Back in March 2010, I had the distinct honor of delivering the keynote presentation at SourceCon on the topic of resume search and match solutions claiming to use artificial intelligence in comparison with people using their natural intelligence for talent discovery and identification.

Now that nearly 2 years has passed, and given that in that time I’ve had even more hands-on experience with a number of the top AI/semantic search applications available (I won’t be naming names, sorry), I decided it was time to revisit the topic which I am very passionate about.

If you’ve ever been curious about semantic search applications that “do the work for you” when it comes to finding potential candidates, you’re in the right place, because I’ve updated the slide deck and published it to Slideshare. Here’s what you’ll find in the 86 slide presentation:

• A deep dive into the deceptively simple challenge of sourcing talent via human capital data (resumes, social network profiles, etc.)
• How resume and LinkedIn profile sourcing and matching solutions claiming to use artificial intelligence, semantic search, and NLP actually work and achieve their claims
• The pros, cons, and limitations of automated/black box matching solutions
• An insightful (and funny!) video of Dr. Michio Kaku and his thoughts on the limitations of artificial intelligence
• Examples of what sourcers and recruiters can do that even the most advanced automated search and match algorithms can’t do
• The concept of Human Capital Data Information Retrieval and Analysis (HCDIR & A)
• Boolean and extended Boolean
• Semantic search
• Dynamic inference
• Dark Matter resumes and social network profiles
• What I believe to be the ideal resume search and matching solution

I am often asked my opinion on the best use of resume search “aggregators” such as those offered by infoGIST, TalentHook, Data Frenzy RAM, DaXtra, AIRS SourcePoint and others.

If you’re not familiar with resume search aggregators, a resume search aggregator is an application that allows a user to enter a search string that will simultaneously execute across multiple free and paid job board resume sites and aggregate the results.

Many HR, sourcing, recruiting and staffing teams tend to use resume search aggregators at the associate level – enabling individual sourcers and recruiters to run Boolean search strings focused on specific hiring needs through a resume search aggregator to to simultaneously search several paid job boards as well as in some cases 100+ free job board resume databases.

While there is a huge convenience factor in using resume search aggregators in this fashion, as each sourcer or recruiter can save time and effort in not having to log into multiple job boards to execute Boolean search strings on each one individually, there are some limitations of resume search aggregators you need to be aware of, and I feel there is a better way to use this technology.

Limitations of Resume Search Aggregators

Limited Boolean Logic Support

Many of the resume search aggregator applications do not support full Boolean logic or even symbols such as the asterisk for root word/stem searching.

That, in and of itself, is a major limitation when attempting to mine information systems such as job board resume databases, because highly effective and precise search strings cannot be created.

Also, some of the aggregators don’t even allow you to “hand code” Boolean search strings – instead, forcing you to create searches using a query builder which often significantly limits the ability to create effective searches.

Similarly, not all job board resume databases support full Boolean logic.

Even if a resume aggregator application supported full Boolean logic, if it is being used to search a site that does not support it – you will not get the results you are looking for (or any results) because the job board resume database can’t execute your search string.

Lowest Common Denominator Search Functionality

If you are attempting to run a single search through multiple paid and free job board resume databases, your search won’t execute across every site you’re searching unless you use the lowest common denominator search functionality. In other words – if you want your single search to work on every site you are searching across, you have to use the syntax and search filters/criteria that each and every site supports.

For example, Monster supports fixed proximity search with their NEAR operator, yet I am not aware of any other major job board resume database that supports the NEAR command. Thus you cannot configure a search in a resume search aggregator using the NEAR command, as it will only execute on Monster and not any site that doesn’t support it.

Another area important for many recruiters is the ability to search for potential candidates by citizenship. Some sites like Dice.com allow you to search specifically for U.S. Citizens, while Monster does not and most others do not (and I still don’t know why – this is a very valuable feature!).

Similarly, some job board resume databases do not support zip code radius or education searching. If you are attempting to use a search aggregator to find candidates in a specific area via zip code radius, or candidates with specific a education level, you will not get results from any job board resume database that does not support that kind of search.

On the other hand, some search aggregators do not support the ability to search by zip code or education. If the search aggregator itself can’t perform these kinds of searches (and/or you can’t configure a Boolean to accomplish the same thing), you simply can’t do it through the aggregator.

One standout solution with regard to this issue is DaXtra – they are one of the few vendors in this space that allow you to search across multiple sites/sources and configure the search for each site to take maximum advantage of each site’s unique search capabilities. For example, you can run the same Boolean string in DaXtra to search Monster and Dice, and configure the applicable citizenship and other different search functionality to

Sorting by Relevance

Some resume search aggregators do not allow users to sort the results returned from a Boolean search across multiple job board databases by relevance.

This is a huge fail – when it comes to search, the whole point of searching for information is to get relevant results, and the most effective way to sort results is by relevance.

Search Limits

Most resume search aggregators have limits on the number of results that can be returned from each search – some are as low as 150.

If a sourcer/recruiter is running a search on job boards and opening up the parameters to search resumes posted in the last 90, 120, or 365+ days, in many cases the number of results can exceed the limits imposed by the aggregator. That means in some cases, sourcers and recruiters simply CANNOT find and retrieve some results because of the artificial limits.

The reason why some aggregators have limits on the quantity of results that can be returned per search and in some cases per DAY from specific job boards (usually the big ones like Monster) is because the big boards don’t want people automatically strip-mining their databases via software.

A Better Way to Use Resume Search Aggregators

In contrast to how many organizations leverage resume search aggregators, I propose that the ideal utilization of these applications involves largely automated 24 X 7 mining of job board resume databases, whereby “broad” Boolean searches are configured and saved into the resume search aggregator application designed specifically to pull relatively general results into an internal database/Applicant Tracking System (ATS) on a daily basis.

Once the resumes are pulled into the internal resume database/ATS, sourcers and recruiters can run more precise, configurable, and powerful full Boolean queries, searching by location, and sort the results by relevance, and find the best candidates available without artificial limits on the number of results. (Note: if your internal database/ATS does not support full Boolean logic and sorting by relevance – it should – but that’s another post entirely).

