Boolean Beware: The Case of the Questionable Quotation Marks

As a recruiter I rely heavily on writing Boolean search strings. This has served me well over the last twelve years, so whenever I find a search string that is not working as I think it should in any database, I become very focused on finding out what’s happening. This article has to do with a situation I encountered in LinkedIn Recruiter. Your input is welcomed!

If most of the Boolean search strings you use in LinkedIn Recruiter (the “not free” version) don’t regularly exceed approximately 100 characters in length, then you might want to skip this article and go to the next.

If you do regularly exceed 100 characters in your Boolean search strings AND you build your long strings in Word before pasting them into LinkedIn Recruiter, then this is something you need to know that could make the difference between a successful and an unsuccessful search.

Need a lawyer?

Let’s start with a non-technical search string. For example, when using LinkedIn Recruiter to search for a Patent Attorney with an Electrical Engineering degree or equivalent, below is a Boolean search string I used:

(Juris OR JD OR attorney OR lawyer OR “legal counsel” OR “licensing counsel” OR “corporate counsel”) AND (patent OR IP OR “IP law” OR “intellectual property” OR license OR licensing OR “patent litigation” OR litigation OR prosecution) AND (BEng OR “B Eng” OR BSc or “B Sc” OR Electronic OR BSEE OR “Electrical Engineering” OR BSCS OR “Computer Science” OR BSCE OR “Computer Engineering” OR Physics OR “Mechanical Engineering” OR Engineering)

The above string has 442 characters.

I write strings that are very long, very tight, and very specific in order to get accurate results. Notice I said ‘accurate results.’ I didn’t say ‘perfect results.’ In this world, the word ‘perfect’ should be used sparingly, if at all! There are many ‘right’ ways to write a good string, and unfortunately, the converse is also true.

When using Boolean you are always playing the trade-off game of high quality / low quantity results vs. low quality / high quantity results. For me, long strings usually equate to my definition of ‘accurate results.’ However, at times I am pleasantly surprised when a three word ‘string’ produces the best results.

For strings of 442 characters and longer, I need to be able to visually see ALL the string at one time as one “big picture.” I do this because I usually go through several (sometimes as many as fifteen) iterations of tweaking strings, adding and deleting keywords in order to create the optimum string. The problem is that in LinkedIn, as with other resume databases, I can only visually see about 100 of the 442 characters at any one time in the Key Word field without scrolling back and forth. In the string above, the first part of the string that I can see at one time in the LinkedIn search field without scrolling is bolded and italicized.

The Copy and Paste workaround

To get around this, I began building my search strings in a Word doc and then copying and pasting the string into LinkedIn Recruiter. I started getting unexpected results that were not predicted by the Boolean search string. Not only were wrong candidates coming to the surface, but I would also get different unexpected results when using the same keywords that were also changing up the order in which they appeared in the string. I didn’t understand why.

Little Creatures!

It’s a long story, but bottom line is that I came to find out that there are three types of question marks that computers use (and for you programmer types there are three different ASCII codes associated with each).

Here’s a blowup of what these three little creatures look like:

When you put a key word phrase in quotation marks in Word, like “technical recruiter,” you’ll notice that the beginning quotation marks are like #1 and the ending ones are like #2.

How it all began while working in Taleo

Taleo has a searchable resume database. While working with this database, I discovered it only recognized the #3 flavor of quotation marks. But how do I use the #3 flavor if all that Word produces/uses is #1’s and #2’s?

So I tried an experiment. I built my search string in Word using the Word quotation marks. Then I would go into Taleo and type a single #3 quotation mark in their Key Word field, copy it, go over to Word, and then paste it over all the #1 and #2 quotation marks in Word. And Shazam!! The Boolean string looked like it was working like it should!!

Back to LinkedIn Recruiter

By this time I had begun using LinkedIn Recruiter very little because every time I copied and pasted a string from Word into LI, I would sometimes get unwanted results that didn’t obey the Boolean rules. [Again I need to emphasize here that if you are writing your strings directly into LinkedIn Recruiter without pasting from Word, this won’t be a problem for you. Everything will work just fine.]

Why not give it a shot!

So one day on a lark, I thought I would give LinkedIn a try using the same trick I had used with Taleo. And lo and behold, it started working!

For the Patent Attorney string above, this meant that after I had typed the string into Word, I had to go over to LinkedIn, type in a single quotation mark, a #3, copy it, go back to Word, and paste it 24 times over each and every quotation mark previously typed into Word. Last step was to copy this modified string in Word and then paste it into LinkedIn and then run my LinkedIn search. I quickly learned that if there were 24 original quotation marks and I only replaced 23 of them, it was a train wreck. 100% always have to be replaced.

I asked the folks at LinkedIn about this and they told me that there was no way you could copy a string in Word, paste it into LinkedIn, and have it work. Their explanation had something to do with “when you go from Word and paste into LinkedIn, you bring along too much unwanted stuff over in the process.” Regardless, it still seems to work.

The 1,900 character LinkedIn InMail

I do know when I compose a LinkedIn InMail, I am limited to 1900 characters with spaces (note: all previous references to number of characters include ‘with spaces’), yet when I compose an 1900 character InMail in Word, and copy it into LinkedIn, I always go over the 1900 character limit. Then if I back up, edit the InMail in Word down to 1850 characters or less before copying and pasting, then it works. I’m not over the limit. So it does look like in this case that 50 (1900-1850) characters of ‘something’ is ‘brought over’ or into LinkedIn from Word. I have no idea what it is, but it appears to be there.

The Genius of LinkedIn Recruiter

I consider LinkedIn Recruiter to be the premier and best searchable resume database I’ve ever encountered to date. The super ease of navigation and availability of an assortment of filtering options make me classify this as a work of pure genius!

So who am I to question the work of genius? All I can say is that I have been sourcing and getting candidates hired using this technique for the past year. Is there a glitch or something I’m missing? Entirely possible. So if you know where and/or how this breaks down, I’d appreciate hearing from you.

Finding Waldo

Below are some example searches (notice the different types of quotation marks in each of the examples):

If I type directly in the LinkedIn Recruiter Key Words field, I get 5,730 results.
If I type in Word and paste it into LinkedIn Recruiter, I get 29,584 results.
If I type in Word, replace all the quotation marks with ‘#3’ quotation marks, and then paste the string into LinkedIn Recruiter, I get 5,730 results.
If I type senior technical recruiter directly into LinkedIn Recruiter with no quotation marks, I get 29,584 results.

Conclusions… at least as of today

From the above it would appear that:

both LinkedIn and Taleo only correctly recognize ‘#3’ quotation marks and
that if ‘#1’ and ‘#2’ quotation marks are used, both LinkedIn and Taleo treat them as if they were not there, resulting in undesired results.

**image source: Quinn Dombrowski