Case story
Your conversion rate didn't crash. Your traffic got polluted.
An EU kids-interior brand running five locale stores (anonymized).
The scare
Mid-August, the brand's US store conversion rate read 0.33%, down hard, overnight, with no site change, no pricing change, no checkout issue. The normal playbook kicks in: panic-audit the site, blame the ads, cut spend.
What Vibel found
Traffic Quality flagged the wave within the data itself. Starting 12 August, a bot catalogue crawl was landing thousands of sessions on product pages, sessions that never touched a cart, spread with machine-like uniformity across the catalogue. Direct traffic alone showed 2,842 sessions converting at 0.11%.
Vibel's detector scores four independent signals (volume vs trailing median, cart-rate collapse, source-vs-CVR divergence, landing-page uniformity) and classified the wave as non-human. It then showed both numbers side by side: the raw 0.33% and the adjusted ~0.87-1.9% once junk sessions were excluded. Real paid traffic was converting fine the whole time.
The outcome
- No spend was cut on healthy campaigns chasing a phantom CVR problem.
- The same sweep surfaced where the money actually leaked: a winning placement at 2.67 ROAS was starved at 6% of budget while a 1.08 ROAS placement ate 32%, plus ~$1,600 sitting in two dead campaigns.
- The report told the merchant explicitly: this is not a site problem, the one sentence that prevents a week of wrong fixes.
Why this matters
Every analytics tool shows you the blended number. Vibel tells you when the number itself is lying, and never silently rewrites it. You always see raw and adjusted together, with the evidence.
Every number, checked against the source.
Vibel reconciles spend and traffic against the platforms themselves, and tells you when they disagree.