I recommend automated daily searches because even in the largest metro areas and when running relatively broad and generic searches, you are not likely to bump into any search aggregator’s max result limit because there are only so many people who will post their resume every day of any given skillset.

I recommend using search aggregator applications to run broader searches because most can’t run precise Boolean queries anyway, so why try to do so and miss picking up candidates in the first place?

Keep in mind that every single search aggregator is limited to the search interfaces they search through, and they will be “dummied down” to the lowest common denominator (the least configurable/capable search engine). In many cases, more complex and precise queries are impossible to achieve anyway.

Instead of a running the risk that a search aggregator (or the job boards I am searching through it) is incapable of returning results of candidates that DO exist but are incapable of returning from a complex Boolean search string, I would much rather get them into my internal database for permanent data capture.

While broader searches can yield some “false positives,” I have not found it to be a problem. I have worked with resume databases from 100,000 to over 3,000,000 records. Statistically, the more, the better – and sometimes the best candidates have lackluster resumes – you never know anything about a person until you get them on the phone/meet them in person and talk to them!

I also believe in automating the input of search results into an internal database/ATS because all too often, sourcers and recruiters over-analyze resumes and unknowingly miss fantastic candidates. This is perhaps the #1 mistake sourcers and recruiters make when searching for resumes.

Benefits of this Utilization of Resume Search Aggregators

1. Increased sourcer/recruiter productivity. Instead of sourcers/recruiters running individual candidate queries across multiple job boards on a daily basis, they can save time by only having to search their ATS/CRM for the majority of their candidate needs.
2. It can automatically build the breadth and depth of your ATS/CRM more quickly and more thoroughly than any team of people could, and at a lower cost.
3. You can automatically build a large passive candidate database – the resumes you capture of active candidates today will evolve into passive candidates over time.
4. Automatic, guaranteed, and permanent data capture. Resume search aggregator applications won’t look at a resume and not enter it into your ATS/CRM because it doesn’t match the position they are currently working on. People who post their resumes on job board databases don’t often leave their resumes up indefinitely – at some point most people remove their resume and if you don’t capture them before they do, you may never be able to find them in the future.
5. Reduced job board costs. With automated sourcing technology, you will not need as many passwords or resume views.
6. Guaranteed use of your ATS/CRM. With automated daily mining and extraction of resumes from the job boards, your ATS/CRM should be the first source of your sourcers and recruiters search.

Final Thoughts

If you already use or are considering using a resume search aggregator application such as infoGIST, TalentHook, Data Frenzy RAM, DaXtra, or AIRS SourcePoint, I highly recommend that you consider the limitations of search aggregators.

You should seriously contemplate the advantages and benefits of using an aggregator application to perform broader, more general automated daily searches to enter results into your internal resume database/ATS for permanent data capture and for more precise search/retrieval from your internal system than your aggregator or the job boards allow.

When you’re searching the Internet for potential candidates, it’s quite common (and practical) to search for resumes of  people who are likely to be local to your opportunity. The two main ways of doing this are searching by area code and searching by zip code range.

While there are limitations of both approaches (not everyone includes a phone number or address), in this post I want to explore an interesting phenomenon that was brought to my attention not too long ago which clearly demonstrates that even when people DO provide a phone number or address, you may not be able to find them by searching for that information.

Yes, you read that right.

Recently I had a recruiter in my network ask me why he could find a specific resume online using Google, but when he added the area code (which was clearly on the resume) to the search string – the search result disappeared.

When he sent me the screen shots, I investigated – and sure enough, he was right.

What I find especially interesting is the fact that he found the resume at all, because he wasn’t using any location criteria in his search to try and find local people. If his initial searches tried to target local people using an area code or zip code range, he would have never found the resume in the first place.

Moreover – he would never know of it’s existence.

Click on the image below to watch a short video clip of the Google resume search phenomenon:

Fascinating and Frightening

Seeing something like this makes me wonder how many people have resumes and other information online that are indexed by search engines, but are never found due to the search criteria used.

My instincts tell me there are many – but the insidious nature of searching is that all searches “work.” In other words – you get results from your searches, which gives you a false sense of security (and perhaps accomplishment), but you’re not aware of available results that your searches simply didn’t return. You can’t be.

In fact, it never occurs to most people to even wonder about available results they have access to, but are never retrieved.

However, just because you don’t find something – does that mean it doesn’t exist?

Why Does Google Fail to “Find” the Search Result?

I have my ideas as to why that specific resume cannot be found when adding the area code to the search, but I am curious to know your thoughts.

Multiple Search Engines

Of course, you should always employ difference search engines, as they don’t all index the exact same pages/sites. And even of they did index the same pages/sites, it doesn’t necessarily mean that they would return the exact same results for any particular search.

For example, while Google, Bing and Yahoo can find the same result without the area code, only Bing and Yahoo find the result that Google fails to return when adding the area code to the search string.

Beware!

I bring this example to your attention because you should always be aware of the fact that you may have access to people that you simply cannot find using certain search criteria, and it’s not limited only to location searching. Searching by skill, title, company, industry, etc., can be similarly affected.

It’s often simply impractical to NOT search with some location criteria – few people have the time to sort through and review results from all over the world that they cannot use. And location-focused searching doesn’t just apply to resume search, but also for any kind of searching in which you are trying to find people that live relatively close to where the opportunity you are sourcing/recruiting for is located.

Hopefully this post makes you think twice about your strategy and tactics to try and identify local people and helps you uncover previously “unfindable” results.

Are you attending or thinking about attending SourceCon 2010 in San Diego in March?

I am going to be the keynote speaker for the event, and I will be presenting on Artificial Intelligence vs. Human Cognition when it comes to sourcing and matching resumes.

If you’re curious to know what kinds of things I’ll be addressing during the session, here is a sneak peek:

1. The intrinsic and often overlooked challenges associated with sourcing resumes
2. What artificially intelligent semantic search and match applications claim to do and how they actually work
3. The limits of artificial intelligence
4. What people can do that semantic search applications cannot
5. The 5 levels of semantic search
6. The 5 levels of secondary/e-sourcing
7. What I believe would be the ideal candidate sourcing/talent identification solution

If you’ve ever wondered about the fantastic claims that some of the semantic search application vendors on the market make as to how their solution can mimic a senior recruiter when finding candidates, then you will be very interested in hearing what I have to say about the reality of what they can do.

