译文 ：手把手教你如何利用“社交媒体分析学”挖掘潜在客户

本文已获授权转载，转载请注明原文链接和联系原作者；

原创译文 | 手把手教你如何利用“社交媒体分析学”挖掘潜在客户

举例来说，用户可以在YouTube上上传视频内容，然后回复视频的评论来与观众互动。通过社交媒体实现的这种双向交流为很多公司企业提供了宝贵机会，能够让他们与终端用户直接沟通。

社交媒体分析，顾名思义，就是通过分析社交媒体数据来进行商业决策。这些数据通常来自于博客、论坛、社交媒体网站，通常使用文本挖掘和自然语言处理技术，将定性数据转化成定量数据。

常见的社交媒体分析目的包括： 

• 扩大业务

• 通过社交媒体发布广告

• 通过社交媒体监控减少客服成本

• 获取产品和服务的反馈

• 获取公众对某个产品或部门的意见

以下是不同公司使用社交媒体分析的主要形式：

• 火焰检测（听取差评）

• 扩大新产品影响力

• 扩大品牌影响力

• 公司形象维护

• 发掘流行趋势

• 分析内容传播力

• 广告效果的测评

• 开发潜在客户

• 政府听取民意

• 正确的维护好公司的官方社交媒体渠道，利用社交媒体平台分享行业知识和专业信息，来扩大公司品牌影响力和认可度。

• 发现目标客户群体的讨论方向，发现热点，抓住潮流，发现产品和服务的痛点。

• 跟踪本公司品牌和竞品的评价，观察用户的评价口碑。

• 发现竞争对手与客户交流的渠道，找到他们对话的社交平台、博客、论坛或者讨论组。

• 通过网络分析发现缝隙市场的主要影响因素。

• 发现是否有公司内部人员未经公司允许参与外部讨论或者在社交平台发布公司信息。

• 找出以往的和现在的竞争对手，关注他们的动态。

• 发现行业中需要发展和改善的领域。

• 招聘到优秀的人才。

• 找到提高用户忠诚度的正确方法。

• 打造品牌大使社区。

• 发现对公司有威胁的领域。

• 评估社交媒体上不同广告的投放效果。

• 观察客户讨论情况，来对产品进行改善。

• 发现销售机会。

• 比较发现公众更喜欢哪种形式的内容。

关键词提取：首先我们需要在初级输入（primary input）中输入种类条目，找到数据提取需要的不同关键词。举例来说，对高价的组合式橱柜，我们需要知道不同品牌的数量、名称和特点，现在流行的款式等等。同时还需要剔除一些干扰词来去掉无关信息。如果我们仅仅使用“橱柜”这个关键词，就能找到“@XXX – 竞争激烈啊! Brand-YYY @Kitchen_Art #生活方式”，这样的信息就是和组合橱柜无关了。

数据提取和数据筛选：一旦关键词列表确定了，我们就需要确定合理的查询模式来抓取有效内容。为了避免上面提到的问题，我们在输入关键词的同时也要剔除干扰词。当然我们还需要筛选正确的信息源，对组合橱柜来说，可以找Houzz.com论坛等。

将定性数据转化为定量数据：接下来我们需要通过文本挖掘和自然语言处理技术来将定性数据转化为定量数据。

演示如下图：

之后还需要通过“测试&学习”方式对上表进行调整，调整后如图：

购买意向&基本分类创建&调整：分析购买意向时，我们需要以样本数据扫描为基础，创建一个初级分类。

分析语气来判断公众心态，看他们对某个产品是持积极、消极还是中立态度。

这种包含了预测模型和习惯分类法的基于Naïve Bayes分类器的分类方法能够有效提高准确度（>80%）。

语气计算流程如下：

文本挖掘：从每条动态中提取主要概念、创建变量并记录100个左右的概念。

人工情感分类：随意提取10%样本或至少5000个数据点、人工对这些数据进行分类。

预测模型：利用决策树、神经网络，SVM等算法，学习样本数据分类，从而确定分类规则。

习惯规则：基于Business Logic、 Naïve Bayes分类器等条件概率算法，建立习惯规则，提高语气情感分析的准确性。

运用分类规则：将预测算法规则和习惯规则应用到整个数据集，获取每个信息点的语气情感信息。

购买意向分析和语气分析结束后，我们可以将内容分为：

• 高可能性客户

• 中等可能性客户

• 低可能性客户

然后着重分析高可能性和中等可能性用户的需求和痛点，然后制定沟通策略来对他们进行重点营销。

解决方案运作模式 

每一个新的数据产生以后，这个数据就将根据现有规则被分到相应的分类条目中。每三个月应该对这些规则进行再验证。

目前可供市场分析师使用的社交媒体分析工具非常多，如Radian6、Sysomos、Poly Analyst (Megaputer)、HootSuite等等。这些工具可以用来分析多个渠道的数据，也有专门用于分析某个网站数据的分析工具，如Twitter分析工具，Facebook分析工具等。所有这些分析工具都能用于将定性数据转化成数据表格，可用于社交媒体监测。

还有一些使用感较好的统计工具，如R语言、SPSS文本挖掘器、SAS统计分析系统，能够用于预测建模等高级分析过程。Naive Bayes分类器可用于提高情感分析的准确度。

Introduction

Conventional media, such as television, radio or newspapers transmits information only in one direction. Users can consume the information which the media offers, but they have very little or no ability to share their own views on the subject.

