Tuesday, March 27, 2018

Why Keyword Scans Don't Work For Either Candidates or Companies when Screening Resumes

The problem with keyword search, when companies are using computers to scan submitted résumés, is that there are just too many variables. Because of all the infinite possibilities, scanners may miss the best candidates.  (This is also the problem with having junior executives who may not be totally familiar with their business.)

Trained and experienced people should be reading résumés, not computers.  The issue is that there are just too many options to describe the same thing.  These variables exist on multiple levels – brands, titles, experiences and companies – the list is endless.

To use an example, suppose a company is screening for a mid-level person to service a package goods account.  Here are possible terms used to describe the experience of working on these products:  package goods, packaged goods, FMCG (fast moving consumer goods), CPG (consumer package goods), and consumer goods.  But they could also be described by their functions – soaps, detergents, toothpaste, health and beauty aids, HBA, food, soup, dessert, confection, confectionary, candy, chocolate, etc.  You get the point – any of these terms is acceptable.  To take it a step further, supposed one has worked on either Procter & Gamble or Unilever, here are variables on those names, all correct:  P&G, Procter and Gamble, Procter, Unilever, Lever, Lever Brothers.  The list goes on.

No person can program a computer with enough variables to cover all the possibilities, including all the hundreds of brand names, which used by themselves would convey consumer package goods.
The number of business-to-business brands or service bands and their nomenclature and descriptions is probably even greater than CPG.  Ditto automotive and car brands, cosmetics or dozens of other categories.

And, of course, most scanning programs only recognize programmed words and cannot give weight to the experience.  While working on Gain Detergent, which is made by Procter & Gamble, may be valuable experience, is it as good as working on Tide or Ariel (Tide in Europe)?  Is Colgate Total better than Crest or Audi better than Volvo?  The answer may have to do with the job being filled.  I remember once early on when I was recruiting, meeting a wonderful candidate who was working on Lincoln.  I introduced him to the agency then handling BMW and got my head handed to me.  “You should know that we are an imported car and we never talk to people who have domestic automotive experience.”  But at least I found out why my candidate was rejected.  With key word scanning, one is never told why they are rejected.

In addition to all the above, there is the proliferation of titles.  In any given industry titles are rife.  Just take digital.  People involved with SEO/SEM may or may not have titles which use those terms.  They can also use other terms: content, engagement, content, UX, user experience – the list goes on and can be endless.  The point is that no programmer can put in all the possible variations and alternatives.  And on top of all this, there is level.  What is a senior at one company may not be so senior at another (I have always been amused by the title senior account executive, which, frankly is meaningless).

Then, there is education.  There are literarily hundreds of wonderful colleges in the country.  Countless studies have shown that where an undergraduate degree comes from has only a small correlation to future success, so companies which program in colleges to their searches are doing themselves an immense disservice. 

Companies that allow themselves to fall into the expediency of keyword scanning may miss their best possible candidates.  The issue is not the brands, brand names, parent companies or product categories.  The issue is always a careful definition of the job, what has to be accomplished and the experience(s) necessary to accomplish the task. Keywords just cannot just determine that because computers are not so intelligent as to be able to interpret the words on a page.


  1. Good words, for sure ... But there is still the issue of AI (Artificial Intelligence), which transcends key words because it automatically connects-the-dots and presumably kicks out a relevant profile. A great tool for very busy or very unqualified hiring executives, but it leaves an awful lot to be desired.

    1. Bill, I guaranty you that there is not a single advertising agency using AI to screen resumes. And probably few companies, if any. It is too expensive and too new to be used for resume screening.

    2. Paul … Since you’re a top-notch agency recruiter of many years, I’ll take your word that what you say is true today. What I will guarantee is that “AI” and “Blockchain” are coming to HR, recruiting, and retention. That is, AI algorithms that aggregate everything about a person from a wide variety of sources, e.g., resume, Linkedin, Facebook, Google, Twitter, YouTube, Snapchat, Instagram, Amazon, TransUnion, and Experian. Blockchain for tracking prospects from A-Z in the process, e.g., e-mail, phone, voicemail, Tweets, text messages, interviews, follow-up, and such. I’m certainly no Stephen Hawking, Carl Sagan, or Steve Jobs, but I’ll tell you … AI and blockchain are coming to human resourcing. Locally, nationally, and globally. Just a matter of time. Happy holiday, Bill

  2. I once hand-delivered a cover letter and résumé to an agency owned by one of the major holding companies in response to a senior-level position posted on the agency's website. Within half an hour, I received an e-mailed request to a submit my résumé electronically instead – a tell-tale sign that an algorithm would be assessing my candidacy. I declined, and have never applied to any of that holding company's agencies since.

    Technology might've improved cars, telecommunications and warfare, but it absolutely ruined the talent-sourcing process.

    1. Sadly, I know this is true. Thanks for sharing.

    2. Anon, I was thinking about your story. One of the things that companies do is to instantly scan resumes into their data base (mostly it is merely the data but is not retrievable). They do this to insure that they do not pay a recruiter for submitting someone they already have (I was once told by an IPG agency that they could not accept a resume from me because they had it in their files already; when asked, they said it did not matter that it was four years prior. They called and hired my candidate and I had to threaten legal action and was then paid.) They also do it to insure that no two recruiters can submit the same person for the same job. In fact, they will only give credit to the first one who submits the resume, even if the second one is the one who causes the interview and subsequent hire.

    3. Thanks for your insight. As it turns out, this was an IPG agency. Still left me with a sour taste in my mouth.

  3. AI if used for this will only compound the problem because humans steadfastly refuse to be modelled accurately or with any degree of predicability by computer systems no matter how much effort and energy is put into creating models. Add language, especially written language on a CV/resume, which is a pretty artificial document anyway and the problem becomes even bigger as outlined in this excellent article. Well written and thank you for some common sense in a lunatic world

  4. This comment has been removed by a blog administrator.


I would welcome your comments, suggestions or anything you would like to share with me or my readers.

Creative Commons License