If you’re a sourcer and you’re concerned that your role/position might eventually be replaced by sourcing software, you will be encouraged by my analysis and supporting arguments that explain why the abilities of creative and investigative sourcers will always be in demand – tomorrow and 50 years from now.

I hope you will be able to attend SourceCon 2010 – I know I’m looking forward to it!

If you’re unable to attend, the good news is that the presentations will likely be streamed. Additionally, I plan on posting my expanded slide deck, including all talking points – so you won’t be stuck staring at some pretty pictures wondering what the heck I talked about.

In response to my recent post about the deficiencies in the search capability of many Applicant Tracking Systems, a few people commented to the fact that resumes stored in applicant tracking systems become stale and outdated over time, which may explain why ATS resume databases are often the candidate “source of last resort.”

While candidate records inevitably age over time and can become outdated, this definitely does not have to be the case.

A candidate record can only truly go “stale” if no one ever makes contact and updates the record with more current information from time to time – and it need not even be every 6 months.

Any recruiter worth their salt will attempt to maintain periodic contact with most candidates and update their information as appropriate, regardless of their job search status. This can also be automated to some extent with strong and effective CRM functionality – so even if the recruiter forgets to follow up with someone every 6 months, the CRM won’t.

Resumes Are Like Wine

While human capital data in the form of resumes and candidate profiles may get outdated, it never truly loses its value. Resumes and candidate records are like fine wine – they only get better with age.

Yes, I believe the value of human capital data actually increases over time.

If I find a resume of a 2 year Unix systems administrator today and permanently capture them into my ATS, over time that person will gain experience and expertise, and likely advance their career along the way. In 5 years I will have a 7 year Unix admin, a Unix systems engineer, perhaps a project manager or even a storage area network specialist – who knows? No matter their career path and progression, I will stay in touch with them and routinely update their information - regardless of their job search status.

The same is true of nearly every profession – accountants, attorneys, physicians, customer service reps, mechanical engineers, recent college grads, etc. – they will all gain experience and advance their careers over time. 

Limited Shelf Life

Did you know that some people who post their resume in online resume databases (job boards and such) sometimes pull their resume down shortly after they post it, rendering it unfindable? Sometimes in a matter of hours!

Social networking sites such as LinkedIn and Twitter can also suffer from a similar effect. Because they are based on UGC (User Generated Content), at any time any user can make their profile private and unsearchable (even via X-Ray searching in some cases!), or simply remove content that may aid you in searching for/identifying them based on their professional skills and experience.

However, if you’re always on the lookout for certain types of professionals, scouring every source available to you, and you permanently capture the information you find into your ATS/CRM, you may essentially be collecting rare vintages (yes, I’m going to continue with the wine analogy) that may no longer be in circulation in the near future. 

In other words, when you find and capture that resume or profile of the 2 year Unix admin (or accountant, or attorney, or recent grad, etc.) who pulls their resume or alters their social networking profile at some point in the future – you may have in your possession a candidate that may never be found by anyone else again.

This would allow you to specifically search for that particular candidate and reach out to them in a year or two’s time – when they are passively looking or not looking at all – and present them with a position that is well aligned with the next step in their career. At this point, you may literally be one of the few people who have quick and easy access to that candidate as their career progresses, regardless of their job search status.

ATS Databases are 70% Passive/Not Looking by Volume

I believe that a well stocked ATS candidate database is likely to consist of mostly (approximately 70%) candidates who are not looking or who are passively looking. That percentage is probably even higher when you consider only candidate records that have been entered/created over 3 months ago. So, if you’re one of those recruiting professionals who believe the hype that active candidates are bad and passive candidates are good, you should be excited about the prospect of building a private “passive” candidate database that you can mine to your heart’s content.

While I am most decidedly NOT one of those people who buys into the idea that passive candidates are the “best” candidates, I can tell you from experience that candidate closing and control is almost a non-issue when you are dealing primarily with people who are not actively looking, are not being called by every other recruiter in the known universe, and don’t have 5 interviews scheduled this week and 2 offers in hand.

So it IS nice to be able to purposefully target and dip into a large pool of well qualified candidates, who are not actively looking, and many of whom no one else has quick and easy access to. 

It’s not just nice; it’s also a significant competitive advantage.

Let Your Candidate Data Age Naturally

If you have a relatively large candidate database (10,000 to 1,000,000+), you needn’t worry about trying to maintain “relationships” with all of them – it’s actually impossible, unless your definition of “relationship” includes automated emails.

If you’re overly concerned with having ultra-fresh information on all of the candidates in your system at all times - don’t be. It isn’t really necessary.

I’ve called and made easy, frictionless hires with candidates who had records/resumes that had not been updated in 4 years. A seasoned sourcer or recruiter can easily make an educated guess at “career trajectory,” and when you make a call to someone whose resume is not on the Internet, not on LinkedIn, not in an online resume database – you essentially have a candidate no one else has practical, targeted access to – and closing/control is a non-issue when you call with the right opportunity, by design.

Final Thoughts

If you permanently capture data on your target professionals relatively early in their careers, you can cultivate their candidate records as their careers progress, allowing you quick and easy access to them as they evolve into more experienced passive or even non-job seekers – the virtually “ungettable” candidates that your competitors wish they had access to.  

If this approach to valuing and leveraging your candidate data doesn’t appeal to you, and you happen to be growing tired of having to store all of those old, stale resumes in your ATS/CRM – give me a ring – I’d be glad to take them off your hands.

This post is second in a series focused on using Google to search for resumes on the Internet.

In the first post I left some unanswered questions, such as why:

• I didn’t talk about searching for CV’s
• I didn’t suggest using the tilde ~ operator in conjuntion with the word “resume”
• I didn’t use -~job when trying to eliminate false positive results
• I didn’t talk about targeting filetypes
• I didn’t talk about just searching for the word “resume” without using it in conjunction with inurl: or intitle:
• I didn’t mention the use of Google Custom Search Engines (CSE’s) to find resumes

Whether or not you had those questions burning in your mind, I will address them all in this post.

