Sourcing

Long Boolean Is Dead: Why Shorter, Sharper Search Strings Get Better Results


Recruiters like very long Boolean search strings. But the reality is that longer doesn’t always mean better when it comes to what your searches turn up.

You need to look no further than the rise of “Boolean builder” tools – software-enhanced thesauri that generate long lists of similar or related search keywords.

Here’s an example many will be familiar with: the long Boolean string used to mine Google’s depths to find resumes. In case you need a refresher, the template looks something like this:

intitle:cv OR intitle:resume OR intitle:vitae OR inurl: cv OR inurl:resume OR inurl:vitae -job -jobs -sample -example filetype:pdf OR filetype:doc OR filetype:docx OR filetype:rtf

Does it work? Technically “all searches work,” but not all work equally well. And there are many variations on the template above…

Even just finding the best way to find a resume can be a task!
Even just finding the best way to find a resume can be a task!

 

But they all rely on the same idea – explicitly naming all of the possible results you could expect to find, to get every possible result without missing any, in one search.

Logically, it would seem this approach to searching would give us better control over the results. And the truth is, searching for what you expect to find is an excellent idea. In fact, the resume search templates relying on long OR lists used to work very well.

However, this is no longer true. The state of search has changed, and recruiters need to update their techniques to adapt.

 

Lesson #1: Less is more

Don’t use long OR statements on Google. If you do, you may get fewer results. If you need to search for alternative terms, search for each in turn.

I will go as far as to say it’s best not to use OR in Google at all. Why?

Let’s take a look at a simple example and compare the search results for a string that spells out synonyms vs. a string that doesn’t.

Long Boolean string: resume filetype:PDF (CPA OR “certified public accountant”) (PwC OR PricewaterhouseCoopers) tax houston

Short Boolean string: resume filetype:PDF CPA PwC tax Houston

Here’s what the search results look like, side by side:

lbid_screenshot2

In theory, the string with ORs (the one on the left) should provide more (and more varied) results, but this approach is outdated!

First and foremost, it’s important to know that Google already understands most common synonyms, abbreviations, and other variations, and includes them automatically. If you search for CPA, your results will also include “certified public accountant.” While Google doesn’t yet automatically understand the relationship between CPA and FASB in our searches, we can almost always drop the most obvious synonyms and variations.

BONUS: Download our free guide listing 10 sources of candidate data you’ve never thought of mining!

Reducing the number of terms makes strings easier to read and understand, and given Google’s search term limits, shorter templates give us more room to include other important details!

Secondly, for Google searches, less is quite literally more.

As we know, Google never provides over 1,000 results. Even when we are aware there are more results available, Google often returns only what it thinks are the most relevant – often 600, 300, or even fewer.

And, counterintuitively, the longer and more complicated our searches become, the smaller the number of results we receive. It seems that Google is increasingly “disliking” complex searches and does less than stellar job providing relevant search results. You can think about it this way: when we give Google the space to make sense of the query (versus controlling the query with overly complex syntax), Google works harder.

Over the last few years, Google has also made advances in the area of semantic search. Last Fall Google told us that their new semantic, machine-learning based component called RankBrain is now the 3rd most important factor in ranking. These changes help to explain why long and bulky Boolean search templates are getting less useful, and providing fewer relevant results, compared to searching in a simpler fashion.

So, if the old methods are outdated, what do we do now?

 

Lesson #2: Make searches clear and concise, and avoid complexity.

While I’ve demonstrated that shorter strings work just as well – if not better – than longer and more complex ones, I must also mention the importance of imagining what you are trying to find before you even start your search. If we think carefully about the information we are seeking to find, then our search results are likely to be more accurate.

In the earlier examples, we have used the filetype: operator to look for PDFs assuming that resumes are in this format. But they could also be Word documents, or on a website, etc. By searching for only specific formats, it’s possible we could miss out on some people that may be a match for our roles.

“If we think carefully about the information we are seeking to find, then our search results are likely to be more accurate.”

What kind of information lies within a resume that we could search for? For starters, most resumes will include:

  • Email address details – gmail.com hotmail.com
  • Phone numbers including area codes or words like cell, mobile or Skype.
  • Phrases like – “reason for leaving” “references available” “home address” “work history” “dates of employment.”

And that just resumes! The rise of social media has completely transformed the online landscape. We are seeing fewer stand-alone sites or pages where a person may post their resume and more and more social profiles – a great change for recruiters, because, arguably, “passive” candidates would never have had an online resume in the first place. But almost everyone is online in some form.

We can apply the same approach of defining our expectations in advance just as easily to profiles (or any other type of information) as to resumes. If the information is online, how will it look? Are there other, less obvious variations on what we were looking for?
Take a look at some comparisons of strings that use a few keywords, versus phrases carefully chosen for relevance:

lbid_screenshot3

(Example 1)

 

lbid_screenshot4

(Example 2)

No question about it, Google’s Boolean search can produce fantastic search results. But the search style we use needs to be different than ten years ago.

This is true beyond Google as well. Though the semantic offerings of sites like LinkedIn pale in comparison to Google, there are still downsides to writing long, complex strings, even when they do produce more results.

First, maintaining the right balance of ()s and “”s to correctly execute a long, complex search is not easy, and missing just one will give you the completely wrong results – which you may not initially catch. It is also very hard to quickly adjust your search, based on the results.

Often the best results come from an iterative, organic approach – start simple, review the first results, add or remove keywords to filter out incorrect matches. Take, for example, a search which names a long list of possible target companies (you could easily include several dozens in one search). If you see many incorrect results, do you know which keyword or phrase is producing them?

It takes much more time to troubleshoot and adjust a long search. Worse still, since your targets and the competitive landscape change over time, trying to save and re-use these painstakingly-built mega-strings requires ongoing maintenance, which nobody has time for.

 

As the pace of change in technology, tools, and the internet itself only continues to increase, we must keep reviewing our search strategies and templates to keep up.

Recruiters who only rely on the same old saved strings from classes they’ve taken, blogs that repeat that content, or “Boolean builder” tools are missing out. The best ways to search on Google (and beyond) keeps changing. And the top recruiters will stay at the top of their game only through regularly reviewing each element of how they search.

Stop chasing the low hanging candidate fruit.
Download our guide to 10 sources of candidate data you’ve never even thought of mining!
 

DOWNLOAD THE GUIDE FOR FREE.