Job Boards = Bad Candidates? Don’t believe the hype.

Posted at October 5, 2008

As stated in a previous post, I continue to see well respected thought leaders in the staffing industry make claims that the value of the job boards is waning and that the quality of candidates on the job boards is low. Not long ago I weighed in on an ERE discussion in response to the question of, “What would happen if the job boards became obsolete?” I noticed that many people in the discussion took the stance that the quality of candidates on the job boards is low. Is it just me, or don’t these types of statements reek of stereotyping?

Wikipedia states that a stereotype can be a conventional and oversimplified conception, opinion, or image, based on the assumption that there are attributes that members of the “other group” (in this case, job board candidates) have in common. Stereotypes are sometimes formed by a previous illusory correlation, a false association between two variables that are loosely correlated if correlated at all. Illusory correlation is the phenomenon of seeing the relationship one expects in a set of data even when no such relationship exists, and when people tend to overestimate a link between two variables. However, the correlation is often slight or not at all.  

Let me be very specific in that when I talk about the job boards – I am referring only to their resume databases.  We’re all painfully aware that the majority of respondents to online job postings are not spot-on (or even close) matches. 

I believe that the job board resume databases have a cross section of all candidates – some would argue a normal distribution (bell curve) and I would agree. From what I can tell, it seems to pretty much be a statistical inevitability. You’ll have a small percentage of horrible candidates, a large percentage of average candidates, and a small percentage of top-notch talent. The same is true of any company at any one point in time – so if you called through a company directory, you’d likely hit the same statistical inevitability: some bad, lots okay/good, some great. I’ll argue that the same is true of Internet sourcing of every type.  

Here are 3 points to think about before saying or believing that the job boards have poor quality candidates:

#1 Statistics

I am definitely not an expert on statistics, but I would argue that the people who enter their resumes into the job board databases are a random sample of the total job seeker population.  With some help from Wikipedia to help me concisely explain these points, a random sample is one chosen by a method involving an unpredictable component (which is fair to say in this case, because who can argue that we can accurately predict the subjective and objective “quality” of people who post their resumes online?). The sample will usually be completely representative of the population from which it was drawn – in this case, job seekers. In the case of random samples, mathematical theory is available to assess the sampling error. Thus, estimates obtained from random samples can be accompanied by measures of the uncertainty associated with the estimate. This can take the form of a standard error, or if the sample is large enough for the central limit theorem to take effect.

The Central Limit Theorem (CLT) states that the sum of a large number of independent and identically-distributed random variables will be approximately normally distributed (i.e., following a Gaussian distribution, or bell-shaped curve, or “normal distribution”) if the random variables have a finite variance. What this all means is that in statistics, it’s generally accepted that if the sample is large and taken at random (selected without prejudice), then it quite accurately represents the statistics of the population, such as distribution probability, mean, standard deviation, etc.

Most of the major job boards claim to have 20M+ candidates in their resume databases – that’s a pretty LARGE sample of the job seeking population, so hopefully you can see where I am going with this. Additionally, I could reference the Law of Large Numbers, which if you boil down all of the technical statistics-speak when you look it up, basically says that the larger the random sample size, the more likely that it “guarantees” stable long-term results for random events. “Stable” results in our case would be that the majority of candidates on the job boards are “average” – with fewer horrible “undesirables” and fewer “A” candidates (see the bell curve coming?).

#2 The candidate’s perspective

And now for the very unscientific side of the equation…why do people post their resumes online? From the perspective of a non-staffing job seeker, many people see the job boards as an online marketplace, not unlike eBay. Most people who are not in the staffing industry and who are not perennially looking for a job don’t view the major job boards with disdain. If a job seeker relies solely on searching job postings online, they are being proactive in seeking employment, but they are reliant on the reactive response of the firms they reply to – and let’s be honest – most candidates do experience the “black hole” effect when they respond to job postings (auto-responders don’t count here). This can lead many candidates to seek to take more control over the process and be actively sought out by opting to post their resume into a resume database so they can be actively found and pursued by potential employers – kind of like posting something on eBay so that people looking for that thing can find it and attempt to acquire it. Many candidates pursue both paths, thinking they’ll cover both angles.

Let’s also realize that some people have not had to switch jobs in the past 5-10 years – most candidates are not professional job seekers. For many of these people, they simply respond to the advertisements of the major job boards as the “new” way of finding a job as compared to the last time they may have had a career transition. Why not let 100’s of recruiters try and find you the best opportunity? Aside from the experience they may have with poor recruiters, this is not a bad value proposition. Many candidates aren’t even aware of how many calls they will get once they do post their resume. But just because they get a large quantity of calls, it does not mean they get a large quantity of quality calls – calls for positions that are very close to their ideal career opportunity.