Challenging Google Resume Search Assumptions

I’ve read my fair share of recruiting blogs and online discussions between recruiters and sourcers. As such, I encounter quite a bit of advice regarding tips and tricks to use when searching for resumes on the Internet using Google.

Some of the suggestions I see make sense at first, but being the inquisitive guy that I am, I don’t just take the suggestions and run with them, assuming they accomplish what they seem to accomplish. I take the time to test search tips, tricks, and suggestions to make sure they add value to my search efforts and that they do EXACTLY what they claim to do.

Today, you get to benefit from some of these tests, as I am going to challenge some of the suggestions I’ve come across over the years when it comes to searching for resumes on the Internet using Google. Let’s get going, shall we?

Using ~Job To Eliminate The Words Job And Jobs From Results

Google has a special operator that allows you to search for synonyms, as well as alternate endings for any word that is preceded by the tilde ~ symbol. For example, let’s see what kinds of results are returned by this simple search: ~car

You can see that Google returned results including the word “car,” but also highlighting words such as “BMW,” and ”cars” - words Google’s search engine thinks are synonymous with the word “car.” The fact that Google thinks the word “car” is synonymous with BMW is powerful and free advertizing for BMW – but that’s for another post.

I’ve seen some sourcers and recruiters suggest coupling the tilde ~ symbol with the word “job” in an attempt to eliminate results with the words “job” and “jobs,” instead of using both -job and -jobs, as I recommend.

HOWEVER - I have found that using -~job does NOT in fact eliminate all results that mention word variants, or words with alternate endings, such as “jobs.” For example – run this on Google: (intitle:resume OR inurl:resume) -~job

On page 2 of the results, you can find this:

Yep – I see the word “jobs” in there, don’t you? —> “I’ve had too many jobs.”

Let’s see what happens if we actually try to use -~job and also search specifically for the word “jobs” in the same search: (intitle:resume OR inurl:resume) -~job “jobs”

Click here for the results. 475,000 results at the time of the search.

If Google actually interpreted -~job as both -job and -jobs, we shouldn’t get ANY results, let alone nearly half a million, because the search is written to actually look for a word we are trying to eliminate.

You will essentially get the same results if you run the search this way: (intitle:resume OR inurl:resume) -~job +jobs

But wait, the Google search weirdness continues!  Let’s see what happens when we use -~job and also try and make Google search for “job” at the same time: (intitle:resume OR inurl:resume) -~job “job”

Click here for the results. You get 5 results that become 10 if you click “repeat the search with the omitted results included.”

I hope you can appreciate how strange those results actually are.

Confused by the results?  Me too!

It actually takes using Google’s “exactly as is” search symbol, the plus + sign to finally see that at least using -~job will in fact eliminate the word “job” from search results.

What this testing demonstrates is that trying to eliminate the 2 words “job” and “jobs” from being returned in search results by using -~jobs DOES NOT WORK.

It appears that the tilde operator on Google does not in fact also search for and return (or eliminate) words with alternate endings, such as plural words. For the word “car,” yes, but not when it comes to the word job. As such, it’s safer to simply write out -job -jobs.

Using ~Resume To Search for Resumes and CV’s

I’ve heard some people suggest coupling the tilde with the word resume, in order to find results that have words synonymous with the word “resume,” such as CV or Curriculum Vitae in their url or the title of web results.

For example: Java (intitle:~resume OR inurl:~resume) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas)

Let’s test this theory.

Search #1 No Tilde

Java (intitle:resume OR inurl:resume) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas)

Click here for the results. 294 results at the time of the search.

Search #2 WITH Tilde

Java (intitle:~resume OR inurl:~resume) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas)

Click here for the results. 2760 at the time of the search.

Wow! 2760 results when we used the tilde vs 294 without the tilde – seems like a no-brainer to always throw in the tilde when searching for resumes, right?

Not so fast…let’s take a look at the search results past result #300 when using the tilde and see what we find:

Click here to see the search results page.

Some pretty ugly results, right?

Yuck!

So where are the 2400 extra resumes we were expecting to get?

Keep searching past the 300 mark and you will see tons of junk results. That’s why I can’t in good conscience recommend using the tidle ~ in conjunction with the word “resume” when searching for resumes.

Where are the CV’s Anyway?

If you’re wondering where the CV’s were in the results, you can simply target them with a search string like this: Java (intitle:cv OR inurl:cv OR intitle:”curriculum vitae” OR inurl:”curriculum vitae”) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas)

Click here for the results. 82 results at the time of the search.

More on Searching for CV’s

You could of course search for both “resume” and CV in the titles and urls in web pages and documents, like this: (intitle:resume OR inurl:resume OR intitle:cv OR inurl:cv OR intitle:”curriculum vitae” OR inurl:”curriculum vitae”)

But when I have searched for CV’s, I have noticed that many people who save their resumes online who use the term CV are in school, are educators at a university, or are international (non-US) folks, whom you may or may not be able to engage depeding on your location and need.

Did you notice that 4 out of the 5 results from the last screenshot have .edu in the url? That means they are from university sites, and I have found many of these .edu CV’s to be of people with no paid work experience – which depending on your hiring needs, you may not be able to hire.

Auto-Stemming: Trying to Use -Job to Eliminate “Jobs” From Results

When it comes to your search terms, Google claims that they will look for some word variants automatically, such as words with alternate endings or pluralization.

However, it does not appear to work on the word “job.”

For example, if you are trying to eliminate false positive results of the word “jobs” when searching for resumes, using -job does not eliminate results with the word “jobs,” although if Google were in fact truly auto-stemming the word job, we could assume it should.

For example, let’s try this: (intitle:resume OR inurl:resume) -job

Here is a result from page 2:

That result does in fact mention the word “jobs.”

So once again I have to recommend simply spelling out -job -jobs when trying to eliminate job-related false positive results, because Google does not auto-stem -job to also include -jobs.

