Skillfishing is when a candidate's resume or interview answers present far more skill than the person can actually deliver. You verify skills by testing work instead of claims: put a short, job-relevant assessment or structured screen at the top of the funnel, then confirm it in a live exercise before anyone gets an offer.
The behaviour is old. What changed is the cost. A generative AI tool can rewrite a resume around the keywords in your job description in seconds, draft a cover letter that mirrors your values page, and rehearse a candidate on likely interview questions. None of that is fraud on its own, but it weakens the signals recruiters have relied on for years.
How common is skillfishing?
More common than most hiring teams assume. In GCheck's 2026 Trust in Hiring survey of 1,500 US adults who had applied for jobs in the previous 18 months, 93% admitted at least one form of embellishment or misrepresentation. The most common were exaggerating expertise in a skill (61%) and inflating the scope of previous roles (59%).
The damage lands after the offer. A hire who cannot do the work slows the team, takes up manager time in coaching, and often ends in a replacement search, which means paying for the same hire twice. If you want a rough number for your own team, the cost-per-hire calculator is a good starting point.
Why can resumes and unstructured interviews not catch it?
Resume screening, human or AI, checks whether the right words are present. If the candidate has tuned the document against your JD, the words will be there. Keyword matching has long been a proxy for skill, and skillfishing breaks the proxy.
Unstructured interviews do not do much better. A conversational "tell me about a time" interview rewards fluency and preparation, and both can now be coached. Reference checks help a little, but a reference usually confirms that someone held a role, not how well they did it.
All three methods evaluate claims. The fix is to evaluate work.
Where should verification sit in the funnel?
Earlier, not later. The instinct when trust drops is to add rounds at the end. That is the expensive option: you have already paid for sourcing, screening and several interviews before you learn the skill is missing, and genuine candidates wait longer.
A better pattern is a short skills check before any human interview time is spent. A 20-minute task that mirrors the job tells you more than a 45-minute conversation about the job, and it scales.
Four practices help. Replace self-reported skill lists with a short assessment based on a real task. Run structured first-round interviews where every candidate answers the same job-specific questions against the same rubric. Use live, observed exercises where the skill can be shown in real time, such as coding. And ask references about measurable outcomes rather than impressions.
Do skills tests hurt candidate experience?
A badly designed one does. A good one is short, clearly relevant, explained up front and applied to everyone. Candidates who have the skill often prefer it, because it is a chance to show the skill rather than compete on resume polish.
Keep the human decision in place. Use verification to decide who gets interview time, not to send automated rejections. That keeps the process fair and easier to defend under rules such as New York City's Local Law 144 and the Illinois Human Rights Act amendment on AI. We cover the wider question in Should AI Be Allowed to Reject a Candidate on Its Own?
How GetHirePlus helps verify skills
GetHirePlus is designed so skills are checked before a recruiter or hiring manager spends time on a candidate. Candidate assessments cover MCQ, coding with a live compiler, puzzles, and open-ended questions, and they are proctored, so a resume claim can be tested in the first step. AI phone interviews ask every candidate the same job-specific questions from your own question bank and return a transcript with per-question scores. For engineering roles, live coding interviews let you watch the skill being used.
Everything lands in the applicant tracking system, where shortlisting stays a human decision made on evidence.
Sources
GCheck, 2026 Trust in Hiring Report (press release, 24 March 2026)
NYC Department of Consumer and Worker Protection, Automated Employment Decision Tools