I’ve been asked for a lot of advice on resumes this year. Recently, an old friend stumped me when he asked ‘should you submit your resume in PDF or Word?’ I hadn’t thought about this question in probably four or five years, which in technology is a lifetime.
So I decided to refresh my knowledge and get the best advice possible by going straight to the CEOs, Presidents, and technical experts of the top 5 resume parsing companies: Hireability, DaXtra, Sovren, Textkernel and Hiretual. These five companies power essentially all of the resume parsers that you’ll encounter applying for jobs on corporate sites globally.
They were each very generous with their time, and shared their expert advice on what you need to know before your resume is parsed.
I spoke to these CEOs and experts in the first week of October 2020, and shared excerpts from our conversations below.
After reading their insights, you’ll want to get free resume templates and a free resume review here on Ladders. You may also want to read my in-depth detailed article on the best resume format for 2020, or have an expert look at your resume.
What does a typical Ladders professional, submitting their resume, need to understand about how their resume is parsed?
Ninh Tran, Founder Member, Hiretual: “My advice for professionals when it comes to resume parsers is try to stay away from fancy resumes.”
Robert H. Ruff, Co-Founder and President, Sovren:
“People want their resume to look unique. I can guarantee you that the longer you spend formatting your resume and making it unique, the less likely it is it’s going to get you hired.”
Christine Watson, Marketing Director, DaXtra Technologies: “A resume parser is a technology that uses Natural Language Processing (NLP) to ‘read’ and convert the text of a resume to language a computer can understand. Resume parsers automatically extract and analyze resume data so the information is able to be categorized, coded, sorted and searched over by the recruiter.”
Steve Kenda, CEO, HireAbility:
“Your best bet for parsing is to use Word or PDF format. Both of those are readily convertible into a text structure that lets the parser do its thing.”
Mihai Rotaru, Head of R&D, Textkernel:
“First of all, make a clean resume. Try to not be very creative. So I do know that all of the parsers struggle with designers’ resumes. Designer resumes are the ones that it’s the hardest for us to parse. Don’t go too designer on it.”
What are common errors people make in their resume formatting or structure?
Steve Kenda: “I think the two biggest ones are not using your contact information early on in your resume. Most of the parsers are looking to be able to identify a name and address, a phone number, to know that, one, they’re actually looking at a resume. And two, they’re starting in the right order.
If they’ve got the contact information early on and have not included a header or footer, the parser just clicks right away and gets started and makes it just a whole lot more orderly.
The other is, people are throwing in jpegs of themselves, pictures, or tables or graphs. That’s going to complicate the process. The easier you can make it for the parser to perform what it’s doing, the more likely you are to have success with your data being more readily found when it comes to search time.”
Christine Watson: “Avoid complexities. The simpler, the better. A straightforward MS Word document with system font text is best.
Include your city, state and postal code so this is picked up in searches. As recruiters typically search by location.
Don’t abbreviate your name.”
Can you discuss how bar charts, graphics, and other visual communications are handled by parsers?
Robert Ruff: “They’re ignored.
Visual information is not going to be understood by the machine. And what you have to think about is, I actually have two resumes. I’ve got a resume that the computer needs to see and understand, and I’ve got a resume that I want to impress a human being with.
It’s okay to have a reference to the fancier resume in your plainer resume. It’s also okay to put information in graphics, as long as it’s not information that is needed to be understood by the machine.
So, I’ll give you a ‘for instance’. I have seen a few resumes where people do not give us the name of the company. They give us the logo of the company. That logo can be read or understood by human, but a parser is not going to get the text from that.”
Minai Rotaru: “We actually do not handle those. Some of the resume generators – there are a lot of websites that generate resumes – those might actually put out resumes with visual bars. At Textkernel, we currently do not process those at all.”
Ninh Tran: “Usually they’re omitted. Graphics and images that are meaningful to the human eye are not meaningful to the parser.”
Christine Watson: “Bar charts, graphics, and other visual communications are not handled well by parsers.”
Why do HR departments use parsers?
Christine Watson: “Parsers are automated, quick and accurate tools that are able to handle large volumes of resumes, breaking them down into a format that can be stored and searched over. A parser takes on rote tasks freeing up time for the recruiter to concentrate on other duties that are more human-focused, like one-on-one interviews. It’s a tool that’s used to benefit HR departments and recruiters by increasing the speed of matching candidates to jobs.”
Robert Ruff: “Nobody can do searching and matching to find out who you are, what you’re best at, until we’ve turned your resume into plain text, and then hopefully extracted the information accurately.”
