Polina Hneletska
“A recommendation block is the shelf of the digital store. I have stocked that shelf by hand — now I audit the algorithms that arrange it.”
Algorithmic bias in e-commerce recommendation systems
An independent research program at AVBR Lab — audited from the outside, with reproducible black-box protocols.
Shopify · BigCommerce · through 2027
Shopify as the first wave, BigCommerce as the second platform, longitudinal waves through 2027.
I lead an independent research program on algorithmic bias in e-commerce: how recommendation algorithms decide what shoppers see, what they pay, and who gets left out. At AVBR Lab we audit recommendation systems from the outside — no platform cooperation, no API access — using reproducible black-box protocols any researcher can verify. The program’s first wave examined 300 Shopify stores; a cross-platform census of BigCommerce is complete, with longitudinal waves scheduled through 2027.
My path to this research runs through fifteen years on every side of the retail shelf. Since 2011 I have executed visual merchandising algorithms by hand on a beauty-retail shop floor, field-audited shelf placement across a national FMCG market, built brand identities and product imagery as an entrepreneur, and run a commercial product photography studio in Dnipro, Ukraine.
I write about what we find — the patterns, the methods, and what they mean for merchants and shoppers alike. For research collaboration or press: polina@avbrlab.org.
My first algorithm fit on a printed sheet of paper.
Fifteen years on both sides of the retail shelf — the planogram years, the studio in Dnipro, the lab.
