Sources and research

Why this page exists

We believe companies that make statistical claims in their marketing should be able to show where the numbers come from. This page lists every claim Vikkla makes in copy, pitches, articles, or ads — with primary source, measurement method, publication date, and our assessment of how robust the underlying data is.

Our rule: a claim that can't be verified against a primary source is not used. A primary source older than 18 months is flagged for re-verification. Older than 24 months — removed until newer data exists.

Active sources

"62% of HR professionals say AI-generated CVs without personalization often lead to rejection"

• Primary source: Resume Now. AI and the Applicant Report, March 2025.
• Method: Survey of n=925 US HR professionals.
• Published: 2025-03-28.
• Framing: The number refers to generic AI CVs — not AI-assisted CVs in general. Used in Vikkla's copy to support per-listing tailoring, not as an argument against AI.
• Status: Verified.

"77% of job seekers worry their CV gets filtered out before reaching a human"

• Primary source: Monster. 2026 State of Resumes Report, December 2025, distributed via PR Newswire.
• Method: Pollfish survey, n=1001 US job seekers.
• Published: 2025-12-16.
• Framing: Perception data — measures the worry, not actual filtering. Used to explain why ATS optimization feels urgent.
• Status: Verified.

"78% of employers say they actively check for AI-generated content"

• Primary source: Resume Now. AI and the Applicant Report, 2025.
• Method: Employer survey, self-reported.
• Framing: This measures what employers say they do — not what they can technically detect. There is no reliable AI text detection for CVs as of 2026. Always used with explicit framing.
• Status: Qualified (perception ≠ technical measurement).

"An active job search typically takes 2–6 months for Swedish professionals"

• Primary source: Missing. Based on industry experience and aggregated observations from recruitment consultants, the Swedish Public Employment Service (Arbetsförmedlingen), and our own customer interviews.
• Method: No unified measurement — different studies use different definitions of "active search" and "hired".
• Framing: Used with the qualifier "typically" or "usually" — not as an exact figure. A range, not an average. Replaces the previous "24 jobs" formulation (LinkedIn Workforce Report) which could not be source-verified.
• Status: Qualified (industry experience, not primary source).

Claims we do NOT use

"62% of recruiters can detect AI-generated content"

This number circulates in several blog articles but we have not found a primary source supporting the wording. We don't use it in our material and recommend other actors in the field run the same check.

Detection rates for AI text in CVs

We have searched for reliable 2024–2026 data and found nothing. Until such a study exists, we don't use numbers around "detectability" — neither for nor against AI CVs.

Methodology

• Before a number is used in external material: primary source is verified, date and method are recorded.
• Every row is reviewed when the publication date is older than 18 months. Older than 24 months — removed.
• New verification is run when an external party cites a number for or against us.

If you find we're using a number without support: email johan@vikkla.com. We correct it within 48 hours.

Last update

2026-05-14 — English version published.

2026-04-17 — Added the "2–6 months" range for job search duration. Replaces the earlier "24 jobs" figure that could not be source-verified.

2026-04-15 — First public version. Extracted from internal claim ledger.