Sebastian Steinker

Who they are

This page records the public professional record only. Nothing adverse is recorded and no motive is attributed.

His affiliation and wider career were not verified for this page, so none is asserted here. The record used is one paper, written with two co-authors, on the value of weather information for e-commerce operations. It is the main evidence on this site for how much a weather forecast adds to an online retailer's sales forecast.

Key papers

Steinker, S., Hoberg, K., & Thonemann, U. W. (2017), 'The Value of Weather Information for E-Commerce Operations', Production and Operations Management, 26(10), 1854-1875, doi:10.1111/poms.12721. Read at the level of the published abstract and a university summary.

His wider publication record was not surveyed for this page.

Key ideas

The authors worked with about two and a half years of daily online demand at a European online fashion retailer. Sunshine, temperature and rain significantly affected daily sales, especially in summer, at weekends and in extreme weather. The effect was asymmetric: good weather cut demand strongly, and bad weather raised it by much less.

Adding weather data to statistical forecasting models improved forecasts against a baseline time-series model that ignored the weather. The abstract reports that errors fell by 8.6 to 12.2 per cent on average and by up to 50.6 per cent on summer weekends. The paper also tested the value of weather information for workforce planning at the fulfilment centre, which this page does not report in detail.

The sign of the effect is worth keeping apart from the size. In Brigitte Roth Tran's 2022 data, the online sales of the brand she studied rose on good-weather days. The retailers, countries and years differ, and neither study can say which difference matters. See weather-sensitive demand and What a Weather Forecast Does for Sales.

The limits are those of a single retailer's data, read at the level of the abstract. The 50.6 per cent figure is the best case, on summer weekends, and should not be read as the typical improvement.

How to read the result

The study is about the online channel, and its findings run against the common expectation. In the retailer's data good weather cut orders strongly and bad weather raised them by much less, with larger effects in summer and at weekends. A planner who assumes the online channel behaves like a mirror image of the shops, or the same as them, is assuming something this paper does not support.

The improvement figures need careful reading. Adding weather data to statistical forecasting models cut forecast errors by 8.6 to 12.2 per cent on average, compared with a baseline time-series model that ignored the weather. The 50.6 per cent figure is the largest reduction, found on summer weekends, so it marks where the weather mattered most and should not be quoted as the typical gain. The comparison point also matters: the baseline ignored weather altogether, so the figures say what weather adds to a weather-blind model, not what it adds to a sophisticated one.

The paper went on to test the value of weather information for planning the workforce at an order fulfilment centre. In a scenario where the weather was known perfectly, excess costs fell by 11.6 per cent against a baseline that ignored it. Because nobody knows next week's weather perfectly, that figure is an upper bound on what the information could be worth, and What a Weather Forecast Does for Sales prints it as such.

What the record does not show

The data are one European online fashion retailer's orders over about two and a half years, read here at the level of an abstract and a university summary. The paper cannot say whether a retailer of different goods in a different country would see the same sign. Brigitte Roth Tran's 2022 letter found the online sales of the brand she studied rose on good-weather days, and neither study can say which difference between the two settings explains it. Co-authors Kai Hoberg and Ulrich Thonemann share the credit for the results.

---