Now-a-days, digital mediums has made it possible to have a two-way form of communication that allows individuals to interact with the information being transmitted. This is known as Social media which encompasses a wide variety of online content, from social networking sites like Facebook, Twitter, YouTube etc to interactive encyclopedias.

For example, social video sites like YouTube allow users to share video content and interact through video comments etc. The two-way communication through social media has created opportunities for lots of organization to communicate with their end consumers directly.

Table of Content

What is Social Media Analytics?

How are different companies using Social Media Analytics?

What are the benefits derived from Social Media Analytics?

How can Social Media be used for Lead Generation?

Tonality / Sentiment Analysis

Lead Generation

Tools used for Social Media Analytics

1. What is Social Media Analytics?

As the name suggests, Social media analytics involves analyzing data on social media to take business decisions. Data is usually gathered from blogs, forums and social media websites (like Facebook, Twitter, Youtube etc.). It is then analyzed by converting the qualitative data to quantitative data with the help of different text mining and NLP techniques.

The most common use of social media analytics is to mine customer sentiment in order to support marketing and customer service activities. Typical Social Media Analytics objectives include:

Amplify the business presence

Running campaign through Social Media

Reducing customer service costs with proper Social Media monitoring process

Getting feedback on products and services

Track public opinion for a particular product or business division.

2. How are different companies using Social media analytics?

Most of the organizations today have dedicated social media practice team to help brands expand on and off the web.

Now-a-days, consumer goods manufacturers, personal technology makers and companies that rely heavily on word-of-mouth referrals to generate business find social media analysis tools crucial to their business strategies.

Social media provides retailers with a wealth of information about their consumers that they might never get through traditional media. Since social media creates a two-way interaction between the brand and the individual, retailers can quickly get an understanding of which of their products are favored by buyers, what features of the brand customers can relate with them etc.

Things get really useful when marrying social media data with internal data of the organization. Displaying custom results for example, for one of the online bookstores based on what a visitor had tweeted about that day might push up sales. Perhaps a large percentage of their Facebook followers under the age of 30 are suddenly searching for particular things on your website, raising the possibility of a targeted campaign.

Following are the main use of Social media by different business entities:

Flame detection (bad rants)

New product perception

Brand perception

Reputation management

Trend Spotting

Analyzing virality of the content

Effectiveness measurement of Social campaigns

Lead Generation

Social listening for Government

3. What are the benefits derived from social media analytics?

Social media listening / monitoring process helps in identifying real-time conversations happening on social media about a product, or a brand so you can respond in a timely manner. Since social media has empowered customers to speak about their experience about a product or brand, it is essential for businesses to assess what is going on in the inter webs by carving out a time to listen for brand/product-related conversations. Here are few benefits:

Any organization with proper social media monitoring process, can use Social Media as awareness generation and amplification tool by sharing knowledge or expertise – where and when needed and if appropriate.

Find out what consumers are talking about in the target market. Spot trends, themes / topics and discover pain points.

Track sentiment regarding own and competitor brands, or product. Identify the tone of voice of what others are saying –positive, negative or Neutral etc.

Discover spaces where your customers and your competitors are hanging out. Where is the discussion happening? Which social media platform? What blog? What forum? What groups?

Find key influencers in your niche market through Network and Author level analysis

Find out if anyone from the company may be participating in discussions and is posting information without company’s approval.

Identify old and new competitors and find out what they are doing using competitive landscape analysis. Who’s talking to them? What terms do they use?

Identify areas in the business process that may need to be developed, or needs improvement.

Recruit/hire right people.

Understand the path leads to “customer loyalty”.

Build a community of brand ambassadors.

Determine areas that could be a “threat” to the company.

Evaluate the effectiveness of different marketing campaigns on Social Media.

Improve product features that the consumers are talking about.

Find sales opportunities (lead generation mainly useful for high-cost products).

Check what type of content is performing well

In this article, I will focus on how Social media can be deployed for lead generation. Following are the details.

4. How can Social media be used for lead generation?

With the introduction of social media, humongous amount of unstructured data is produced every second on the internet. This data contains very relevant & useful information about brands, competitors, Industry, consumers’ perception about different product, brands, services, etc.

Now-a-days, social media is the connection between brands and consumers. However, most marketers think of social media as a brand amplification or awareness generation tool (and not sales). But social media is an integral part of today’s sales process which helps to know the prospects and establishing relationships.

