Nearly two decades ago, just as e-commerce was taking off, a group of players emerged to claim their share of the home-delivery market. Remember Webvan, Urbanfetch, Kozmo, and HomeGrocer? In 2000, in this magazine, we analyzed these and several other startups and found they faced insurmountable hurdles. Limited online sales, high delivery costs, entrenched competitors, and an unacceptable trade-off between speed and variety would combine to doom many of the early home-delivery companies (we called it “the last mile to nowhere”). And in fact, most flamed out in spectacular fashion.

Many of the same challenges persist today, often with added complexities. By 2014, Internet sales in the U.S. had reached US$300 billion, an impressive growth rate averaging 18 percent for 15 years. Yet e-commerce accounts for just 7 percent of total U.S. retail sales — the physical store is still alive and well. Delivery costs continue to be driven by variable labor costs, delivery density, and average order size. The established competitors (UPS, FedEx, and the U.S. Postal Service) have become increasingly dependent on e-commerce to replace the business lost from the digitization of letters and other documents. And their position has been further complicated by companies that use crowdsourced delivery models.

But the most important change since our earlier analysis has been the evolution of the trade-off between speed and variety. In the late 1990s and early 2000s, home-delivery startups focused on speed at the expense of variety: They could get you a small selection of goods relatively quickly. Today, when retailers approach the last mile, they make more nuanced trade-offs among speed, variety, and convenience. The right combination entails a complex set of compromises that depend on the product type, consumer segment, shopping occasion, and retailer positioning.

That said, the fundamental economics of the last mile haven’t changed. Companies have to offer a solution with costs equal to or lower than the customer’s willingness to pay (the “cost to serve”). It’s easy to delight customers with a free offering, and it’s not hard to cover your cost by charging a high premium. But finding the sweet spot that resonates with consumers and drives sales growth proves far more difficult. If retailers can get that right — admittedly, a big challenge thus far — they can make the last mile a competitive advantage.

To help companies better understand these complexities, we conducted a bottom-up analysis of the cost-to-serve for an array of retail models, including traditional store-based sales, curbside pickup, crowdsourced shoppers, “white glove” delivery, and pure-play e-commerce. We also surveyed 2,000 online U.S. shoppers to determine their willingness to pay for each of those last-mile options for a variety of goods purchased online. The results revealed some of the winning approaches in categories such as groceries, durable goods, and apparel.