Only 23% of Recruiters Take Your Word. Give Each Bullet a Receipt

Your resume used to get read as your own account of your work. Now it gets read as a claim that needs checking. TestGorilla's 2026 recruiter survey found that only 23% of US recruiters, and 15% of UK recruiters, would take a candidate's word that they can do the job. Most of them then spend an hour or more checking each person by hand. You won't win that hour by writing better. You win it by making every claim checkable in a couple of minutes.

What did TestGorilla's 2026 recruiter survey find?

TestGorilla's 2026 survey found that recruiters now treat a resume as unverified until they've confirmed it themselves, and confirming it costs them real time. The US sample was surveyed in August 2026, and 483 UK recruiters were surveyed between 10 September and 7 October 2026.

The headline numbers:

• 23% of US recruiters and 15% of UK recruiters would take a candidate's word that they can do the job.
• 67% of US recruiters spend an hour or more checking a single candidate by hand before they trust them. Nearly 30% spend two hours or more. In the UK it's 73%.
• 75% of UK recruiters believe most CVs are now AI-written.
• 33% of UK recruiters say every profile for AI roles reads the same.
• 57% name verifying whether resume skills are real as their top sourcing challenge.

Read it this way: recruiters aren't mainly hunting for AI. They're checking whether the skills hold up, because too often they haven't.

Why do recruiters stop taking resumes at face value?

Recruiters stopped taking resumes at face value because self-reported skills kept coming out weaker than claimed, and AI writing tools made every resume sound equally sure of itself. When every candidate sounds strong, sounding strong tells a recruiter nothing.

The problem isn't new, and the tools haven't solved it. Verifying skills ranked first in TestGorilla's Secret Source Report 2025 too, at 59%. After two years of AI tools, the top complaint has barely moved. TestGorilla CEO Wouter Durville sums it up: "Recruiters got a better search engine and a worse index in the same 12 months."

Search tools help recruiters find more profiles faster. But the profiles they're searching are full of polished, generic text, so each match is worth less. Durville again: "A CV used to be a candidate's account of themselves. Increasingly, it's an AI's account of what a recruiter wants to read."

Other surveys found the same thing:

• Robert Half (2,000+ US hiring managers, March 2026): 67% of US HR leaders say reviewing AI-generated applications has slowed hiring, and 20% report delays of more than two weeks.
• Resume Genius: 76% say AI-written resumes make it harder to understand what a candidate actually did, and 60% want to test, discuss, or see proof of AI skills rather than take resume claims at face value.

Why won't sounding human make your resume stand out from AI-written resumes?

Sounding human won't help because recruiters can't reliably tell AI writing from human writing. The filter that actually works is generic versus specific, and only specifics get through.

Tilburg University researchers found recruiters did no better than chance at spotting AI-written cover letters. Meanwhile, 80% of managers in a separate survey say they can tell. That gap matters. A recruiter who thinks they can spot AI, but can't, ends up treating all fluent writing as suspect. Your hand-written resume gets the same discount as a chatbot's.

So if recruiters think your resume was written by AI, a warmer tone won't change their mind. What does change it is a detail a language model couldn't have made up: the name of the internal system you rebuilt, the quarter it went live, the four people on the team, the manager who'd confirm it. A named system plus a date or a number can't be generic.

What counts as a receipt on a resume bullet?

A receipt is any detail attached to a claim that a recruiter can check on their own, without having to trust you. The test: could a stranger confirm this bullet in under five minutes using only what's on the page?

Five kinds of receipt work:

1. A named system. "Rebuilt the event pipeline behind the checkout service" beats "improved data infrastructure." A name can be checked in an interview. A category can't.
2. A public artifact. A GitHub repo, a published case study, a design file shared by link, a conference talk, a live dashboard, an app store listing.
3. A dated launch. A month and year of shipping lets a recruiter find the changelog, press release, or launch post.
4. A team size and your role in it. "Led 3 of a 7-person team" shows your scope and is easy to confirm with a reference.
5. A named reference who can confirm it. Not "references on request." A line like "reference: former eng manager, available" tied to the work you're claiming.

Every bullet doesn't need all five. Every bullet needs at least one.

This lines up with what recruiters say they want. TestGorilla found that 92% would run an objective check on a sourced candidate before interviewing them, with a short job simulation and a 10-minute skills test tied as the preferred method, at 51% each. A bullet that links to working output is that check, already done: here's the work, look for yourself.

Why isn't a certificate or skills badge a receipt?

A badge asks the recruiter to trust a third party instead of you, and that trust backfires. TestGorilla's data shows that recruiters who trust pre-tested candidate data get let down more often, not less.

In the survey, 63% of recruiters believe most or all of their candidate data has been independently tested. Within that group, 65% say claimed skills still turned out weaker than advertised. Among recruiters who trust their data less, it was 36%.