For high-cost product marketers (for example, very high-cost modular kitchens), it is all about lead generation. The sales team is more interested in good and high-quality leads than anything else. The extensive reach of social media grants it potential as one of the most powerful lead generation tools, as social media allows sales people to see what prospects are saying about their brand and competitors. But the enormous size and dis-organized nature of the data makes it very cumbersome to generate actionable insights manually. Luckily, we can overcome this situation easily with the help of Analytics.

Social analytics taps and analyzes consumers’ opinions converting them into insights, which helps businesses & marketers in identifying potential leads, areas of customer satisfaction or any customer grievance for a product etc.

Solution Approach

Keyword Generation: To start with one needs primary inputs about the category to get idea about different keywords to be used for data pulling. For example, for premium Modular Kitchen segments we need information like number and name of the brands, features in different brands, prevailing models’ name etc. One also needs to create a list of noise words so that it is easier to remove the irrelevant conversation. For example, for premium Modular Kitchen, if we use only “Kitchen”, it can capture post like “@XXX – all competition! Brand-YYY @Kitchen_Art #TheLifesWay” which is irrelevant for the context of Modular Kitchen.

Data Extraction & Data Cleaning: Once the keyword list is finalized, one needs to formulate the query in proper manner to capture right content. To avoid the problem mentioned in previous example, we need to mention the query in proper order comprising of keywords and noise words (for example Kitchen and not competition or Modular Kitchen & Brand XXX Or Brand YYY etc.).

Also, one needs to select the right sources. For example, for premium Modular Kitchen category, forum like “Houzz.com”, “planned5d.com”might be more helpful than “Generic blogs” etc.

Converting Qualitative data to Quantitative data: Next, we need to convert qualitative data to quantitative data using text mining as well as Natural Language Processing (NLP) based techniques. Below are the examples of converting Qualitative data to Quantitative data:

The taxonomy needs to be fine-tuned based on test & learn approach. For example, the taxonomy for the purchase intent for premium Modular Kitchen looks like:

Purchase Intent & Basic Listening Taxonomy Creation & Fine Tuning: To analyze the purchase intent, one needs to create an initial taxonomy (the science concerned with classification of texts) based on some secondary researchers or sample data scan.

5. Tonality / Sentiment Analysis

Analyze the tonality to understand consumer expectations as well as pain points. Since Social Media data is more inclined towards Neutral content, predictive models alone will not suffice as a classification technique to classify “Positive”, “Negative” and “Neutral” tonality.

An ensemble approach comprising of predictive modeling along with custom classification rules based on Naïve Bayes Classifier would help to achieve higher accuracy (>80%).

Please find below the description of tonality calculation process:

6. Lead Generation

Once the Purchase Intent & tonality analysis are over, we can classify the content as:

High Probability Lead

Medium Probability Lead

Low Probability Lead

We can figure out the Author Name corresponds to High Probability Lead & “Medium Probability Lead” and analyze their needs & pain areas based on the conversation and accordingly design the communication strategy to target them.

Operating Model of the Solution

Every time, there is new data, the data will be automatically classified based on existing rules. It is advisable to validate the rules in every three months.

7. What are the tools used for social media analytics?

There are a variety of social media analytics tools to help marketing experts do all the analysis. Few of them are Radian6, Sysomos, Poly Analyst (Megaputer), HootSuite, etc.  They can be used for multiple channels, while others focus on particular networks such as Twitter/ Facebook etc. All tools are useful for converting the qualitative data into the quantitative format as well as social media monitoring work.

There are also statistical tools like R, SPSS Text Miner, SPSS Modeler and SAS which helps in different advanced analytical work like predictive modeling, Naive Bayes Classifier is used to boost-up the accuracy of Sentiment / tonality analysis.

End Notes

With the help of Social Media Analytics, organizations can identify social leads, influencers and advocates on a daily basis. The potential leads can be segregated into different segments based on the conversation themes and tonality. This “Persona Analysis” helps to understand the demographics and psychographics of the prospects and influencers.

Social Media Analytics can capture profiles of prospects based on the well-defined taxonomy. Generally, the social media experts / analysts verify the leads and sort them into different segments as per business requirement and create personalized communication strategies.

Social media is a powerful way to grow your followers, gain trust, and increase overall revenue by acquiring more customers. Using Social Media Analytics, one can definitely increase the ability to grow to the maximum potential through social networks.

翻译：灯塔大数据；

版权声明：本号内容部分来自互联网，转载请注明原文链接和作者，如有侵权或出处有误请和我们联系。

商务合作｜约稿 请加qq：365242293 。

更多相关知识请回复：“ 月光宝盒 ”；

数据分析（ID : ecshujufenxi ）互联网科技与数据圈自己的微信，也是WeMedia自媒体联盟成员之一，WeMedia联盟覆盖5000万人群。