Not Searching for the Word “Resume” in Titles and URLs

I’ve heard some sourcers and recruiters say that it is unecessary to search for the word “resume” specifically in titles and URLs, through the use of (inurl:resume OR intitle:resume).

This is because simply searching for the word “resume” should return results with the word “resume” in the body of the website/page as well as in the url and title.

This is accurate, for example - let’s see what happens if we run this search: Java “resume” -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas)

Click here for the results.  About 1400 at the time of the search.

You can see that the word “resume” is in fact highlighted in the blue title and green url lines, as well as in the “body” of the search results.

If you remember, the search I used previously that did specifically target (inurl:resume OR intitle:resume) only returned 294 results.

When I searched for “resume” without specifying that the word had to be in either the url or title, I got 1390 results.

That’s a big difference!

However, let’s not get too excited about the “extra” 1100 results we get from just searching for the word “resume” and not limiting the search to only results mentioning “resume” in the url and/or title.

Let’s take a look at the results past #300 to see what we’re really getting: Click here for the results.

Umm…those aren’t resumes – those are false positives!

Quality vs Quantity

Don’t ever be impressed by large quantities of results until you check in the “deep” end and make sure that they are just as high quality as the first few pages.

This test has shown that simply searching for the word “resume” anywhere and not forcing Google to specifically target results that mention the word “resume” in the title and/or url of results does not in fact yield more high quality results.

What it does is get you more junk in most cases.

So I recommend always beginning your resume searches using the (intitle:resume OR inurl:resume) syntax like this, specifically targeting urls and titles that contain the word “resume”: Java (intitle:resume OR inurl:resume) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas)

Here is a snippet from the the LAST page of the results:

Yep – resumes all the way to the last page.

Searching for Filetypes

The last assumption I will challenge is that it’s always a good idea to search for specific file types when searching for resumes on the Internet.

Google does support searching for results of specific filetypes – here is an example of a search targeting PDF files:

Java (intitle:resume OR inurl:resume) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas) filetype:pdf

All of the results are PDF files.

While searching for specific filetypes can be handy, there are many filetypes available to people when saving their resume online.

Back in 2007, I read Jim Stroud’s Resume Sourcing Survey and it was an eye-opener – there are so many different filetypes people use when creating and saving resumes online. While you may think you are uncovering a lot of resumes when searching for doc, html, php and pdf fileytpes, you’re probably not searching for ASP, XML, PID, PHTML, MHT, DOT, CGI, XSL, WPD, and SDW filetypes.

I am not exactly sure why some people suggest searching for specific filetypes when searching for resumes. If I had to guess, I would say it has something to do with trying to get rid of false positive non-resume results.

Perhaps the thought process is that job postings and such are not typically DOC or PDF files?

In any event, searching for specific filetypes is unnecessary, because when you use a search string that doesn’t specify filestypes such as this one: Java (intitle:~resume OR inurl:~resume) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas) …you’ll get every resume result available, REGARDLESS OF FILETYPE.

You can see from the results below that it returns ALL results, regardless of filetype.  In the screenshot, you can see we snagged 4 different filetypes from the first 4 results: HTM, PDF, DOC, and HTML.

That’s why I don’t recommend targeting specific filetypes – because if you do, unless you actually search for every possible filetype available, you will be missing results.

Perhaps the only time I would strongly recommend targeting specific filetypes is when you are specifically looking to find people who do not title or save their resume with the word “resume.”

Using Google Custom Search Engines (CSE’s) to Find Resumes

There has been much written on using Google Custom Search Engines to find resumes on the Internet.

Creating your own and using others’ Google CSE’s can be a convenient way of automatically building in core and essential search logic (title/url search, false positive term removal, etc.) so that you don’t have to keep entering it into your searches.

However, when it comes to using someone else’s CSE, I’d advise that you not blindly use it without adapting it to your specific use.

Only you know what it is you are looking for specifically, where you would like to get your results from, and the specific locations you would like to recruit from.

I believe the best CSE’s are those that have location-specific logic built in, so that all you need to enter is your keywords and nothing else.

Implicitly trusting that someone else’s custom search engine has the optimum search logic would be a mistake. Use CSE’s designed by others as a starting point to modify and create your own that suits your specific need.

Conclusion

I hope you enjoyed me walking you through testing some of the more common Google resume search suggestions I’ve come across over the years.

It can be both easy and dangerous to follow search advice from anyone, regardless of their experience or reputation, and take their suggestions and immediately begin putting them to use.

I strongly recommend that you take the time to thoroughly test any sourcing/recruiting advice you read or hear about before making the assumption that it works as intended and that can help you achieve your goals.

As I’ve shown you today – some suggestions such as searching for specific filetypes can actually prevent you from finding all of the resumes available online, and others such as searching for ~resume can yield more junk than viable results.

Want to learn how to find resumes on the Internet using Google?  You’ve come to the right place!

Whether you are new to searching the Internet for resumes or you are a veteran Interent sourcer, I’ve included some tips, tricks, and observations for the novice and expert alike.

Targeting Resumes

When using Google to search specifically for resumes, it’s a good idea to begin by searching for the word “resume” in the title and/or the url of web pages.

For example: (intitle:resume OR inurl:resume)

Here is a sample result to illustrate how this works – we can see the word “Resume” in the blue TITLE line, as well as in the green URL line. The first line of any search result is the title of the webpage, and the url is the specific web page’s address.

Targeting the word “resume” in the title and/or url is very handy, because for many people, it’s simply common sense/instinct to either title the web page containing their resume with the word “resume,” and/or save their resume using the word “resume” in the name of the file.

Eliminating False Positive Results

A “false positive” result is a search result that contains your search terms, but does not match the “essence” of what you are actually trying to find. For example – if you’re searching for resumes, there are many sites that will be returned in your search results that are in fact not resumes, such as sites advertising resume samples and job postings that mention phrases such as, “submit resumes to…”

In order to remove most non-resume false positive results, you can use Google’s version of the Boolean NOT operator, which is the minus sign, to prevent your search from returning results with words such as sample, job, etc.

