• People Analytics
    马海刚:I时代HR大数据思路与腾讯实践 作者:马海刚 本文由马海刚先生授权HRTechChina发布,转载请注明文章作者及出处 引语: HR管理经过几十年的发展,理论基础仍是工业时代的科学管理经验。近年来面对汹涌而来的移动互联网大潮以及层出不穷各种新的管理挑战,HR管理的理论和方法并没有出现相得益彰的创新内容。唯一引起了广泛关注的HR管理遇上大数据的话题,目前公开的研究也多聚焦在概念阶段,能够真正应用到企业HR管理实践的案例却不多见。本文旨在结合腾讯在HR大数据领域的探索历程,来说说大数据将如何助力HR管理升级,迎接这个崭新的时代浪潮!   I时代,传统的HR将被颠覆,你造吗? 2012年12月12日,CCTV中国经济年度人物评选颁奖现场,万达集团董事长王健林同阿里巴巴董事局主席马云给大家留下了1亿元的赌约。     一年以后同样的颁奖现场,格力电器董事长董明珠和小米科技董事长兼首席执行官雷军在另一个赌约上把赌注提升到了10亿。不同的赌约,相同的内涵——移动互联网将挑战甚至颠覆传统行业。     又一年之后的胡润富豪榜,移动互联网挤掉房地产成为前10大富豪中人数最多的行业。     前段时间,又流行起了一个段子,描述当前苦逼潮人的生活,“每天乘地铁,用小米手机,穿凡客T恤,上3W咖啡听创业讲座,在家看耶鲁大学公开课,知乎果壳关注无数,36氪每日必读,马云的创业史了如指掌,张小龙的贪嗔痴如数家珍。肉夹馍只吃西少爷,约朋友得去雕爷牛腩,喜欢Kindle胜过iPad,手机里没游戏全是GTD的APP。”     实际上,在我看来,这些无一不是因为我们正在迎来人类发展史上一个在量级上可以同工业时代相媲美,但在理念上却与工业时代完全相悖的新时代,我把这个时代简称叫I时代。     起名叫I时代的缘由,是因为我认为这个新时代是一个由internet(互联网)、individualism(个体主义)、innovation(颠覆式创新)聚合而成的时代。这个时代的特征完全颠覆传统经济中的大鱼吃小鱼和快鱼吃慢鱼理论,抱着传统管理思想的企业和管理者将不断面临着生死考验。     这已经是个革命者层出不穷的时代,很多企业通过颠覆式思维,利用先进的技术和跨界的创新,使事情变得更简单,从而完成弯道超越老牌大企业,实现令人难以至信的突破式增长。这也是个更加注重情感链接和用户感观的时代,了解人性、捕获人心成了商业的制胜法宝,也成为管理上的核心要素。     今天的我们已经处在这样一个经济、社会与技术的大变革之中,面临这场变革大潮,HR将会遇到什么样的挑战?我个人的总结是:挑战很多,有两点最关键。     挑战一:I时代下HR管理的理论基础——管理科学将被重新定义。 在这点上我非常认同中国工程院工程管理学部副主任郭重庆院士的观点:“传统的管理将被颠覆,……从管理学界来看,是历史难得的大机遇,大数据是最接近映射真实世界的手段,云计算是社会化配置的计算服务工具,以及无所不在的互联网,开启了管理科学研究的新范式,是管理科学发展史上最接近现代科学的一次机遇,……”在传统的管理被颠覆之际,HR管理是不可能独善其身的。     挑战二:从工业时代过渡到I时代,HR管理研究的方向发生了变化。 通过对比可以发现工业时代和移动互联网时代在组织和人的研究方向上不仅是不同,甚至是完全相悖。 在迎接移动互联网的过程中,HR管理的变革在所难免,那么,我们又该如何应对?     HR,老板喊你转型升级了! 有一点可以确定的是:传统的HR管理已经无法满足变革时代的需求。新时代的HR管理需要转型升级,而转型升级的重点,我建议从三个层面着手:     一是HR组织模式的升级:需要改革传统的按照“选育用留”这种功能模块设置的HR组织模式,打造由COE(Centers of Experts)、BP(Business Partner)、SDC(Shared Delivery Center)共同组成的三支柱模式,提升HR对战略的驱动力,对业务的支撑力,以及对员工的影响力,让HR真正成为企业的变革推动者、领导者、业务伙伴和HR业务专家。     二是HR信息化的升级:HR信息化的目标将不再仅仅是信息化办公或者提升工作效率,而是通过移动端、云、BI等新技术的使用,打造成能够有效连接COE、BP、SDC以及HR所服务的管理者和员工的信息高速公路,促成HR管理的颠覆性创新。     三是HR数据能力的升级,这也是本文我的阐述重点。 当我们所处的环境都被数据化以后,管理决策所依赖的将更多的是数据而不是经验。这也要求HR的数据能力不再是传统的数据统计,而是包括了数据的分析、挖掘、建模、训练、验证、管理改进等一系列的完整活动。类似于谷歌的People Analytics团队、腾讯的活力实验室、人平数据哥这类研究HR的大数据应用的团队将会出现在越来越多企业的HR队伍中,并发挥越来越重要的作用。     如果将转型升级后的HR管理体系想象成一个智能机器,那么组织模式就是机体,信息化是连接机体各个部位的神经网络,大数据就是“大脑”,这三者相辅相成,缺一不可。     那么问题来了,HR大数据挖掘技术哪家强? 搜索一下“HR+大数据”,可以轻松得到几百万条记录,可见大数据在HR领域并不是一个陌生的话题,遗憾的是,热度有余而深度不足。北大光华的穆胜博士在其写的《大数据为何走不进人力资源管理?》一文中提出“HR可能误会了大数据”,这一点我也是比较认同的。HR的大数据需要有自己的玩法,其不同于传统的HR数据分析的功能可以概括为三个方面:     一是养成平台的能力:大数据的特征概括为4V,Volume(大量)、Velocity(高速)、Variety(多样性)、veracity(真实性)。这也决定HR的大数据绝不仅仅是把一些数据拿过来分析,而是一个涵盖数据的产生、存储、抓取、清理、分析、挖掘、建模、训练、验证、呈现的全过程的综合平台。     二是要有连接的效能:与传统的数据分析只需要得出一个数据性的管理结论不同,HR的大数据分析包括了提出概念、分析框架、数据准备、数据清理、数据挖掘、模型创建、训练验证以及管理行动,其过程充分卷入了HR三支柱的COE、BP和SDC,乃至于管理者和员工,其目标是推动HR管理的持续改善。     三是能够牵引HR的方向:传统的数据分析多是事后的总结,是一种滞后的管理。而HR的大数据分析则要求能够帮助HR进行预测,实现前置的管理。     例如传统的人力资源通过绩效管理来识别高绩效的员工并帮助员工持续提升绩效,而在大数据模式下的思路则是通过数据的挖掘找到高绩效员工的特征要素,让企业的每一个员工都能够持续产生高绩效。     由于多数企业在HR的数据领域缺乏规划,要实现上述突破对HR部门而言将是一个漫长而艰难的过程。     HR大数据领域腾讯的实践与探索 腾讯在HR领域的大数据实践最早可以追溯到2012年,通过People Soft搭建起了HR的统一结果库,并开展了第一期的数据清理工作。     