Steve Kenda: “Whereas in the old days, the resumes would just come in and they get stored in a searchable text database, and when someone ran a search, they were just getting a lot of information that they really didn’t need. So instead of getting 30 resumes now when a search is run, they get three because you can have tagged data.”
Ninh Tran: “HR departments use parsers because they don’t have time to process a thousand resumes at once. It’s a problem of too many applicants, too little time.
Resume parsing is deconstructing whatever file that your resume is in. Then we convert that into raw text.
They want to have your data, your title, your location, your email, all in their nice fields inside their tracking systems that is central, because it makes it more searchable. So let’s say that they want to find, hey, how many candidates do we have with Python in San Diego? They want to index your data in a more meaningful way.”
Mihai Rotaru: “HR departments need to take the information that you have given to them and put it in their applicant tracking system (ATS) so they can track your information. And, especially when they have many applicants for a position, to be able to search.
This transformation from a format that it’s easy to read by people to a structured format that is used by computer is the actual parsing. And it’s crucial for being able to be found by recruiters.
So if you have a recruiter that needs to screen 200 CVs, they search for the information rather than review each resume. So parsing is essential from that perspective.
Another thing that parsing helps is to extract information that might be implicit in the document. So for example, the number of years of experience that you have, or your experience in a certain domain, how much you’ve used a skill, other meta information, like, ‘Do you have international experience? Have you changed many jobs?’ and so on. This type of information can be automatically derived and inferred from your document.”
Should I submit my resume in MS Word, Google doc, or PDF? How do parsers manage these different file formats?
Steve Kenda: “MS Word is good. Google Doc’s good. Text or RTF is good. What you want to avoid are OCR (Optical Character Recognition) situations, jpegs or pngs, or any other picture type formats.”
Robert Ruff: “So, obviously PDF is the big issue. PDF has always been a problem. People don’t understand that PDF is not a text format. It’s actually an image format. And what I mean by that is that there is no concept underneath a PDF, inside of it, of text. There is literally just specified fonts and where those characters, using whatever font, are placed like a pixel on a picture.
So, to extract text from a PDF is actually part science and part guesswork.”
Ninh Tran: “Any of those would work. MS Word is probably the best. Or Google Doc.
Because PDFs are locked usually, we have to have a PDF reader, an OCR system, to read it.
So if I were advising a friend, I’d say MS Word or Google Doc.”
Chrstine Watson: “The best way to submit your resume is in a simple word processor format like an MS Word document or a Google doc. PDFs are a bit more complicated because they can be saved a couple ways – as a text document or as an image document. With text, there is no problem. It’s the image document that can be tricky. Although some systems implement Optical Character Recognition (OCR) tools that can be integrated with a parser and could technically work, optically reading the document then parsing that text, it is not as common and is not worth the risk of your resume not parsing. Therefore I suggest not submitting an image PDF at all.”
Mihai Rotaru: “Parsers usually have converters and transform your resume into HTML and then into text. Most of the machine learning works directly on this text. Our experience is that sometimes MS Word documents are easier to process, especially when it comes to tables, because the table information is retained in the document. While for PDFs, the table information is implicit, so that can be a bit more difficult.
So MS Word documents, especially if you use tables, are better.”
How do resume parsers handle borders, horizontal lines, tables and text boxes?
Robert Ruff: “Text boxes will either show up in random places, or not at all. Don’t use them.
Tables should not be used. If you want to put information that’s in a tabular fashion, you should do a poor man’s table. Use tabs and spaces and not the actual table object in Word.”
Christine Watson: “Borders, horizontal lines, tables and text boxes are complications the parser will need to deal with. Keep it simple and avoid all of these. The parser is looking for text, words and sentences. Graphics only get in the way.”
Mihai Rotaru: “Parsers do not really pay that much attention to the lines, all these visual clues like horizontal lines and so on. I do know that the more tables and the more columns you use, the harder it is to parse those documents.”
Should I use one column or two on my resume?
Christine Watson: “Keep things as simple as possible. Although it is possible to parse two columns, it is more difficult. One column is preferred.”
Ninh Tran: “One. If your objective is to be parsed by these resume parsers, make it simple.”
Steve Kenda: “Just one. Well, the pdf converters are good, but I still call it an evolving technology for handling columns. It’s just an awkward structure for pulling out experience.
My recommendation is you only use one column. I think it will become less of an issue over time. I’m just saying for right now, one column only.
Columns don’t make it any easier for the parser. So you’re not helping yourself when it comes time to the applicant tracking system that’s being used by whomever you’re applying to.”