For example:

(intitle:resume OR inurl:resume) -job -jobs -sample -samples

Other ideas of terms you can try to eliminate that may return false positive results:

-eoe -submit -free -”resume service” -template -”resume service” -”resume writers” -”resume writing” 

Targeting Local Resumes

Area Codes

Some people who decide to make their resume available on the Internet also decide to publish a phone number. To find these folks, you can try and include local area codes in your search in attempt to find them.

Here is my favorite area code lookup site

Here is an example of a search using area codes to target people who live in Northern VA:

(intitle:resume OR inurl:resume) -job -jobs -sample -samples -”resume service” (703 OR 571)

What you’ll notice, however, is that searching for area codes can easily produce false positive results, as can be seen in the screenshot below – which are the first 4 results from the search. Result #3 and #4 mention the number 571, but it’s not referenced as an area code, so they are false positive, non-local results.

To clean up the false positive results that mention 571 or 703 but are not of resumes of people who live in VA, you can add the state and state abbreviation to the search as well as exclude more non-resume terms and phrases:

(intitle:resume OR inurl:resume) -job -jobs -sample -samples -template -”resume service” -”resume writers” -”resume writing” (703 OR 571) (VA OR Virginia)

While there are still a good number of non-resume false positive results, this can be expected because we didn’t even search for any keywords/skills. Once we do, we will notice a significant drop off in false positives.

For example:

(intitle:resume OR inurl:resume) java oracle -job -jobs -sample -samples -template -”resume service” -”resume writers” -”resume writing” (703 OR 571) (VA OR Virginia)

Zip Codes

Another way to search for local resumes is to target a zip code radius.

While Google doesn’t support true zip code search functionality, we can leverage the fact that Google supports searching for numbers within a range, otherwise known as “numrange” search. You can perform a numrange search on any sequential (low to high) set of numbers by typing two periods between the numbers without any spaces. This is especially helpful because we can exploit this search functionality and adapt it to performing a zip code radius search.

For example, take your target zip code and use this handy zip code radius lookup application, which you can use to search for zip codes in a radius and sort the zip codes within the radius from low to high.

Using a central zip code of 75201 in Dallas, TX and targeting a 20 mile radius, we can use that zip code radius application to get a range of zip codes from a low zip code of 75001 to a high zip code of 76155.

Using Google’s numrange search syntax, here is an example of a search string targeting people within a 20-mile radius of 75201 in TX:

Java (intitle:resume OR inurl:resume) -job -jobs -sample -samples -”resume service” 75001..76155 (TX OR Texas)

Sample results: you can see the zip code search worked beautifully:

Be Aware

Here are a couple of things to keep in mind when searching for local resumes:

• Not everyone will publish a phone number on their resume
• If they do publish a phone number on their resume, it could be a cell phone with a non-local area code if they relocated from another area – this is why is can be helpful to search for zip codes as well as area codes
• If they don’t publish a phone number at all, they should at least mention a city and/or state

Targeting Resumes That Are Not Titled As “Resume”

Although plenty of people do title their resume or the web page containing their resume with the word “resume,” there are some people who don’t. After searching for all of the results that DO contain the word “resume” in the URL or title of the web page, you need to try to eliminate them from future search results using Google’s NOT operator, the minus sign, to specifically target the results that don’t mention resume in the title or URL. For example:

-inurl:resume -intitle:resume

Don’t be tempted to try and write that as -(inurl:resume OR intitle:resume), because Google will not read that search string in the manner in which you intend. Instead of eliminating results with the word “resume” in either the URL or the title, Google will in fact will do the exact opposite. Click here to see the sad reality. Sorry Google.

Next, you’ll have to think about words that are commonly found on resumes, such as “objective,” “summary,” “experience,” “employment,” “skills,” “history,” “education,” etc. Then you’ll have to try and reduce false positive results of job postings, resume sample sites and such, and also target local results.  Putting it all together, your search string could look like this:

Java -intitle:resume -inurl:resume (objective OR summary) (experience OR history) education -job -jobs -sample -samples (301 OR 240 OR 410) (MD OR Maryland)

First page results:

Analyzing the Results

We can see that the first result is essentially of a resume, yet the person saved their work history as “biodata.” See?  Some people don’t use the word resume.

Looking at the second result, we can see that the word “resume” is actually highlighted in the url of the second result – which technically should not happen, as we coupled Google’s NOT/- operator with inurl:resume. Yet more proof that Boolean search does not = Internet search.  That’s 2 strikes, Google.

The third result is definitely not a resume – it is a LinkedIn directory by name, within which is at least one positive hit on our search terms.

When You Stray, You Pay

While there are no doubt resumes on the Internet that do not mention the word “resume,” anywhere within them, I have found that when you stray from targeting the word “resume” in web page titles and URLs, you can quickly find yourself swimming in a sea of false positive results, which is a huge time drain.

I believe the 80/20 Pareto principle solidly applies here: You will get 80% of your viable results from targeted resume searches and the other 20% from searching without targeting the word “resume.” Likewise, you can easily waste 80% of your time trying to sort through the sea of false positives in order to dig up the 20% of resumes that do not mention the term.

Conclusion

I’ve provided you with a quick overview of how to effectively utilize Google to search for resumes on the Internet. Be aware though - all is not perfect in when hunting for resumes in Google land – if you have not already, please read these 2 posts: The Internet has Free Resumes – So What?and Boolean search does not = Internet search.

Have Unanswered Questions?

For those of you who read this post and asked why I:

• didn’t talk about searching for CV’s
• didn’t suggest using the tilde ~ operator in conjuntion with the word “resume”
• didn’t use -~job when trying to eliminate false positive results
• didn’t talk about targeting filetypes
• didn’t talk about just searching for the word “resume” without using it in conjunction with inurl: or intitle:
• didn’t mention Google CSE’s

…you’re in luck.  All you have to do is read this post in which I debunk common Google resume search suggestions and assumptions.

In response to my post of Resumes on the Internet: Monster vs. Google one of my readers commented that “While it may be true that Monster has more resumes than Google, using a zip code search is not a fair comparison for Google. People who post their resumes on Monster are required to enter their zip code, while people who resumes are stored online will generally only put their email and/or phone number. Also, even using the term resume can be limiting in Google. Because it was not built to only index resumes, you have to get more creative to filter out the noise. You can try the ~CV or ~Resume, you can also take that out completely and search for types of documents, .DOC, .PDF, etc. and look for words commonly found in CV’s like education, objective, etc.”

