7/11/2014 Forget the CV, data decide careers - FT.

com

By continuing to use this site you consent to the use of cookies on your device as described in our cookie policy unless you have disabled them. You
can change your cookie settings at any time but parts of our site will not function correctly without them.

Home World Companies Markets Global Economy Lex Comment Management Life & Arts
Business Education Entrepreneurship Business Books Business Travel Recruitment The Connected Business Tools

Welcome to the Financial Times, news and analysis for global decision makers.
Register now and receive 8 free articles per month.

July 9, 2014 9:14 pm

Forget the CV, data decide careers

By Tim Smedley

AFP

Automatic for the people: algorithms are filling jobs, such as those in call centres

I no longer look at somebodys CV to determine if we will interview them or not, declares Teri Morse, who oversees the recruitment of
30,000 people each year at Xerox Services. Instead, her team analyses personal data to determine the fate of job candidates.

She is not alone. Big data and complex algorithms are increasingly taking decisions out of the hands of individual interviewers a trend
that has far-reaching consequences for job seekers and recruiters alike.

The company whose name has become a synonym for photocopy has turned into one that helps others outsource everyday business
processes, from accounting to human resources. It recently teamed up with Evolv, which uses data sets of past behaviour to predict
everything from salesmanship to loyalty.

For Xerox this means putting prospective candidates for the companys 55,000 call-centre positions through a screening test that covers a
wide range of questions. Evolv then lays separate data it has mined on what causes employees to leave their call-centre jobs over the
candidates responses to predict which of them will stick around and which will further exacerbate the already high churn rate call centres
tend to suffer.

The results are surprising. Some are quirky: employees who are members of one or two social networks were found to stay in their job for
longer than those who belonged to four or more social networks (Xerox recruitment drives at gaming conventions were subsequently
cancelled). Some findings, however, were much more fundamental: prior work experience in a similar role was not found to be a predictor
of success.

It actually opens up doors for people who would never have gotten to interview based on their CV, says Ms Morse. Some managers
initially questioned why new recruits were appearing without any prior relevant experience. As time went on, attrition rates in some call

http://www.ft.com/intl/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html#axzz379n7ZGoM 1/7
7/11/2014 Forget the CV, data decide careers - FT.com
centres fell by 20 per cent and managers no longer quibbled. I dont know why this works, admits Ms Morse, I just know it works.

Organisations have long held large amounts of data. From financial accounts to staff time sheets, the movement from paper to computer
made it easier to understand and analyse. As computing power increased exponentially, so did data storage. The floppy disk of the 1990s
could store barely more than one megabyte of data; today a 16 gigabyte USB flash drive costs less than a fiver ($8).

It is simple, then, to see how recruiters arrive at a point where crunching data could replace the human touch of job interviews. Research
by NewVantage Partners, the technology consultants, found that 85 per cent of Fortune 1000 executives in 2013 had a big data initiative
planned or in progress, with almost half using big data operationally.

HR services provider Ceridian is one of many companies hoping to tap into the potential of big data for employers. From an HR and
recruitment perspective, big data enables you to analyse volumes of data that in the past were hard to access and understand, explains
David Woodward, chief product and innovation officer at Ceridian UK.

This includes applying the data you hold about your employees and how theyve performed, to see the causal
Members of one or
links between the characteristics of the hire that you took in versus those that stayed with you and became two social networks
successful employees. Drawing those links can better inform your decisions in the hiring process. stayed in their job longer
than those belonging to
Data sets need not rely on internal data, however.
four or more

Social media data now gives us the ability to listen to the business, says Zahir Ladhani, vice-president at
IBM Smarter Workforce. You can look at what customers are saying about your business, what employees are saying, and what you
yourself are saying cull all that data together and you can understand the impact.

Most recruitment organisations now use social media and job-site data, says Mr Ladhani. We looked at an organisation which had very
specialised, very hard to find skill sets. When we analysed the data of the top performers in that job family, we found out that they all hung
out at a very unique, niche social media site. Once we tapped into that database, boom!

Ceridian, too, has worked with companies to effectively scan the internet to see what jobs are being posted through the various job
boards, in what parts of the country, says Mr Woodward. If youre looking to open a particular facility in a part of the country, for
example, youll be able to see whether theres already a high demand for particular types of skills.

Experts appear split on whether the specialisation required for executive recruitment lends itself to big data.

I hire 30,000 call-centre people on an annual basis we dont hire that many executives, says Ms Morse, adding theres not enough
volume. However Mr Ladhani disagrees, believing that over time the data set an organisation holds on senior management hires would
become statistically valid.