Mihai Rotaru: “One column guarantees that your parsing is not affected. Two columns can choke some of the parsers.”
Robert Ruff: “Do not use columns. Do not use fancy tables of data. Keep that simple.”
Should I use a template I found online?
Robert Ruff: “Don’t use anybody’s template.
If you take the built-into Microsoft Word resume template that’s been there for 30 or 40 years, it is the least readable template out there. And the reason is, the left hand column is intended for you to put the date. And then to the right of that, in the table, you put the information about each job.
Problem is, visually, those dates align with the data that you put to the right, because you made it that way. But logically, all of the dates are stacked on top of each other in a column. When a parser looks at it, it will find just a whole list of dates first, then followed by all the other information.”
Christine Watson: “Templates can be complicated with headers, footers, graphic elements and so forth. They are unnecessary. Resume parsers best work with straightforward text.
That said, and having a degree in design myself, I know how important resume design is to creatives. For creatives, I recommend having two resumes. One with your preferred design, to present as a portfolio piece or in an in-person interview and the other, a straightforward Word Doc to submit online.”
Ninh Tran: “You can use a standard template. Don’t use one of those super fancy ones that a lot of designers often use.”
Mihai Rotaru: “Layout’s actually quite important. It’s also important for people. Layout – you actually use it so people, when they get a quick look at your resume, they can find something. And those type of signals, for example, indentation when you describe your work experience, that helps both the humans and also the parsers to realize, ‘Okay, this is a whole unit for me to understand.’
In general, indentation helps. Consistent font usage also helps in that perspective. Spacing is important.
Absolutely headers for the sections. Headers for sections are essential. ”
Robert Ruff: “It’s important to segregate the data in your resume by using a header for it.
So, if you’re going to have a summary, say ‘professional summary’ or ‘summary.’
If you’re going to tell your work experience, say ‘work history’ or ‘employment history.’
Whatever comes to your mind first is probably right. If you sat there and thought about something unique, it’s probably not worth using. I think my favorite header of all time that was not understandable was a user that had a section called ‘Community Infusion’. Which I thought was a coffee.
But what they were saying was, ‘These are volunteer organizations that I’ve worked with.’ Okay. Well, how about ‘Volunteering’ as a header?”
Can I use lots of different fancy fonts?
Mihai Rotaru: “Yes. It’s no problem to use fonts as long as you’re also consistent. As long as people can see it, then the parser usually can also see it.”
Steve Kenda: “Yeah, fonts aren’t an issue.”
Robert Ruff: “Yes. Unless it is a PDF, in which case it will mess it up.
So one of the things we see on PDFs is that people will put “Marc Cenedella,” and they’ll put the M an inch high, and the rest of the text will be normal size. That M, I can guarantee, will come out separated from the other letters.”
Ninh Tran: “You should put standard fonts on your resume. Some fonts, especially if you are using paid fonts, they don’t always get captured during the conversion.”
Christine Watson: “Stay away from fancy fonts. It’s best to use system fonts that are compatible with any system used.”
How should I handle contact information?
Steve Kenda: “You’re best off if you can keep your contact information out of headers and footers by using a traditional tombstone layout.”
Does keyword stuffing hurt my resume?
Steve Kenda: “I don’t think so, unless they’re not relevant, in my view. I think the more specific you are to the job you’re applying, the better off you are.”
Christine Watson: “Keywords are important and repeating them plays a factor in ranking, but use them wisely. A small part of how a parser ranks resumes is dependent upon the number of times a keyword is used. But the other part of the ranking process deals with the context around the keyword. Yes, a parser can read and understand context! The other part of the ranking deals with how recent the job title or skill is. The more recent, the more relevant. “
Robert Ruff: “Not necessarily, but it’s also not going to help. So, if you can honestly put keywords into your job and education histories, that helps.
The other thing that I’ve seen is, people trying to do tricks, like repeating the same skills 500 times, and then one-point white text on a white background. Those resumes will be deleted because they come up on every search, and the recruiters will be like, ‘Oh right. You know what? I’m not going to do this every time. I’m going to delete you out of our system.’”
How do parsers handle dates?
Steve Kenda: “So it’s fairly straightforward… the date that you use for your history, your job experience, they all handle the dates in all different formats. The five primary vendors are all working internationally. And so they’re used to different date configurations. They can pick up the phone numbers and the zip codes and the postal addresses from all over the world.
So those issues are not a problem.”
Robert Ruff: “Let’s say that I’ve been with IBM for 15 years, and I’ve had four different jobs there. It is preferred to say ‘2005 to present, IBM Corp’. And then say, right below that, ‘2015 to present, executive vice president.’ And then list the other jobs.