I’ll also address the idea that all good candidates have a magical network of people who can automatically find them their next optimal career opportunity without them having to look online. Some people have this magical network – but even so, there is no guarantee that this network can provide the ideal job opportunity at the right time.  I have a large network – and if I were to leave my employer – I would certainly leverage it. However, I do not for one second think that this network can be guaranteed to offer me the best possible match for me, nor all of the other fantastic opportunities out there that neither I nor my network can provide me.  A strong analogy goes back to eBay.  If I am looking to sell something, why would I only limit myself to the people I know?

My main point here is that it is not only the “bad” candidates that decide to post their resumes online – I’d go back to the random sample concept.  However, it is easy for staffing professionals to assume this is the case, especially if their primary method of recruiting is cold calling – they’re not going to hit many people who have their resumes posted online.

#3 Sourcer/Recruiter Talent and Ability

I’d also like to take this time to comment on database and talent mining expertise. I have recruited and placed many “A+” candidates from the job boards that my clients and competitors also had access to. For a look into a real world example of how I accomplished this  – read this post about a Google Network Performance Tester position that hundreds of agency and contract recruiters had been working for 4 months.

How is it that no one else found these people? Did I get lucky? Only if you can get “lucky” consistently.   Just because many people have access to a database, it is not safe to assume that everyone can find the same candidates, or find ALL of the qualified candidates, or find the BEST candidates in that database. Perhaps the people who are always claiming the job board resume databases have low quality candidates lack the proficiency to actually FIND the high quality candidates.

Conclusion

I firmly believe that many people believe the hype that job board candidates are low quality, and that anyone with access to them can magically find every candidate available, but it’s simply not true.  Take a moment to consider the laws of statisitics, the candidate’s perspective, and the widely varying levels of sourcer/recruiter talent mining ability before you are quick to assume that the job boards have low quality talent.

Back to the normal distribution – the exact shape of the bell curve could be disputed, flatter in the middle or more sharply peaked, but I hope I at provoked some thought by challenging the apparently widely held belief that most job board candidates are not desirable, and that conversely most of the “good” candidates are not on the job boards. I hope it helps that I drew upon some statistical and mathematical theories rather than sticking to subjective opinion only.

Related Posts Plugin for WordPress, Blogger...

 

 

  • Pingback: 15 Common Secondary Sourcing Mistakes | Boolean Black Belt

  • Pingback: Do you have Talent Intelligence? | Boolean Black Belt

  • http://www.fastcompany.com/blog/joshua-letourneau/fc-talent-manifesto-sex-lies-and-recruiting-0 Joshua Letourneau

    Glen, I thought this conversation was solely at ERE, so it took me a while to stumble across this.

    Personally, I did not suggest ‘bad candidates’ are only on job boards – this is a political re-framing of a statement. However, in the sake of debate, I understand the value of extrapolating one single comment and interpreting it in such a way as to paint a picture for your point or counter-point. I can tell you that I have a pretty strong background in statistics and including Wikipedia definitions of statistical terms does not indicate anything more than a theoretical understanding.

    I will leave you with this, because it is the only truth in this situation: It is the ‘observer’ that judges candidate quality. In the case of a large candidate database, you and I can be looking at the same population, yet come to different conclusions as to the quality of each and every candidate on an individual basis. Why? Because we are different observers. Another example would be Company X versus Company Z. If Company Z is more specialized and needs more niche skill-sets, then they may evaluate the candidate population differently than Company X.

    Therefore, what may amount to a normal distribution to you may not amount to a normal distribution to me (i.e. 20% A-level candidates, 20% B-level candidates, 20% C-level candidates, 20% D-level candidates, 20% E-level candidates). While your evaluation of the candidate population may yield a normal distribution, mine may not. It’s all relative. In addition, my standard deviation (sigma) may be greater than yours. Again, it’s all relative. And in that sense, stating that job board databases are ALL ‘normally distributed’ is the same as saying that ALL job board resumes belong to ‘bad candidates’. Both statements are wrong.

  • Boolean Black Belt

    Joshua,
    While I don’t have a degree in statistics, I have taken college level courses in statistics and got straight A’s. I reference Wikipedia so that my readers can learn more about some of the statistical concepts I wrote about. I had to reacquaint myself with them as it’s been over 10 years since I’ve done any hands-on statistics work, and the Wikipedia articles are very good overviews of the concepts.

