The research, published in Discover Cities by Archiman Biswas of the World Resources Institute and the Indian Institute of Technology Kharagpur, together with Purbita Samanta of IIT Kharagpur, combined geographic information system measurements with survey data from 484 consumers in Jamshedpur, a planned Tier-II industrial city in eastern India. The authors geotagged each respondent’s residence, mapped the city’s two dominant retail agglomerations — Bistupur and Sakchi — and computed the shortest motorised road-network route from every home to the retail centroids using ArcGIS and an OpenStreetMap-derived network validated against municipal records. Crucially, they rejected straight-line Euclidean distance, which systematically understates the actual travel burden in planned urban grids with restricted crossings, in favour of a measure that reflects how people actually move through the city.

The study’s methodological centrepiece is a data-driven threshold calibration. Rather than imposing an arbitrary one-kilometre or two-kilometre cutoff — the convention in prior work — the authors applied a Chow-test segmented regression, scanning 384 candidate thresholds across the 10th to 90th percentile of the observed distance distribution. The procedure identified a structural break at 1,844 metres, where the combined residual sum of squares of two sub-sample regression models reached its minimum. Substantively, this distance corresponds to roughly 20 to 22 minutes of walking, the outer boundary of the pedestrian catchment around the retail nodes. It is precisely where, the authors argue, the cognitive and physical cost of a spontaneous store visit transitions from negligible to non-trivial.

The behavioural discontinuity at that boundary is striking. Consumers living within 1,844 metres of the retail agglomerations — 105 of the 484 respondents — reported webrooming in an average of 6.54 of their last ten fashion purchases, compared with 4.15 among those living farther away. Showrooming showed the mirror image: distant consumers averaged 2.62 showrooming episodes per ten purchases, against just 0.64 among proximate consumers. Both differences produced effect sizes exceeding 1.75 in Cohen’s d terms, far beyond the conventional threshold for a large effect. Age, income, and digital literacy did not differ significantly across the two groups, supporting the interpretation that the behavioural gap reflects spatial rather than socioeconomic sorting.

Continuous regression analysis reinforced the pattern. In ordinary least squares models with heteroskedasticity-consistent HC3 standard errors — chosen because the bounded 0-to-10 outcome variables induce heteroskedastic residuals that inflate Type I error rates under homoskedastic assumptions — log-transformed road-network distance emerged as the dominant predictor in both models. Each unit increase in log distance reduced webrooming frequency by 1.749 units and increased showrooming frequency by 1.438 units, both highly significant. The webrooming model achieved an R-squared of 0.624 and the showrooming model 0.604, substantially higher than the 0.12 to 0.19 range reported in earlier spatial e-commerce studies that treated online and offline shopping as binary alternatives rather than directional sequences. Poisson and negative binomial robustness checks confirmed that the direction and significance of the distance effect persist across distributional assumptions.

The study’s most theoretically consequential finding concerns the cognitive pathway through which distance operates. Using bootstrapped mediation analysis with 5,000 resamples, the authors tested whether perceived accessibility — the subjective evaluation of how easy it is to reach and use retail outlets, measured with three Likert-scale items — mediates the distance-behaviour relationships. For webrooming, it does: the bootstrapped indirect effect was −0.310 with a 95% bias-corrected confidence interval of [−0.453, −0.167], excluding zero and accounting for roughly 18% of the total distance effect. For showrooming, the confidence interval [−0.058, 0.181] includes zero, indicating no significant mediation. The asymmetry suggests that spatial friction reaches the two behaviours through different cognitive routes.

The authors interpret this asymmetry through the Stimulus-Organism-Response framework and activity-space theory. Webrooming is a deliberate, pre-planned sequence: the consumer commits in advance to a store visit as the purchase endpoint, making anticipated accessibility cognitively salient at the moment of channel planning. Showrooming, by contrast, typically unfolds reactively during or after a store visit, with the online purchase decision emerging from in-store product comparison rather than prior planning. When a retail centre falls within a consumer’s habitual activity space, the in-store transaction can be appended to routine movement at low marginal cost, supporting webrooming. When it lies beyond that boundary, the visit becomes a discrete spatial effort justified only by the non-substitutable benefit of tactile evaluation — and the transaction migrates online to avoid a return trip, generating showrooming.

The findings carry direct implications for retail location strategy and urban planning in India’s rapidly urbanising secondary cities, which house an estimated 35% of the country’s urban population and contribute around 65% of its incremental online shoppers. The 1,844-metre boundary around a retail agglomeration represents not merely a trade-area edge but a behavioural inflection point: beyond it, the physical store shifts from transaction venue to evaluation venue, potentially reducing in-store conversion while raising the importance of experience quality. The authors suggest that retailers serving distant consumers might prioritise in-store product presentation and comparison support, while India’s Open Network for Digital Commerce initiative could consider last-mile transport and mixed-use zoning interventions that reduce effective friction beyond the threshold. The positive association between income and both channel sequences also introduces a spatial-equity dimension, suggesting that the benefits of combining digital and physical retail are disproportionately available to economically advantaged consumers.

The study’s limitations define its agenda. The cross-sectional design precludes causal inference, the single planned city may not generalise to organically grown morphologies, and the product scope is restricted to fashion and apparel, where tactile evaluation plays a distinctive role. Self-reported channel frequencies introduce recall bias, and car ownership, local retail density, and residential self-selection remain potential confounders. Still, the Chow-test calibration procedure offers a replicable, data-driven tool for future multi-city comparisons across the Global South. What the study establishes is that even as digital retail expands at over 25% annually in India, the spatial configuration of physical stores remains a structural force in consumer channel choice — placing distance, quite literally, back at the centre of omnichannel behaviour research.

Subject of Research: A GIS-based study of 484 consumers in Jamshedpur reveals that road-network distance, not attitude, is the dominant predictor of cross-channel shopping behaviour, with opposite effects on webrooming and showrooming.

Article Title: Analysing influence of retail accessibility and spatial organisation of in store retail on cross channel shopping behaviour in tier II Indian cities

Article References: Biswas, A., & Samanta, P. (2026). Analysing influence of retail accessibility and spatial organisation of in store retail on cross channel shopping behaviour in tier II Indian cities. Discover Cities, 3(1), Article 202. https://doi.org/10.1007/s44327-026-00381-2

Image Credits: AI Generated

DOI: 10.1007/s44327-026-00381-2

Keywords: retail geography, webrooming, showrooming, spatial friction, perceived accessibility, Tier-II cities, India, consumer behaviour, Analysing, influence, retail, accessibility