A certificate raises expectations without giving the recruiter anything to look at. When the candidate then underperforms, the disappointment is bigger because the badge promised more. A repo, a dated launch, or a named reference works the other way. The recruiter checks it themselves, so the expectation is set by what they actually saw.

Keep your certifications if the posting asks for them. Just don't treat them as proof of the bullet next to them.

How do you add proof to resume bullet points, step by step?

Go through your resume one bullet at a time, find the claim in each, and attach the most checkable detail you honestly have. Cut any bullet that can't hold a receipt.

1. Find the claim. Rewrite each bullet as one sentence: "I can do X, and here's what happened when I did." If you can't find the X, the bullet is filler.
2. Name the thing. Swap categories for proper nouns: the system, the tool, the product, the client segment, the market.
3. Add a date. Month and year of launch or delivery. Dates make claims searchable and match your timeline.
4. Add scope. Team size, budget owned, number of users or accounts, using only figures you could defend in an interview.
5. Link the artifact. Where something public exists, point to the single artifact behind that bullet, not a homepage, and write a short visible URL. Check every link before you send: a dead link is worse than none, because now the recruiter has caught you in a small lie.
6. Assign a reference. For your three strongest claims, know who would confirm each one and check they're willing.
7. Order by the posting. Move the bullets whose receipts match the job's must-haves to the top of each role.

That last step is where most people stop, because doing it for every application is tedious. It's also the work Refolk takes off your hands. It writes your resume from your own history, then rewrites it for each posting you paste in, so the bullets with the matching receipts lead.

Here's what the change looks like, with illustrative bullets:

Every "after" row gives the recruiter something to open, search, or ask about. None of them depends on tone.

Where do resume receipts matter most?

Receipts matter most in the UK and for AI roles. UK recruiters are both scarcer and stricter, and AI-role profiles have become hard to tell apart. In both places, a resume you can check in five minutes stands out more.

In Refolk's index of professional profiles, there are 122,778 current US profiles titled Recruiter, Technical Recruiter, or Talent Acquisition Partner, against 9,373 in the UK. That's about 13.1 times as many recruiters in the US. Yet the smaller UK bench is less willing to take your word and more likely to spend an hour checking.

The third row applies survey percentages to index counts, so read it as a rough sense of scale, not a measurement. Still, the point holds: roughly 1,400 of the 9,373 UK recruiter profiles belong to people who'd take a CV at its word. Every hour of checking eats into a thinner bench, so a CV that saves that hour counts for more.

AI roles are the other hot spot. Refolk's index holds 16,051 current US profiles titled AI Engineer, ML Engineer, or AI Product Manager, which are exactly the roles where a third of UK recruiters say every profile reads the same. If you work in AI, "built LLM-powered features" is the most generic line you could write. Name the model, the eval you ran, the latency or cost you hit, and the date it shipped.

Why do receipts beat gaming the screen?

Receipts are built to be inspected, while tricks like hidden text are built to avoid inspection, and recruiters now go looking for the second kind. When the reader is already suspicious, anything that can't be checked counts against you.

Greenhouse data shows 22% of recruiters have caught hidden prompt injections in resumes, and 39% are running more in-person interviews to check authenticity. Every hidden keyword and invisible instruction adds to the suspicion that already costs everyone time. Petra Wilton of the Chartered Management Institute warned that "playing detective over a single candidate's claims slows recruitment down."

That slowdown is what job seekers experience as silence. Robert Half's numbers show it: 20% of HR leaders report AI-application review delaying hires by more than two weeks. Your application isn't rejected. It's sitting in a pile waiting for someone to find an hour, while you send more into the same pile.

FAQ

How do recruiters verify resume claims in 2026?

Mostly by hand. TestGorilla found 67% of US recruiters and 73% of UK recruiters spend an hour or more checking a single candidate, and 92% would run an objective check such as a job simulation or 10-minute skills test before interviewing. They search for named systems and launches, open linked work, call references, and increasingly run in-person interviews. The more of that your resume does for them, the faster you move.

Do I need a GitHub or portfolio for every bullet?

No. A public artifact is one kind of receipt, but a named system, a dated launch, a team size, or a named reference also count. Aim for at least one checkable detail per bullet and a link only where real public work exists. For NDA work, a short sanitized write-up or a reference who can confirm it is enough.

Will an AI detector flag my resume if I used AI to draft it?

Detection is unreliable, so it's the wrong thing to worry about. Tilburg University researchers found recruiters spotted AI-written cover letters at chance level. What hurts you is being generic, not using a tool. If every bullet carries a specific name, date, or reference you can defend in an interview, how you drafted it matters far less than whether it holds up.