ATS Resume Tips: What Actually Gets You Past the Filter

Jobscan’s 2025 State of the Job Search report surveyed 384 recruiters and found that over 99.7% use keyword filters in their ATS to rank applicants before a human ever opens a file. And yet most advice about resume writing treats ATS as an afterthought – a checklist you skim at the end, after you’ve spent three hours choosing the right font.

That gets the sequence backwards. The filter runs first. If your resume doesn’t clear it, the font is irrelevant.

Why ATS rejects resumes that look fine to you

An applicant tracking system isn’t reading your resume the way a person would. It’s extracting structured data – job titles, skills, dates, institutions – and then scoring that data against the job description. The match score drives where you land in the recruiter’s queue.

Harvard Business School’s 2021 Hidden Workers study (conducted with Accenture across 8,000 employers) found that 88% of companies acknowledged their ATS eliminates qualified, high-skilled candidates whose resumes don’t use the precise wording the system expects. The companies knew. They kept using the filters anyway, because volume demands it. A corporate job posting in 2025 gets an average of 250 applicants, according to Jobscan’s own analysis.

Two things cause most of the rejections.

First: terminology mismatch. You wrote “led infrastructure modernization.” The job description says “cloud migration.” These mean the same thing to you and to any recruiter reading the document. They don’t mean the same thing to a parser doing exact or near-exact string matching. Some modern ATS platforms have improved their semantic understanding, and I’ll admit I’m not certain how consistent that is across Workday, Greenhouse, and Lever – their vendor documentation is vague on specifics. But if you’re submitting to any of the 97.8% of Fortune 500 companies running a detectable ATS (per Jobscan’s 2024 audit), betting on semantic generosity is a bad strategy.

Second: formatting that breaks the parser. Tables, text boxes, headers and footers, multi-column layouts – these can cause an ATS to misread or drop entire sections. Skills listed in a sidebar may never be extracted at all.

The title match is more important than most candidates realize

Jobscan’s 2025 report surfaced a specific finding that surprised me when I first read it: candidates whose resume job titles matched the target role’s title had interview rates 10.6 times higher than those who didn’t. That’s not a small lift. That’s the difference between getting a callback and disappearing into the rejection queue.

The practical read: if you’re applying for a “Senior Software Engineer” role and your current title is “Software Developer III,” the discrepancy may cost you. You can address this without being dishonest. The title on your most recent role stays accurate. But your resume summary, your skills section, and your professional header can all use the exact language the job description uses.

This applies to skills too. Don’t write “proficient in cloud platforms” when the description lists “AWS” and “Azure” separately. Write both. Include the acronym and the full form when there’s any ambiguity – “ML” and “machine learning,” “JS” and “JavaScript.”

Formatting rules that actually matter

A few things that cause measurable problems, based on what’s documented by ATS vendors and widely corroborated by recruiter accounts:

• Use .docx, not PDF. PDF rendering in many ATS platforms is inconsistent. Some extract text cleanly; others lose formatting entirely depending on how the PDF was generated. DOCX gives you more predictable output.
• Single-column layout. Multi-column resumes are visually appealing to humans but frequently confuse parsers. Information in a right-hand column can get read out of order or skipped.
• Standard section headings. “Work Experience,” “Education,” “Skills.” The parser is looking for these labels. “My Background” and “Where I’ve Been” sound interesting and will likely cause the system to misfile the section.
• No tables or text boxes. Content inside these containers often doesn’t parse at all.
• No headers or footers for contact info. Some systems don’t read header/footer zones. Put your name, email, phone, and LinkedIn URL in the body of the document.

Tailoring without starting from scratch each time

The standard advice is “customize your resume for every application.” That’s technically correct and practically exhausting if you’re applying to 30 roles. Here’s what’s actually worth customizing versus what you can standardize.

Standardize: your work history bullets (impact and scope don’t change), education, certifications, and the overall structure.

Customize per application: your professional summary (two to three sentences at the top), your skills list (reorder to match what the job emphasizes), and any phrasing in your most recent role that uses language that doesn’t match the target description.

This usually takes 20 to 30 minutes per application if you have a clean base document. If it’s taking 90 minutes, your base document needs work.

What we’ve seen on LastRoundAI

One thing that’s come up repeatedly from candidates using LastRoundAI’s Resume Builder: when people draft their experience bullets the way they’d talk about their work in a conversation – loose, confident, full of context – the language often doesn’t match what appears in job descriptions. The gap isn’t dishonesty or underselling. It’s just that conversational language and ATS-optimized language follow different conventions. Working through that gap is most of the actual work of resume writing.

This also affects how candidates describe their experience in interviews. If your resume uses one set of terms and you use a different vocabulary on the call, that inconsistency reads oddly to interviewers who have your resume open in front of them. Practicing answers that use the same language as your resume – which you can do with mock interview sessions – closes that gap faster than most candidates expect.

Three things not worth obsessing over

Some resume advice circulating in 2025 and 2026 is overcorrected or just wrong.

Keyword density targets. Some tools give you a “match score” and tell you to hit 80% keyword coverage. I think chasing a score above roughly 60-65% produces resumes that read awkwardly to humans, which matters a lot once you’ve cleared the ATS. Optimize enough to get through – then make sure a person actually wants to read it.

One-page rules for experienced candidates. This came from an era of fax machines and paper stacks. If you have 12 years of relevant experience, cramming it onto one page by using 9pt font and removing whitespace makes the document worse, not better. Two pages is fine for senior engineers and above.

Elaborate visual formatting for non-design roles. If you’re applying for a UX or graphic design position, a visually distinct resume makes sense. For engineering, product, data, and finance roles, it mostly signals that you’ve prioritized aesthetics over clarity. Recruiters at high-volume companies have said this directly – and many of the visually complex templates circulating on LinkedIn don’t pass basic ATS parsing anyway.

If you’re actively applying right now and want to stress-test both your resume and your ability to talk about it under pressure, LastRoundAI’s AI interview copilot can help you connect the dots between what you’ve written and what you’ll say on the call.