    I partially agree with you on the “observer” argument. I agree that candidate quality ranking and labeling is highly subjective and in the eye of the beholder. But I would not go so far as to stereotype entire companies with regard to the quality of candidates recruited.

    I think it really comes down to the individual recruiter level, regardless of firm, and you can’t truly compare candidate quality unless you and another recruiter had evaluated exactly the same candidate and labeled them – then compared notes.

    In the end, I feel debating candidate quality is moot unless you can compare the exact same candidates. However – even that would be open to subjective differences of opinion. No single recruiter can stamp a candidate as “A” and say factually and objectively that they are an “A” candidate – it’s just a personal opinion.

    Why I can only partially agree with your “observer” position is that I feel the only relatively neutral (unbiased by any individual recruiter’s personal opinion) and objective measure of candidate quality is the client’s assessment of candidates. The client/manager is the ultimate judge of candidate quality. When I consistently achieve 1 candidate submitted/1 hire – in competitive scenarios (other recruiters are submitting candidates as well) – I can feel pretty confident my “A” is also my client’s definition of “A”. I could care less how recruiters at other firms rate their candidates, as I am confident in my assessment of candidates. I’ve placed many high level, highly niched skillsets in critical positions (Vignette/Interwoven portal architects responsible for redesigning Sprint/Nextel’s entire corporate portal, EMC SAN program managers for EMC, TS/SCI cleared identity and access management directors and architects, EAI project managers, etc.) – so I’m not talking about low level generic profiles.

    You raise a good point with regard to the shape of the normal distribution curve and how it can be shaped differently based on standard deviation. However, I would interpret the standard normal distribution curve to represent approximately 10% “A” talent, 15% “A/B”, 25% “B”, 25% “C”, 15% “C/D”, and 10% “E” – not the 20/20/20/20/20 you suggested.

    My whole point is that the job boards, as well as LinkedIn and any other source of candidates that has no barrier to entry or built in quality filter (an an internal database should) is going to be a true random sample of the total population, and thus a standard normal distribution will apply – regardless of any recruiter’s individual opinion. So if Monster had only 10% “A” talent, that may seem like a small percentage, as it should, because top talent is exactly that – “top.” However, when Monster claims to have 38M resumes, 10% of 38M is still 3.8M “A” candidates – a pretty big pool.

    I do see many respected staffing/recruiting leaders throwing around phrases such as “the job boards have low candidate quality,” and to me – that’s a broad, sweeping, opinion based statement that is statistically impossible when it comes to large sample sizes (10′s of millions). That’s why I wrote the article and participated in the original ERE discussion – to try and show people that simple statistics disprove the idea that a population of 38M people cannot have such a skewed distribution of 0% “A”, 20% “B”, 50% “C”, 30% “D + E” (for example, as an interpretation of low quality).

    Thank you very much for your passion and interest – I sincerely appreciate the challenge to my post. If we all agreed on everything, the world would truly be a boring place with very little critical thought.

  • http://www.fastcompany.com/blog/joshua-letourneau/fc-talent-manifesto-sex-lies-and-recruiting-0 Joshua Letourneau

    Glenn, great response. We agree on more than you might think :) Especially the below paragraph, which as you know, is a selling point of someone who wants to sell a candidate that doesn’t have a resume on a job board. You know, there are quite a few dogmatic beliefs in our industry that are bunk, and the notion that all job board candidates are poor quality is insane.

    “I do see many respected staffing/recruiting leaders throwing around phrases such as “the job boards have low candidate quality,” and to me – that’s a broad, sweeping, opinion based statement that is statistically impossible when it comes to large sample sizes (10′s of millions). That’s why I wrote the article and participated in the original ERE discussion – to try and show people that simple statistics disprove the idea that a population of 38M people cannot have such a skewed distribution of 0% “A”, 20% “B”, 50% “C”, 30% “D + E” (for example, as an interpretation of low quality).”

    P.S. Please don’t consider my statement about recruiters judging candidates differently as anything more than an example. There are great recruiters and bad recruiters at most companies – maybe even a bell-curve if the population is big enough :)

  • Pingback: Top 10 Candidate Sourcing Best Practices | Boolean Black Belt

  • Pingback: Job Posting vs. Searching for Candidates

  • Pingback: Resumes Are Like Wine

  • Pingback: Why Sourcing is Superior to Posting Jobs for Talent

  • Pingback: Why Sourcing is Superior to Posting Jobs for Talent « Venkatesh Kothapalli

  • Pingback: Do You Have the Proper Perspective in Recruiting?