The Invisible Wall: What Actually Happens to Your CV Before a Human Sees It
My first three applications after my career break went nowhere. Not rejected. Not even acknowledged. They disappeared into whatever happens to applications when no one is looking at them. I checked my sent folder to confirm they had gone. They had. I checked the job postings to confirm they were still live. They were. I refreshed my email for three days and then accepted that something had happened that I did not understand.
Here is the uncomfortable part. I spent fifteen years on the hiring side of that process. I was Head of Talent at a large telecoms business and then People Director at a FTSE 250 retailer, where my team processed over 40,000 applications a year. I bought several of the systems I am about to describe. I sat in vendor demonstrations and nodded along, thinking about time-to-hire metrics and recruiter workload, not about the woman on the other side of the process who would never know her application had been evaluated in milliseconds by software she could not see.
Then I took two and a half years out to care for my mother, came back, and became that woman.
What I learned, sitting on the wrong side of a process I helped build, is that most returning professionals are applying for jobs without understanding the single most important fact about modern hiring: a piece of software makes the call before a human ever sees your name. You cannot fix what you cannot see. So let me show you what you are dealing with.
The hiring process you remember no longer exists
If you left work before 2018, 2019, or even 2022, the process you are returning to is not the one you left.
The one you left worked roughly like this: you sent a CV, a person read it, that person decided whether to call you. The pile was large and the attention was partial, but it was human. A gap in your employment history was something a person noticed and made a judgement about, sometimes unfairly, sometimes not, but at least with the full picture in front of them.
The one you are returning to works like this: you submit an application online. Before it reaches a person, it passes through an applicant tracking system, or ATS (software that ingests your CV, parses it into fields, and scores it against the job description). If your score clears a threshold, a person sees it. If it does not, no one does. You will never know which happened. You will receive either an automated rejection or nothing at all.
According to research from Jobscan, 98 per cent of Fortune 500 companies use ATS software to manage their hiring. Many mid-sized employers do too. The majority of the roles you will apply for are gated by one.
And the gap penalty is built into most of these systems by default. An ATS reads your CV by looking for dates of employment and calculating the intervals between them. A gap of six months or more is typically flagged. The system does not know why the gap is there. It does not ask. It applies the flag, the flag affects your score, and if your score drops below the threshold, the recruiter does not see you.
Studies comparing identical CVs with and without employment gaps consistently show substantially fewer interview callbacks for the version with the gap; some estimates put the reduction at around 45 per cent for gaps of a year or more. The gap itself, on a well-written CV presented to a thoughtful human reviewer, would often not be disqualifying. The gap as processed by an ATS frequently is.
The AI layer on top
In the past three or four years, a second layer has been added at many larger employers. AI-powered screening tools go beyond keyword matching to assess career trajectory, language patterns, and predicted performance. Some analyse video interview responses. Some generate candidate rankings that the recruiter uses as a starting point rather than reviewing the full application pool.
These tools learn from historical hiring data. That sounds neutral, and the vendors would like you to think it is. It is not entirely. A system that learns which candidates a company has historically hired will learn to favour candidates who look like those people. If the company has historically hired people with linear career trajectories, recent employment, and no gaps, the AI will learn to deprioritise candidates with non-linear trajectories and gaps. It is not making a judgement. It is pattern-matching. The effect on you is the same either way.
This is not malicious, and I want to be precise about that, because I sat in the meetings where these systems were configured. Most of the recruiters and HR teams running them have not thought carefully about what their configuration does to candidates with career gaps. They used the default settings, or copied the setup from a previous implementation, or took the vendor’s recommendation and did not interrogate it. The system that filters you out was not built with malice. It was built with inertia.
The law is starting to catch up, slowly. New York City has required bias audits of automated employment decision tools since 2023. California introduced regulations in October 2025 requiring employers to disclose the use of automated decision tools in hiring. Colorado’s AI Act came into force in June 2026 with similar requirements. In some jurisdictions you now have the right to ask whether automated tools assessed your application, and to request human review. These rights are new, inconsistently enforced, and not widely known. Exercising them requires knowing they exist, which you now do.
LinkedIn has been making decisions about you too
There is a third filter, and it was the one that caught me most off guard.
I avoided LinkedIn for two years. When I finally went back with serious intentions, I discovered it had been making decisions about me in my absence. My profile still existed. My old job titles were still listed. But the algorithm had quietly moved me towards the back of the queue for recruiter searches. Two years of inactivity had a cost, and the algorithm had invoiced me for it without sending a notification.
LinkedIn is a search engine, and that is the part most people miss. When a recruiter searches for candidates (which many do proactively for mid-to-senior roles, before any job is even posted), what comes back is a ranked list, ordered by an algorithm that weighs profile completeness, keyword density, and recent activity. A profile untouched for two years, with a gap where current employment should be, scores lower than an equivalent profile that has been recently updated and regularly active. The algorithm does not know you were caregiving. It knows you were absent.
The recruiter who eventually called me admitted she had to scroll further than expected to find someone at my level. A former colleague with a nearly identical background appeared above me in her results. Because she had been active. Because the algorithm had kept her warm.