而完整意义上的HR大数据体系探索则到了2014年初,在SDC内部成立了HR大数据团队。这里我将从平台建设、连接效能和方向牵引这三个方面简单介绍我们在HR大数据领域的探索经验,希望能够给同样在研究HR的大数据的HR同行们带来思想碰撞的火花。     一、腾讯的HR大数据平台由应用层、功能层以及团队三个部分组成。 1、应用层主要解决HR大数据如何支撑HR业务的问题,阐述的是大数据的应用场景,以及需求如何被响应和落地(如下图所示)。 2、功能层主要解决HR大数据在后台如何运作的问题,阐述的是如何去科学的管理和使用数据,保障数据的质量和价值,包括元数据管理、数据质量管理和逻辑建模规划三大核心模块。     3、从应用层和功能层我们可以看到HR的大数据涉及了HR专业以外的IT系统、数据库、数据分析、产品设计等多个专业,这也意味着仅凭专业的HR是无法搭建起HR的大数据平台的。     以腾讯SDC的大数据团队为例,其成员由SSC、E-HR、区域中心的员工共同组成,是一个拥有人力资源、HR信息化、数据库、HR咨询复合工作经验和背景的团队。     二、在连接效能上我以我们正在开展的某项目举例。 该项目由COE最先提出概念,先后卷入SDC和BP,执行迅速成立了项目联合团队。     其中COE团队负责政策、资源的协调以及专业方向的把控,BP团队负责模型验证以及落地研究,SDC团队则负责数据清理、质量建设、特征挖掘以及模型的搭建和训练。     在这个项目中,不仅COE、BP和SDC的人被连接起来,同时连接的还有对应的“事”和“信息”。     三、在牵引HR的方向上我以腾讯社招候选人稳定性分析为例。 传统的HR数据分析会围绕离职率展开分析,而在HR的大数据分析中则是将腾讯历史上所有的员工按照稳定程度分成多个样本,通过数据的挖掘找到与稳定性相关的典型特征,建立起能够识别候选人稳定性的数学模型。     其目标之一是希望通过应聘者的简历自动对其稳定性给出评估建议,也为后续招聘以及保留环节提供参考。     在此,还有几点建议给到准备进行HR大数据探索的同行们: 一、从现在开始,夯实数据基础。 以腾讯的某个HR大数据项目为例,一次调用的数据就超过了600万条,400多个字段,一般的PC机以及excel、spss等工具都无法支撑此种量级的数据挖掘,但是其量级又达不到使用TDW的程度,加上数据敏感性等诸多因素,最终发现需要搭建用于HR大数据分析的服务器。     二、数据质量决定数据的价值。 涂子沛在《大数据》一书中用了整整一个章节来阐述数据质量,足见数据质量的重要性。在此我想用一句话来补充说明:在一堆错误的数据中,你能指望得出正确的分析结果吗?     三、是挖掘数据而不是统计数据。 仅从统计学的方法上看就可以看到差别,传统的HR数据分析用的最多的统计方法就是描述统计、箱型图等。     但是到了HR的大数据分析,相关性分析、方差分析、回归分析、聚类分析、决策树模型等用的会更多。其原因就像维克托.迈尔-舍恩伯格在其《大数据时代》中强调的,大数据研究的“不是因果关系,而是相关关系。”     对于企业的HR而言,当HR遇上大数据,我们更应该抓住这个机会,在大数据平台能力,连接的效能,牵引HR方向这三方面寻求突破,进行创新性的研究和探索,提升HR之于企业的价值和影响力。     最后,借用狄更斯的名言“It was the best of times, it was the worst of times”,I时代带给HR的不仅仅有挑战,同样也有机会。     亦如郭重庆院士所言,“管理学界应该抓住这个机会,实现自己的历史使命和担当。”
    People Analytics
    2014年10月30日
  • People Analytics
    HR Data Analytics – Case Use by HR Organziations 作者:William Chin 授权发布 The Chapman Consulting Group just completed their Spring HR leaders meeting in Beijing on May 15. This time Lenovo hosted the session at their Beijing headquarter office. The topic for this round is centered around Managing Global HR in the age of ‘Big Data’ What companies are doing to optimise talent and HR systems in parallel with the advent of global and regional Centres of Excellence; The increased use of data and analytics as another tool of Global HR management; and The effect this is having on the type of HR Leader progressing within the profession.  This theme is consistent with their #1 trending HR focus areas for 2014. I have captured key points from the meeting below.     Lenovo – Shared Services Lenovo, the world’s leader in the PC industry, had just implemented a global HR system, making the switch by eliminating several disparate systems into a global solution. While they have done all the requisite change management requirements with organisation stakeholders, they are seeing that people still like to do things the “old way.” How true! People hate to change. While Lenovo made a clear stand that all everyone need to adopt and utilise the new system they do have a VIP process for their top key executives. The VIP allows for telephone hotline and/or email communication to a HR professional for assistance. However, everyone else is expected to utilise the new self-service model. The benefit of going global with their new HR system is now they have the ability to manage their workforce under one roof. Previously, HR was unable to access “real time” data and instead, was managing people with spreadsheets.   