Because of the overriding ‘I was there for 15 years’ first piece of information, we will understand that all of those date ranges below it go with IBM. If you don’t do that, then you need to put ‘IBM’ in every single job, so that we know, ‘Oh, that was another job at IBM.’
We call it nesting. So, if you’re going to have jobs that are all at the same company, nest it under the overall range. Give us that first-to-last year date range and tell us the company.”
Can I use headers and footers?
Christine Watson: “Avoid headers and footers. They are unnecessary elements when it comes to parsers and could hinder parsing of a resume.”
Robert Ruff: “No. That is such an outmoded concept. It’s terrible.
Back in the old days, if you printed out a paper resume and it was eight pages long, or any document, it was important to know what pages in what order. So you had headers and footers to tell you what the document was, and what the page number was. I can assure you that the machine knows the pages without you giving us headers and footers.
And the problem with the headers and footers there, that text will come out interleaved with all the other data, and it will look bad. Recruiters hate that.”
Ninh Tran: “If you use the standard format, it’s fine. With more customization, each customization widens the margin of error.”
Steve Kenda: “In the conversion to text, whether it’s the PDF or the Word doc, when you convert into the text, you’re getting something that’s getting repeated. If it’s a true footer, it’s breaking up the rest of your document with irrelevant info.”
Can I just submit a screenshot of my resume?
Steve Kenda: “Stay away from that.”
Christine Watson: “Never submit a screenshot of your resume. Same goes for an image PDF. Although some systems implement Optical Character Recognition (OCR) tools that can be integrated with a parser and could technically work, it is not as common and is not worth the risk of your resume not parsing.”
How should I name my resume file?
Christine Watson: “The best way to name your resume would be with your name — first name and last name, maybe followed by ‘resume.’ Kind of an industry standard recommendation.”
What is something that our typical professional at Ladders doesn’t realize about resume parsing that might be important?
Christine Watson: “A parser doesn’t care how your resume looks, it is only looking at text, words and sentences. It can contextually understand what you’ve written and this is in part how you are ranked as a qualified candidate. “
Steve Kenda: “Well, the dilemma really is people still view the resume as an opportunity to stand out, or distinguish themselves, or individualize. I mean, nobody wants to just be a fill-in-the-box, LinkedIn situation where, here’s my profile and it looks like everybody else’s profile. And people think of their resume as a way to stand out as a person, to keep the human element in technology. So they’ll try to craft their resume in a unique way that personifies them.
And to that extent, okay, I get it. I understand the sentiment and everything. But from a parser’s perspective, you just want to keep it fairly simple and straightforward. That’s the easiest way to make sure your information is accurately collected.
Robert Ruff: “Another thing that we see that’s related to PDFs that’s happening more and more, is people are using their LinkedIn profile as a resume. I agree that the content is there, but LinkedIn has, for several years now, been on a campaign to make their PDF profiles not accurately readable by parsing software. Which is their right to do, but it’s a disservice to their users. And it’s a mistake for people to then use their PDF LinkedIn profile.
Because although we stay up on those changes, in general they aren’t going to parse as well.
I think we saw at one point they had made 40-something changes in 100 days to make the resumes hard to parse out of those profiles. So that’s a mistake.”
[MC: Please note, Readers, that our resume templates take into account, and avoid, these common template errors.]
MC: And, why are they making so many changes?
Robert Ruff: “Well, what they’re trying to do is there was this big lawsuit that did not go their way. And the lawsuit pretty conclusively established that the candidates own their data on LinkedIn.
I don’t have a problem with LinkedIn trying to keep people from taking that data off of LinkedIn. I don’t think it’s unethical, but I think it’s unfortunate. And I think the victims of it are their customers who don’t understand that those PDFs are intended to be very inaccurate when they’re parsed now.”
Robert Ruff: “The other thing is, if you’re going to have multiple versions of your resume, it’s really hard to do when they all live on the internet. And people are like, ‘Well, I don’t know. Who are you? Are you this guy, or that guy, because you keep telling me different stuff.’”
Overall, any final advice for professionals as they write their resumes?
Robert Ruff: “I will tell you that in every case, the cure for the bad resume is to tell less, more powerfully.”
Christine Watson: “Keep things simple and clear. Write with honesty and use keywords like skills job titles wisely. List your most recent jobs at the top in descending order.”
Thank you so much to our friends in the industry at Hiretual, Textkernel, Sovren, DaXtra and HireAbility for their terrific help in understanding more about how resumes are parsed with these phenomenal insights!