The six formatting choices that get you auto-rejected
None of this is catastrophic, and all of it is fixable, so let us get on with the fixing. Start with the document itself, because an ATS does not read your CV the way a person does. It parses it: extracts the text and attempts to classify it into fields. If your formatting makes that difficult, the parsing fails and your application is scored inaccurately or rejected outright. You submitted a CV you were proud of and heard nothing.
These are the choices that cause it most often.
- Creative section headings. The system looks for standard labels: Work Experience, Education, Skills. If you use “My Journey” or “What I Bring,” it may not recognise them as sections at all, and the content beneath may be ignored. Use standard headings. Save the creativity for the content.
- Tables, columns, graphics, and text boxes. A two-column layout looks polished to a human. An ATS frequently reads it left-to-right across both columns simultaneously and produces complete nonsense. Single column, plain layout.
- Inconsistent dates. The system uses dates to calculate gaps. Omitting end dates, or using only years to minimise the appearance of the gap, backfires: a system that cannot determine an end date may assume the gap is longer than it is. Use month-year formatting throughout, and list the career break as a dated entry rather than an unexplained interval.
- The wrong file format. Newer systems handle PDFs, but not universally. Unless the posting says otherwise, DOCX is the more conservative choice.
- Decorative fonts and styling. Standard fonts, 10 to 12 point, parse reliably. If it is not formatting you would find in an ordinary business document, remove it.
- Missing keywords. This is the most consequential mistake and the most fixable, so it gets its own section.
The keyword problem, illustrated
An ATS scores your CV against the job description by looking for matches. If the job description says “stakeholder management” and your CV says “managing relationships with senior partners,” you have the same skill in different language, and the ATS scores a miss.
Consider a returner I will call Claire. Marketing manager, two years out caring for her father, applying for a senior marketing role at a technology company. The job description asks for “content strategy,” “SEO,” “marketing automation,” “HubSpot,” “cross-functional collaboration,” and “data-driven decision-making.” Claire’s CV, written in the language of her pre-break role, describes “developing content plans,” “search engine optimisation activity,” “email marketing campaigns,” and “working with product and sales teams.”
The skills are essentially the same. The language is not. Her match score comes back at 52 per cent. The threshold most recruiters set sits somewhere between 70 and 80. She is not getting through.
So she makes the specific changes: “content plans” becomes “content strategy,” “SEO” appears alongside the longer phrase, a skills section explicitly lists “HubSpot” and “marketing automation.” Her score comes back at 79 per cent. She is now in the zone, and not a single one of those changes was dishonest. She genuinely did all of those things. The problem was never her experience. It was her vocabulary.
You do not have to guess how your own CV will perform. Tools exist that simulate ATS scoring: Jobscan offers five free scans a month and produces the most detailed keyword breakdown in its category; Resume Worded is a strong free alternative. Take a specific job posting, run it against your CV, and read the results. Run the check for every role you apply for, not once generically, because different employers use different language for similar roles. The check takes ten minutes. It is ten minutes that determines whether a human ever sees your application.
One warning: do not optimise to the point where the document reads unnaturally. You are writing for the algorithm first and then for the recruiter who reads what the algorithm surfaces. A CV crammed with keywords at the expense of clarity will clear the ATS and then underwhelm the person. The goal is a document that reads naturally and contains the right language. Those two things are compatible.
The human on the other side
Everything above is about the automated layer, and it is worth remembering what that layer is for. It is a filter, not a decision. The decision is still made by a person. The ATS surfaces a shortlist; a recruiter reviews it; a hiring manager interviews from it.
I spent fifteen years being one of those people, and I can tell you what the automated facade hides: the humans in this process are mostly busy, under-resourced, and genuinely relieved when a strong candidate appears in their inbox. They are not looking for reasons to reject you. They are looking for reasons to move forward. The candidates who stayed in my memory were not the ones with the most perfect CVs. They were the ones whose documents told a coherent story, clearly enough that I did not have to work to understand it.
Which means you have more agency than the online portal makes it appear. If a company you genuinely want to work for seems to be filtering returners through process design rather than intent, it is entirely reasonable to contact them directly, explain your situation, and ask whether your application can be reviewed by a person. Some will say no. Some will say yes. Most will not expect the request, and the act of making it may itself prompt a human to look.
The invisible wall is real. It was built by people like me, mostly without thinking about people like you. But it has a shape, and a shape can be learned, and once you have learned it you stop walking into it blind. Get the format right. Match the language. Test before you submit. Reactivate the profile. That is the price of admission. Once you are through, you are back in a room with a human being who is hoping you might be the right person, and everything from that point is a conversation you know how to have.
This article draws on my book, Relaunch: The Modern Woman’s Playbook for Returning to Work in the Age of AI Hiring. It covers the whole of the return, from the first honest look at where you stand, through the CV, LinkedIn, returnships, interviews, and the salary negotiation most returners never have. I spent fifteen years building hiring systems and two and a half years locked out by them; the book is everything I learned from both sides of the wall. You can find it on Amazon.