As more companies start to analyse their employee data to make hiring decisions, could recruitment finally become more of a science than
an art?

The potential is clearly much greater now than ever before to crunch very large volumes of data and draw conclusions from that which
can make better decisions, says Mr Woodward. The methods and computing power being used in weather forecasting 10 years ago are
now available to us all...who knows where this may go.

It is a trend worth considering to get your next job, perfecting your CV could well be less important than having carefully considered the
footprint you leave in cyberspace.

-------------------------------------------

Case study: Demographic drilling down helps LV= recast recruitment ads

Kevin Hough, head of recruiting at insurance firm LV=, was a pioneer of Big Data before he had heard the term.

The insurer has 5,800 employees across 17 UK sites. A year ago, the question of where best to target its recruitment advertising provided
an innovative answer. We took all of the postcode data from our HR database where our current staff live and where they are located
and split that by their [seniority] explains Mr Hough. Using software called Geo-Maps, which works similarly to Google Maps, we could
zoom in and out of clusters of our people to see where they are willing to travel from to get to work.

On top of that we were able to overlay where people were applying from, taken from our recruitment data. The next iteration was where
our Facebook and LinkedIn followers are from: the demographic of people following us, what interests them, the diversity and inclusion

http://www.ft.com/intl/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html#axzz379n7ZGoM 2/7
7/11/2014 Forget the CV, data decide careers - FT.com
breakdowns, male and female and so on, on which social media platform.

We were able to build up a profile of who our successful candidate was, the age demographic, and where people were hunting for jobs.
What was really interesting was the reach some of our advertising was having, and more importantly some of the gaps.

Recruitment advertising at LV= has subsequently been redesigned. The investment and expertise required to analyse the data were
negligible, says Hough. For Geo-Maps it was literally as basic as getting a licence for the software and uploading our data to it from a
spreadsheet. Sometimes with all the clever systems that people have in organisations you can be blinded to the simple, raw data that is sat
there.

Next, LV= will add performance review data.

The ongoing piece is to say of that group we recruited a year ago, which are still there? It helps to shape not only how we attract the
people, but will even start to shape some of the roles themselves.

RELATED TOPICS Big data, UK employment

COMMENTS (30)

Sign in

Submit Comment

By submitting this comment I confirm that I have read and agreed to the FT Terms and Conditions. Please also see our commenting guidelines.

Newest | Oldest | Most recommended

Percy Street 1 hour ago

When algorithm meets public policy there are some very interesting problems. If certain types of discrimination in hiring are illegal, should policy be to wait to
see the actual outturn of the algorithm? Or should the algorithm be published? Or, somehow, certified by a responsible authority? How is the legal system,
how are courts planning to cope? Are they?

Note that the use of algorithms for hiring, lending, insuring, security purposes is--in fact--entrenching a new authoritarianism. There is no one to reason with
when the computer says no. Responsibility becomes diffused in such a way that senior management--is this surprising?--escapes blame. Is this what we
want?

Recommend Reply

cockney dave 1 hour ago

Traditional "success analytics" will never tell you that the guy who is applying from Stevenage for a job in Blackpool is doing so because that is where his grand
children live.

One way or another these methods homogenise the workforce and discriminate against INDIVIDUALS with non-standard motivations, backgrounds and
desires.

In a rapidly changing workplace more diversity gets you more flexibility. Homogenised workforces dont innovate much as they think and do the same.

2 Recommend Reply

Carola Hoyos, Recruitment Editor FT 1 hour ago

This certainly has struck a nerve, not just with me. Given how quickly things move in this space, we should look at revisiting what is happening in a few months.
Many thanks for all your interesting comments, they have added valuable insights around the practicality, morality and reality of this trend.

Recommend Reply

Econ. 2 hours ago

http://www.ft.com/intl/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html#axzz379n7ZGoM 3/7
7/11/2014 Forget the CV, data decide careers - FT.com
This big data approach may only work if you view labour as machines.

Recommend Reply

O.J. 3 minutes ago

@Econ. I don't think that call centre workers are mere machines. I have to say that being answered by a call centre worker who can understand and
listen is far far better than Siri or any other Amazon-like person.

Recommend Reply

asmyth28 3 hours ago

Stick your databases up your filthy slave driving holes you dimwitted cockroaches. Im on record.

