流程的两个或多个版本,在同一时间维度,分别让类似访客群组来访问,收集各群组的用户体验数据和业务数据,最后分析评估出最好版本正式采用。A/Btest的流程如下:

(1)现状分析并建立假设:

分析业务数据,确定最关键的改善点,对优化改进进行假设,提出优化建议;比如,我们发现用户转化率不高,我们假设是因为大众化落地页带来的转化率太低。接下来,我们需要找到一种改进的方法。

[En]

Analyze the business data, determine the most critical improvement points, make assumptions for optimization and improvement, and put forward optimization suggestions; for example, we find that the conversion rate of users is not high, and we assume that it is because the conversion rate brought by the popularized landing page is too low. Next, we need to find a way to improve.

(2)设定目标,制定方案:

设置主要目标来衡量每个优化版本的优缺点,设置辅助目标来评估优化版本对其他方面的影响。

[En]

Set the main goals to measure the advantages and disadvantages of each optimized version, and set auxiliary goals to evaluate the impact of the optimized version on other aspects.

(3)设计与开发:

制作2个或多个优化版本的设计原型并完成技术实现。

(4)分配流量:

确定每个上线测试版本的分流比。在初始阶段,优化方案的流量设置可以很小,并可以根据情况逐步增加流量。

[En]

Determine the diversion ratio of each online test version. In the initial stage, the traffic setting of the optimization scheme can be small, and the traffic can be increased gradually according to the situation.

(5)采集并分析数据:

收集实验数据,判断有效性和效果:如果统计显著性达到95%或以上,并保持一段时间,则实验可以完成;如果低于95%,则可能需要延长测试时间;如果统计显著性长期不能达到95%甚至90%,则需要决定是否暂停测试。

[En]

Collect experimental data and judge the effectiveness and effect: if the statistical significance reaches 95% or more and is maintained for a period of time, the experiment can be finished; if it is less than 95%, the testing time may need to be extended; if the statistical significance cannot reach 95% or even 90% for a long time, it is necessary to decide whether to suspend the test.

(6)最后:

根据测试结果,确定是发布新版本、调整分流比继续测试,还是继续优化迭代方案,如果没有达到测试效果,重新开发上线测试。

[En]

According to the test results, it is determined to release the new version, adjust the shunt ratio to continue the test, or continue to optimize the iterative scheme and re-develop the on-line test if the test effect is not achieved.

流程图如下:

9种常用的数据分析方法

八、象限分析

通过对两个或多个维度的划分,以坐标的方式表示期望值。从价值直接到战略,从而进行一些落地推广。象限方法是一种战略驱动的思维,经常与产品分析、市场分析、客户管理、商品管理等联系在一起。

[En]

Through the division of two or more dimensions, the desired value is expressed by the way of coordinates. From the value directly to the strategy, so as to carry on some landing promotion. Quadrant method is a kind of strategy-driven thinking, often associated with product analysis, market analysis, customer management, commodity management and so on.

象限法的优势:

(1)找到问题的共性原因

通过象限分析方法,对具有相同特征的事件进行分析,总结出共同的原因。例如,在上述广告案例中,第一象限中的事件可以提取有效的推广渠道和策略,而第三和第四象限可以排除一些无效的推广渠道。

[En]

Through the quadrant analysis method, the events with the same characteristics are analyzed, and the common reasons are summarized. For example, in the above advertising case, the events in the first quadrant can extract effective promotion channels and strategies, while the third and fourth quadrants can rule out some ineffective promotion channels.

(2)建立分组优化策略

针对投放的象限分析法可以针对不同象限建立优化策略,例如RFM客户管理模型中按照象限将客户分为重点发展客户、重点保持客户、一般发展客户、一般保持客户等不同类型。给重点发展客户倾斜更多的资源,比如VIP服务、个性化服务、附加销售等。给潜力客户销售价值更高的产品,或一些优惠措施来吸引他们回归。

如下图是一个广告点击的四象限分布,X轴从左到右表示从低到高,Y轴从下到上表示从低到高。

9种常用的数据分析方法

点击率高、转化率高的广告,说明人群比较精准,是一个高效的广告。

[En]

The advertisement with high click-through rate and high conversion shows that the crowd is relatively accurate and is an efficient advertisement.

点击率高、转化率低的广告显示,大部分点击的人是被广告吸引的,而低点击率广告的转化率表明,针对广告内容的人与产品的实际受众有些不一致。

[En]

The advertisement with high click-through rate and low conversion shows that most of the people who click on it are attracted by the advertisement, and the conversion of the low-click advertisement shows that the people targeted at the advertising content are somewhat inconsistent with the actual audience of the product.

高转化率、低点击率的广告,说明广告内容符合产品的实际受众,但需要优化广告内容,吸引更多人点击。

[En]

The advertisement with high conversion and low click shows that the advertising content is in line with the actual audience of the product, but it needs to optimize the advertising content to attract more people to click.

点击率低、转化率低的广告可以放弃。

[En]

Ads with low click-through rate and low conversion can be given up.

还有经典的RFM模型,把客户按最近一次消费(Recency)、消费频率(Frequency)、消费金额 (Monetary)三个维度分成八个象限。如下图

9种常用的数据分析方法

九、留存分析

用户留存指的是新会员/用户在经过一定时间之后,仍然具有访问、登录、使用或转化等特定属性和行为,留存用户占当时新用户的比例就是留存率。留存率按照不同的周期分为三类,以登录行为认定的留存为例:

第一种 日留存

每日保留期可细分为以下类别:

[En]

Daily retention can be subdivided into the following categories:

(1)次日留存率:(当天新增的用户中,第2天还登录的用户数)/第一天新增总用户数

(2)第3日留存率:(第一天新增用户中,第3天还有登录的用户数)/第一天新增总用户数

(3)第7日留存率:(第一天新增用户中,第7天还有登录的用户数)/第一天新增总用户数

(4)第14日留存率:(第一天新增用户中,第14天还有登录的用户数)/第一天新增总用户数

(5)第30日留存率:(第一天新增用户中,第30天还有登录的用户数)/第一天新增总用户数

第二种 周留存

每周保留率是指与第一周相比,每周仍在登录的新用户数量。

[En]

The weekly retention rate refers to the number of new users who are still logged in each week relative to the first week.

第三种 月留存

每月保留率是指相对于第一周,每个月仍在登录的新用户数量。保留率是针对新用户的,结果是一个矩阵半报告(其中只有一半有数据),每个数据记录行是对应于日期和列的不同时间段的保留率。在正常情况下,保留率会随着时间的推移而下降。

[En]

The monthly retention rate refers to the number of new users who are still logged in each month relative to the first week. The retention rate is for new users, and the result is a matrix half-report (only half of which has data), and each data record row is the retention rate for different time periods corresponding to the date and column. Under normal circumstances, the retention rate decreases with the passage of time.

以下是以月度留存为例生成的月度用户留存曲线:

[En]

The following is the monthly user retention curve generated by taking monthly retention as an example:

9种常用的数据分析方法

Original: https://blog.51cto.com/u_15668438/5576053
Author: lanxiaofang
Title: 9种常用的数据分析方法

原创文章受到原创版权保护。转载请注明出处:https://www.johngo689.com/500789/

转载文章受原作者版权保护。转载请注明原作者出处!

(0)

大家都在看

亲爱的 Coder【最近整理,可免费获取】👉 最新必读书单  | 👏 面试题下载  | 🌎 免费的AI知识星球