His comment inspired me to get these industry heavyweights into the ring for a second battle and experiment with not using zip code ranges or the word “resume” when searching for resumes on the Internet using Google. Let’s begin with the same searches as Monster vs. Google Round 1.

Search #1 – Java, Oracle, Sprint or Nextel, State of MD

Google #1 Zip range (original search) = 4 results

(intitle:resume | inurl:resume) java oracle (sprint | nextel) 20601..21930 (MD | Maryland) -~job -~jobs

Google #2 No zip range, using area codes instead = 3 results

(intitle:resume | inurl:resume) java oracle (sprint | nextel) (301 | 410 | 443 | 240) (MD | Maryland) -~job -~jobs

Google #3 No zip or state = 6 results, 3 are not local to MD, but we did gain 1 new resume

(intitle:resume | inurl:resume) java oracle (sprint | nextel) (301 | 410 | 443 | 240) -MD -Maryland -~job -~jobs

Because we know that not everyone actually titles or saves or refers to their resume as a “resume,” we’ll try to search for words commonly found on resumes. Note – I am not a fan of ~resume or ~cv because I tend to get lots of junk. I am not fooled by a larger quantity of results when the quality goes down – it’s not about quantity, it’s about relevance. Also, not every mention of “cv” is actually referring to a curriculum vitae – I get all kinds of strange false positives with the letters CV. “CV” is more common internationally than in the U.S., and in this case, we’re searching for U.S. based candidates. Additionally, as one observant reader pointed out, many online resumes are of people at educational institutions (.edu) – and many of those contain mention of CV. For most positions, people with only educational experience are not viable candidates. So, instead of shooting for the word “resume,” we’ll go for common resume words:

Google #4 No mention of “resume,” using zip range = 33 results, however – no resumes I can see

NOT “resume” (objective | summary) education (history | experience) java oracle (sprint | nextel) 20601..21930 (301 | 410 | 443 | 240) (MD | Maryland) -~job -~jobs

Note – for you search geeks out there, using -intitle:resume -inurl:resume did not work as the syntax suggests it should – it still includes the word “resume” in the url. For example, this search:

-intitle:resume -inurl:resume (objective | summary) education (history | experience) java oracle (sprint | nextel) 20601..21930 (301 | 410 | 443 | 240) (MD | Maryland) -~job -~jobs

Yielded results like this:

Verdict: Monster crushed Google

Search #2 Java, Oracle, State of MD (very simple and broad search)

Google #1 – Zip range (original search) = 220 results

(intitle:resume | inurl:resume) java oracle 20601..21930 (MD | Maryland) -~job -~jobs

Google #2 No zip range, using area codes instead = 218 results. HOWEVER, these results have significant overlap with Google search #1

(intitle:resume | inurl:resume) java oracle (301 | 240 | 443 | 410) (MD | Maryland) -~job -~jobs

I tried to shoot for mutually exclusive results by negating the zip code range from search #1, however, I could not negate the zip code range and get it to run properly. For example, I tried all 3 variations below to no proper effect:
-20601..21930
-(20601..21930)
NOT 20601..21930

Google #3 No mention of zip or state – just area codes = 375

(intitle:resume | inurl:resume) java oracle (301 | 240 | 443 | 410) -MD -Maryland -~job -~jobs

While this search returned 375 results – most of the results are false positives of non local candidates. See below:

As you can see, the MD-specific area code numbers come up as false positive hits in non-local resumes

I will not bother with not using (intitle:resume | inurl:resume) – as stated previously, although I concede that targeting the word “resume” is limiting, whenever you go beyond it, you get tons of garbage false positive results (see Search #1 above). I (and most sourcers and recruiters) should have better (higher yield) sources to tap for Talent rather than having to dig through large quantities of false positives to find a handful of relevant results.

Monster – 592 results @ 1 month, 1000+ @ 3 months

Java and oracle – in MD

Verdict: Monster destroys Google

Conclusion
After seeing the results of Search scenarios #1 and #2, I will not bother to repeat the other searches I executed in Monster vs. Google Round 1. I’ve seen enough to know that regardless of how you spin Booleans searching for candidates on the Internet (zip codes, area codes, resume or not resume, etc…), the quantity of candidates with resumes on the Internet is not as large as many would believe or assume, and certainly not even close to as large as even 1 of the major online job boards.

The reader who inspired this post made an excellent point – Google and other search engines are not designed to index resumes. That’s why I have never been crazy about finding resumes in the Internet – while it certainly can be done, it’s like using a screwdriver as a hammer – you’re trying to use a tool for something other than it’s original or intended design.  That certainly explains the results of this exercise.

If you’re looking for resumes – why not leverage systems specifically designed to index resumes, like the job boards (paid and free), as well as your own internal database/ATS? Which, by the way, is also free.

And if you happen to believe the hype that the job boards have poor quality candidates – please read this post on the quality of job board candidates that will provide you with a reality check.

I hope you enjoyed round 2!

How do you value a database? I say that the value of a database lies not in the information contained within, but in the ability of a user to extract out precisely and completely what the user needs.

When talking about the value of a company’s internal candidate database or the online job board resume databases, we must always be aware that their value is more accurately quantified by the user’s ability to find and retrieve any and all candidates who are appropriately qualified for their needs.

There are 3 major contributing factors as to why most sourcers and recruiters inaccurately value the resume databases they have access to.  #1 Talent Mining capability, #2 The search interface/engine, and #3 Believing the hype that the job boards only have desperate, active or poor quality candidates.

#1 Talent Mining Capability

I’ve had the opportunity to work with and train many recruiters who are responsible for sourcing and who come from a variety of corporate (large and small) and agency (contract and executive search) environments, and I continue to be struck by the fact when they first come to my training, the vast majority of them are not highly proficient at Talent Mining. In fact, most are not proficient at all, and many come from environments where they had access to large internal resume databases as well as online job board databases.