1 Recommend Reply

wilovitch 4 hours ago

'Big Data' is, simply put, the analysis of the masses of data accumulated since data can be digitalized and stored, and transforming this data into some sort of
meaningful information to improve decision making.

Being the analysis of existing data, 'big data' can tell us a lot about what happened in the past and why. It can eventually help us understand why we are where
we are today. But what does it tell us about the future ? Nothing.

I'm still with the profound belief that the success of a business lays in anticipating its own needs based on its customers' wants and needs. Looking at past and
today's patterns to determine patterns of tomorrow, consists of looking at the future in the rearview mirror - even for recruiting in call centers which are one of
the biggest plagues I know.

'Big Data' is yet another form of robotization - and a big revenue generator for the IT industry - that could easily be used by managers as a shield for hapless
decisions. And it is a tricky one because it may contribute to erode our capacity of thinking critically.

2 Recommend Reply

prateet thai 9 hours ago

Execs would like to replace all call centre jobs themselves by machines; hence the screening (press *1 now) process even before you get to the computer
selected human. And most call centres are located in garden spots like North Dakota, the Philippines or Bangalore. (See excellent movie, "Outsourced") So, the
zip code search seems a definite starting point (duh!). (Bismark North Dakota is 58501).

This is an interesting article, but it only serves to show how expensive/un-efficacious are low end functions (like the usually hapless HR department). The
article fails to show the economics of human vs machine processing, or even mention the cost of interviews. It must mean that the value of a good call centre
operator is a)hard to determine, and therefore, b)worth less than the price of x number of interviews. Dig deeper, Smedley - they'll be screening your
replacement by computer soon enough.

1 Recommend Reply

lemon squeezy 12 hours ago

Recruiters may have found another way to discriminate. Understandably enough, it is completely rational for them to do so, even if everyone starts from an
equal position or the most complex algorithms are being used. The question is, which social group will be the target of all discriminations in this case?

Recommend Reply

JenNOLA 17 hours ago

They have a relatively unbiased data set now - it includes candidates who became both successful and unsuccessful employees. Over time, they will only hire
from the attributes of the successful bunch, so the data will become more uniform and the unusual bits will be more random (i.e., not good predictors). They
will need to hire small groups with the unsuccessful attributes and compare those to a control group with the successful attributes. That's the only way to keep
the data working for them over time.

2 Recommend Reply

http://www.ft.com/intl/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html#axzz379n7ZGoM 4/7
7/11/2014 Forget the CV, data decide careers - FT.com
monnoo 14 hours ago

@JenNOLA no, that's wrong. It is the usual misunderstanding, caused by applying "over-generalization" and the (inappropriate) assumption of
overall (space & time) stationarity. There always will be successful and not successful items, because enterprises develop, the expectation of
"successful" changes, because not all employees can behave the same way. All this and more is leading to the necessity of continuous remodeling.
Models are not laws, particularly not in the domain of the social.

2 Recommend Reply

DavidHS 12 hours ago

@JenNOLA Please define "successful."

I know of no field, possibly except for professional sport, in which performance, can be assessed objectively or can be attributed to an individual given
the power of social influence.

Recommend Reply

DuchessofTesco 4 hours ago

@JenNOLA successful or unsuccessful based on what criteria? All that happens is the prejudices and consequent misjudgements of recruiters
about prospective employees are programmed into the process of unnatural selection that now dominates the job market. It is not such a recent
phenomenon either but has operated for years. It goes a long way to explaining the absence of competence in the workplace at every level and across
a wide variety of industries. It also explains the frustration of many, many people in the work place who see ways that performance / sales could be
improved but are ignored because "computer says 'no'". Initiative and good judgement once sought after at work are now too risky. Better to keep
your head down or nod in agreement even if you know what is being proposed is rubbish e.g. subprime.

Recommend Reply

andrewguest 17 hours ago

"Computer says 'No'".

4 Recommend Reply

Paulus4701 18 hours ago

A behavioural filter being applied, based on social and economic circumstances will basically determine your employment future. A good reason not to go
bonkers on social media (this is my last comment ever)

Recommend Reply

Beefeater1980 20 hours ago

I think that at the very basic level this is a good thing in that it replaces subjective judgements with objective ones and probably results in more accurate
information being used as the basis for decisions. That said, I wonder to what extent the algorithms compensate for the fact that algorithms are being used
(and therefore diminish the potential hiring pool). If the algo designers get the causal relationships wrong then they are excluding combinations of
traits/experience in candidates that might be more effective than the combinations thrown up as ideal by the filtering.