Pfizer – Improving Retention Employee retention is a huge risk in the pharmaseutical industry in China. Industry average is around 25-30% turn-over each year. Pfizer is the global largest pharmaceutical player and is also the largest in China. Even Pfizer is not immune to the high turnover rate. In fact, competitor companies target their employees, because they are the largest. To combat turnover and improve retention they turned to “big data” to better understand drivers of turnover – they created an employee profile of turn-over drivers. The profile Pfizer developed is employee specific with a “risk score.” Pfizer partnered with a consulting company to develop the analysis tool combining existing employee data and against employees who left the company. By looking at former employee profiles they then were able to map those to existing employee enabling Pfizer to see trending issues that may cause turnover. Seeing this information ahead of time allows HR to partner with BU leads to take proactive actionable steps. Some examples of high risk dimension include: employees where are a rehire (they have already left once), short-time with a manager (have not developed a strong bonding with the direct manager), and long tenure in a role (it’s time to refresh with a new focus). I was thinking these are all indicators of high risk turnover by itself. So, why do you need to do a study? The genius is that employee turnover is multi-dimension. Not one thing by itself are drivers of turn-over but, by combining all the various turnover drivers and employee profile, you begin to see a multifaceted profile of their employees – HR and BU can then take multiple tracks to drive retention.   Qualcomm* – Use a Data Analytics Qualcomm has a dedicated data analytics team. That team started in 2008 and was a small group who was responsible for generating large HR data but, on spreadsheet format. Over the years, Analytics Team went on to focus on benchmarking to creating data visualisation and now focusing on predictive modelling for the company. Qualcomm human resources has the ability to pull up dashboard data an a click of a button. This is information is globally accurate and with the ability to do drill downs by organisation, business function, geography and employee types etc. This enables Qualcomm HR, at all levels – HRVP to HR specialist, to have the same data points, at any time. The analytics team also conducts research projects analysing