How to tailor your CV for both ATS screening and human recruiters.

Past the ATS, the first human pass is fast. On a role with hundreds of applicants nobody reads your CV top to bottom the first time round. They check a handful of things and decide whether to come back to it.

In those 30 seconds they scan for:

What they look for

• Current title and company (top of the CV)
• Years of experience and career progression
• Whether your most recent role matches the role they're hiring for
• One or two named achievements with metrics
• Signal that you understand their kind of work, language, scale, domain

What humans don't care about: keyword density, fancy formatting, length beyond two pages, decorative graphics. The ATS and the recruiter want different things, but you can satisfy both with the rules below.

1. 01Single-column layout. No tables, no columns. ATS parsers process text top-to-bottom, left-to-right. Multi-column CVs scramble that order.
2. 02Standard section headers. “Experience”, “Education”, “Skills”, not “My Journey” or “Toolkit”. ATS systems look for these exact labels.
3. 03Save as .docx or .pdf. Most ATS support both. Avoid .pages, .odt, or image-only PDFs.
4. 04Action verb + metric in every bullet. “Led”, “Built”, “Reduced”, “Grew”, followed by a measurable outcome. Both ATS and humans reward concrete achievement language.
5. 05Match the JD's exact phrasing for skills. If the JD says “Kubernetes”, don't write “K8s”. If it says “cross-functional”, mirror that exactly.
6. 06Parseable date format. “Jan 2022, Present” or “01/2022, Present”. Avoid fancy date formats with abbreviations the ATS might miss.
7. 07No images, headers, footers, or text boxes. ATS parsers strip these or misread them. Your name and contact details should be plain text at the top of the document.
8. 08Quantify impact wherever possible. “Reduced p99 latency from 4s to 800ms” reads better to a human and carries more matchable terms for the parser (latency, p99, performance) in a single line.
9. 09Group skills by category. Don't dump a 50-item comma-separated list. Use sub-headers: Languages, Infrastructure, Leadership.
10. 10Keep length to 1-2 pages. ATS doesn't care about length, but humans do. A first pass measured in seconds favours clarity over completeness.

The STAR method (Situation, Task, Action, Result) was designed for behavioural interview answers, but it's the single best structure for CV bullets too. A STAR bullet contains scope (Situation/Task), an action verb (Action), and a measurable outcome (Result). All three give the parser something to match on and give a human something to remember you by.

Three before/after examples, same principle across engineer, EM, and PM roles:

Notice what changed and what didn't. Every strong version invents nothing: the team sizes, the dates and the percentages were all available to the person who wrote the weak one. What got added is structure. A verb at the front, the scope it happened at, the outcome in numbers, and one line about how the work was done rather than only what was delivered. If your bullets already contain the facts, this is an editing job, not a writing one.

Everything above works with a text editor and an afternoon. This section is about Calibrd, so skip it if you came for the rules.

Most CV tools give you one score. We give two, recruiter readability and ATS screen, and the interesting part is the distance between them. A CV that reads as 88 to a recruiter and 64 to a parser is not a “72 average” CV. It is a CV that a person would like, filtered out before a person ever sees it, and averaging the two hides the only fact worth acting on.

Three things make our approach different from generic ATS scanners (Resume Worded, Jobscan, etc.):

1. 01JD-aware, not generic. Every gap and edit is tied to the actual job description you're targeting rather than a checklist of universal “good CV” rules. The same CV scores differently against a Senior SWE at Stripe vs an EM at a Series-C health-tech startup, and Calibrd's scoring reflects that. Paste the JD; we read it directly.
2. 02The gap is named rather than buried. When the two scores diverge, the report says how many points your CV isn't claiming yet and gives a one-sentence diagnosis of why. You get the specific mismatch instead of a generic rubric.
3. 03Bullet rewrites that keep your facts. Every edit leaves your real numbers, team sizes and timeframes alone and changes the framing to match what the posting asks for. It won't invent a metric or hand you a skill you don't have. That's the step that turns the weak examples above into the strong ones, and it's the paid part.

The free scan gives you both scores and the specific edits. For a lot of people that is the whole useful part, and it is worth running before you decide whether anything else here is worth paying for.

Seen most often

• Tables for layout. ATS parsers may misread cells, dropping content entirely or stitching it incorrectly.
• Custom or display fonts. PDFs embed fonts; ATS may struggle to render or misinterpret. Stick to system fonts (Inter, Arial, Calibri, Georgia).
• Skill “soup.” A 50-item comma-separated skills dump. Both ATS and humans want grouped, prioritised skills tied to the role.
• Identical CV for every application. The single biggest score lift comes from tailoring keywords and reordering experience to match the specific JD.
• Sentence-fragment bullets. “Database design.” gives the ATS no signal and the recruiter no context.
• Buried current role. The first thing both ATS and humans see should be your most recent role with a clear title and dates. No experimental layouts.
• AI bullets that read optimised but not credible. The tell is when every bullet has the same polished shape and none of it sounds like your actual work. AI is useful for translating your experience into the job's language, but the final pass has to add the specifics only you would know: the tools, the constraints, the trade-offs, the outcomes that were messier than the bullet suggests. Skip slop like “results-driven team player.”