Google Maps buries new London restaurants before they can survive. Researcher Lauren Leek reverse-engineered the platform's three-factor ranking system, relevance, proximity, and prominence, and found that prominence is the decisive variable. It measures interaction frequency, brand recognition, and review volume, not quality. New independents get no traffic because they have no reviews. No reviews means no traffic. The loop is closed before they open.
Leek built a machine learning model to escape that loop and published a public dashboard at laurenleek.eu/food-map that surfaces restaurants outperforming their Google Maps weight. The model scores London venues across five dimensions: density, ratings, surprise factor, cuisine diversity, and independent share. Hexagonal heatmaps cluster the city into hub tiers. The five top-scoring outer-London zones are Ealing-Acton, Kingston Upon Thames, Enfield, Bromley, and Havering, not the neighborhoods Google's algorithm pushes. Paid placement in Maps results, which Google does not disclose to users, compounds the distortion.
The most valuable section of the original piece is not the tooling. It is Leek's finding that restaurant diversity maps directly onto settlement history, affordable high streets, and displacement patterns. The algorithm does not create inequality in London's food geography, it locks it in. Read the full Substack post for the methodology behind the prominence deconstruction and the implications for any city where a single platform controls discovery.
[READ ORIGINAL →]