There’s no escaping the fact that when you try to use something and you can’t get it to do what you want, there’s a tendency to blame it on the thing you’re trying to use instead of yourself. There’s also the tendency to simply get frustrated and assume it’s no good because you don’t get the results you want or expect. However – not all searches are created equal, and in most cases, the reason why sourcers and recruiters don’t get the results they want is due to user error, or at the very least, unsophisticated usage. If you can’t find it – it doesn’t mean it’s not in there. 

It’s kind of like deep sea fishing – you can go out for 4 hours and not catch a single fish – but just because you didn’t catch any fish, it doesn’t mean there aren’t fish in the ocean. The same goes for resume databases.

#2 The database’s search interface

There are few Applicant Tracking Systems (ATS) that have resume databases that support full Boolean logic (free text or structured field) searching, and fewer still support extended Boolean. Without the ability to create highly configurable, precise and specific Boolean queries, a user is limited to largely non-specific and imprecise results. In fact, with simple and unsophisticated Boolean searching, a user simply cannot retrieve all appropriately qualified, let alone the best qualified candidates available in the database. This is true whether the user is aware if this or not. 

All of the major job board databases support full Boolean, although a while ago Dice made a change to how they process the asterisk (*) and it’s never functioned the same way since. Supporting full Boolean logic is great – but there are many sourcers and recruiters who are not capable of fully harnessing its power – so I would refer you to factor #1 above.

Monster is the only major job board to my knowledge that supports any extended Boolean functionality (kudos to Monster!) – but it’s limited to non-configurable proximity searching in the form of the NEAR operator. Hey, I’m not complaining – I’ll take any extended Boolean I can get. Some people aren’t even aware that Monster supports the NEAR operator, and most people do not know exactly how powerful this Boolean operator actually is – I’ll be writing a post in the near (pun intended) future focusing on the NEAR operator and all of the cool things you can accomplish with it. CRAZY stuff. You’ll have to see it to believe it.

#3 Believing the hype that the job boards only have desperate, active or poor quality candidates

There is plenty of negative hype floating around about how the quality of the candidates on the job boards is declining, or simply has never been good - and a good amount of it is comes from staffing industry thought leaders. I’m not really sure where people get this opinion – I strongly suspect most of it comes from  anecdotal evidence or hearsay. However, sound statistical laws and theories (and coincidentally, my personal experience) support the reality that there are plenty of high quality candidates on the job boards – read this post to learn more. I think the real issue at hand is that most people run basic and unsophisticated Boolean searches which do not enable them to find the best, let alone all of the qualified candidates available in the database they are searching. Read this post on Hidden Talent Pools to learn more.

The next time you hear someone (or yourself!) comment on how poor a particular source of candidates is, please take a second to think – it may not actually be the source itself, but the user’s ability to extract out exactly what they are looking for, or the search interface of the source. And I’ll say it again – just because you don’t find what you’re looking for, it does NOT necessarily mean it’s not in there. The user simply may not have the ability, or the capability (enabled by the search interface) to find them.

With the buzz I continue to see and hear surrounding Twitter, social networks, Internet sourcing (blogs, articles, etc.) and such, it’s easy to look at resumes as dull, outdated, or at least “uncool” when it comes to sourcing and recruiting. I fear there are many people who get blinded by the “shiny object” factor of each and every “next new thing” that will supposedly revolutionize staffing, leading them to overlook the significant and tangible advantages that resumes have over other sources of talent identification information.

The limitations inherent with using Twitter, blogs, Internet articles, LinkedIn profiles and similar sources for talent identification is that they are what I classify as “shallow” sources of candidate information. In most cases, they contain very little information regarding critical candidate variables such as skills and responsibilities, quantity and quality of experience, career history and accomplishments, education, precise location, etc. Many shallow sources of candidate information simply do not provide ANY information regarding some of these details. With little or no information to go on, it is extremely difficult to search for and identify candidates who have a high probability of at least meeting the minimum requirements for your opening, let alone exceeding them.  

Sniffing out and following candidate leads based on shallow information comes with the territory of being a Sourcer or Recruiter, and the thrill of the hunt can be quite satisfying. However, it can be an incredibly slow and time consuming process, as well as result in a significantly low return on time invested. I refer to this as “low yield” sourcing, and its cause is the shallow depth of information available – which cripples your ability to control or even identify critical candidate variables. Searching for talent using shallow information sources is intrinsically limited in that with less information available, the less effective text-based search is. And if it’s simply not there, you can’t find it, let alone interpret it.

In contrast, while certainly not perfect representations of people, resumes do tend to provide what I refer to as “deep” information. This typically includes a work history including career progression, skills and experience (at least to some extent – some resumes are certainly more detailed than others), environment/group/division/project information, education, and precise location. In many cases, resumes contain information detailing specific responsibilities and responsibility level, as well as accomplishments and achievements.

Being able to search for and control these critical candidate variables enables “high yield” sourcing – a high volume of more accurately and appropriately matched results in less time. In the hands of an adept Talent Miner, a resume database that supports full Boolean queries (and ideally extended Boolean) can yield a high volume of results with a high percentage of candidates who have specific skills and experience, specific responsibilities, specific years of experience, specific environmental/project experience, and who live in a specific area. More information enables more effective searching – if it’s there, you can find it and interpret it.

I am certainly not saying that you can’t find great candidates using shallow information sources, nor am I advising that you should not use them. However, I am painfully aware that many staffing professionals and organizations have a tendency to be blinded by (or at least distracted by) every shiny new recruiting tool, source, or method – wanting to believe they will magically help them fill their human capital needs so much better than the “old” ways. Being interested in new sourcing and recruiting methods is a good thing – but it should not be at the cost of forsaking or underestimating more “traditional” methods such as resume database recruiting, which offers significant sourcing advantages. Especially when there are a shocking number of staffing professionals who aren’t even capable of fully exploiting their own internal resume database, let alone a job board database. See this post on Hidden Talent Pools. 

You can run to your shiny new objects – but the first place I’ll head to is my dull old resume databases – and find large quantities of more precisely qualified candidates in less time.