Sharad Borle

Who they are

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

The affiliation is the one printed on the 2005 paper. His co-authors were Peter Boatwright, Joseph Kadane, Joseph Nunes and Galit Shmueli, and two of them, Boatwright and Nunes, were the authors of the 2001 paper the study revisits. His wider career was not surveyed for this page.

Key papers

Borle, S., Boatwright, P., Kadane, J. B., Nunes, J. C., & Shmueli, G. (2005), 'The Effect of Product Assortment Changes on Customer Retention', Marketing Science, 24(4), 616-622, doi:10.1287/mksc.1050.0121. The abstract and the full text were read for this page.

His wider publication record was not surveyed.

Key ideas

The study built a model that estimates the effect of an assortment change on sales at the store level and the category level at the same time, and applied it to detailed household purchase records from an online grocer. The cuts ran from 24 to 91 per cent of the items in a category and came in the last six months of a year of data. The 840 households offered the reduced range were compared with 378 that kept the full range.

The results were that households were expected to wait about 25 per cent longer between orders and to spend 4.8 per cent less per order, that the effect on frequency produced the larger loss, and that the impact varied widely by category. Less frequently bought categories, such as aluminium foil and refrigerated pasta, were hit harder. Staples such as milk, bread and cereal were hit less. The cut affected how often households bought in about 74 per cent of categories.

The paper is explicit about why it differs from the 2001 analysis. The earlier work used a subset of categories and households, tending towards frequently bought staples, and fitted to the earlier households with all categories the new model returned the new result. The practical advice the authors draw is for managers to compare selected categories in order to moderate the overall loss in sales.

It states its own limits. The principal experiment was in a single city where delivery fees did not vary, and it reflects changes in usage by existing customers and not effects on customer acquisition. See assortment reduction.

What the 2005 paper adds for a shop

The paper is the one that asks what a cut costs after the first week. Its measure is how often households order, and its finding is that frequency fell more than basket size. For a shop owner the implication is about what to measure. Basket size is the figure most tills report and the one that moved least. Visits per household moved most, and it takes weeks of records to see.

The paper also names where the loss falls. Categories bought less often, such as aluminium foil and refrigerated pasta, were hit harder, and staples such as milk, bread and cereal less. The authors' advice, to compare selected categories in order to moderate the overall loss in sales, follows from that spread. It treats a cut as something to test in part of a range first.

Its reading of the earlier result is stated as a reason and not a verdict. The 2001 analysis used a subset of categories and households that leaned towards frequently bought staples, and the new model, fitted to the 2001 households with all categories, returned the new result. A reader can accept that account and still note that the correction note to the 2001 paper was not read for this page.

The limits are the authors' own. One city, one delivery fee, existing customers only. The paper cannot say whether a thinner range would have drawn fewer new customers.

---