Also, at what point do the data become extensive enough that they can be used to determine promotion within the HR function?

2 Recommend Reply

Old School Canuck 21 hours ago

It all goes so well with the financials when algos rule the roost... Do all staffing professionals (everywhere) believe that number-crunching will be the next "cure
all" approach to dealing with the flood of CVs that plague many corporate and government personnel recruiters?

Recommend Reply

MarekRiacek 21 hours ago

@Old School Canuck Can this really be so much worse than what HR is doing now? Our HR people at least are so clueless that random selection
seems attractive in comparison. Headhunters will never propose anyone other than a candidate who has already done the exact same job elsewhere,

http://www.ft.com/intl/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html#axzz379n7ZGoM 5/7
7/11/2014 Forget the CV, data decide careers - FT.com
so you hardly ever get fresh ideas, which is why you recruit from outside your network in the first place.

7 Recommend Reply

Firdaus Abbasi 21 hours ago

Deciding by algos'!? On one hand we are told that the brain is too complex. So I ask whether like in history there were the alchemists, now recruiters of the
future are becoming like Estate Agents or the Bankers of recent past; fat cats and too lazy to use head.... Conformity and bland!

Recommend Reply

Global Citizen 22 hours ago

Is there a danger that this turns into discrimination against certain groups?

Recommend Reply

compayee 21 hours ago

@Global Citizen

You mean certain groups that do not fit the 'ideal profile', which shall by now be established in advanced?

3 Recommend Reply

Carola Hoyos, Recruitment Editor FT 21 hours ago

@compayee @Global Citizen - Yes. Though that might not be such a change - humans are already overtly and unconsciously discriminatory when
it comes to recruiting without the 'help' of the algorithms they write.

4 Recommend Reply

Econ. 2 hours ago

@Carola Hoyos, Recruitment Editor @compayee @Global Citizen Would it be appropriate to conduct alog audits to check whether companies
are filtering out racial groups?

Recommend Reply

JenNOLA 17 hours ago

@Global Citizen The discrimination will be for and against certain personality types. Not a problem - there already exist personality traits that make
for better salespeople in every race on earth. This will give ALL potential salespeople the chance to sell those cars. Likewise for analysts - all races and
cultures certainly produce members who are good at figuring out what is going on. Now it will be easier to find those analysts among a group of
applicants rather than being drawn to one because of the college attended (same alma mater as the CEO!) or "smart-sounding" name ("J.W. Thorpe
III" or "Daniel Steinberg" vs "Tiffany Jones").

Now if we can just get rid of CVs altogether!! List your university degree(s) and describe your work experience in concrete terms for the past 10 years.
Done!

Recommend Reply

Carola Hoyos, Recruitment Editor FT 22 hours ago

stopatlngme - hmmm. I'm not willing to call BS. I think this is just the start. and agreed with stolto, there's a lot to do in the space and it is becoming more
sophisticated quickly

1 Recommend Reply

Carola Hoyos, Recruitment Editor FT 22 hours ago

RDD - good point re CV, wish I could say I thought of that. Grrr. All this makes my head hurt like after watching a very fast doubles volley rally.

http://www.ft.com/intl/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html#axzz379n7ZGoM 6/7
7/11/2014 Forget the CV, data decide careers - FT.com
Recommend Reply

Carola Hoyos, Recruitment Editor FT 22 hours ago

contracts - yes, struck me too. Raises the question about how accurate it will be.

Recommend Reply

stopeatingme 23 hours ago

It's for a call centre position. Jobs that were always seen as expendable, basically. I mean, how many of these 'staff' were temporary workers recruited through
agencies anyway. I call BS.

Recommend Reply

stolto 1 day ago

Big data and recruitment. Surely there is a lot to do there. But nothing related to the silly pseudo-correlations that are described in the article. It is not surprising
to see companies like Xerox apparently using the most simplistic and nave models to filter and assess candidates; it just shows how low this company went.

Depressing...

2 Recommend Reply

contracts 1 day ago

The methods and computing power being used in weather forecasting 10 years ago are now available to us all...who knows where this may go.

Hmm, Ironic?

2 Recommend Reply

Printed from: http://www.ft.com/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html

Print a single copy of this article for personal use. Contact us if you wish to print more to distribute to others.
THE FINANCIAL TIMES LTD 2014 FT and Financial Times are trademarks of The Financial Times Ltd.

http://www.ft.com/intl/cms/s/2/e3561cd0-dd11-11e3-8546-00144feabdc0.html#axzz379n7ZGoM 7/7