the success of a merger and acquisitions project. The team created a social network analysis / model indicating the strength on network and social ties. In a M&A, one would typically want to see the newly acquired company integrate into the overall company. The faster employees integrate the greater the success outcome. Creating such a model allows Qualcomm to analyse and visualise social interactions to gain insight on who were the “bridge builders,” those who were the best at helping with integrating after a merger.   Doosan – Don’t Over Do It With Data Doosan is a Korean-based conglomerate. The HRVP reminded us that sometimes over use of data can be detrimental to business decision. Instead of using judgement, managers often ask project analysts or HR for more data to help with their decision making. With the data provided, business will ask for more next level data, to back up the high-level data. Analyze the data, analyse more data, then the data paralyses you. By the time the data is complete that the information is out of date and decision window is closed. How often have we faced this before? The presenter was so right on with this point. In HR, we also have metrics and data to measure our performance. The roundtable participants all have HR KPI scores they manage to. One hotelier HR said that after a HR systems implementation that their HR satisfaction scores dropped. I thought that after any large project implementation that one would expect a drop (remember that people hate changes). Instead of managing to the dropped score HR should be managing to improving the score and maybe, that scoring criteria will be different from the prior standard but, the processes and systems have changed. Doosan further explains that in a business downturn, for example, HR is expected to manage employee reductions. So, what happens if HR is successful with meeting the employee reduction targets and morale KPIs are on track yet, the business continues to decline. Doosan HR further suggests that we use experience and judgement to help the business. Data is only one part of the story.   This wraps the summary of key points by each presenter. The Chapman Consulting Group always does a good job with brining together a group of HR leaders from various industries for sharing and networking.   作者声明: *I am employed by Qualcomm. However, the information contained within is the opinion of the author and not that of my employer. All company and/or product names may be trade names, trademarks and/or registered trademarks of the respective owners with which they are associated. Furthermore, this blog post does not represent an endorsement of their products and services and I have woven my own experience in this post. This is for informational purposes only. There is no representations as to the accuracy or completeness of any information.
    People Analytics
    2014年05月17日