Of course, that's just a crude map — it's not actual research. So what engineers Erica Wygonik and Anne Goodchild did in their paper (pdf) was to take a look at the actual stores and homes around Seattle and simulate thousands of deliveries and grocery runs around the city. They then analyzed a random sample of those runs.
They found that delivery trucks bringing groceries to people's doorsteps emitted between 20 to 75 percent less carbon dioxide per customer, on average, than having all those people drive their cars from home to the store and back again. That's bigger than previous estimates:
The biggest variable here, interestingly enough, was whether the customers or the grocery store had more leeway over delivery times. If the store could choose its drop-off times (that's the column on the right), then the delivery truck can follow a much more efficient route through the city. The carbon savings are quite large — 80 to 90 percent.
By contrast, if customers are choosing drop-off times (the column on the left), then the delivery truck spends more time on the road, and the carbon savings are smaller — between 17 and 75 percent. But even in the latter case, the authors found, the delivery truck was more efficient.
Now, there are still lots of caveats here. The study only modeled one city, Seattle. It didn't compare delivery service against, say, taking the bus or walking to the store. (Presumably the latter would be the greenest option.) It didn't look at the effects of refrigeration, since so few delivery trucks currently use this. It also didn't consider that many people may stop by the grocery store while running multiple other errands anyway, or on the way back home from work. And so on.
What's more, it's easy to come up with plenty of counterarguments. What if, for instance, online grocery shopping encourages people to make more shopping "trips" over time? Wouldn't overall emissions actually go up?
In some ways, questions like these help illustrate the logic behind proposals for a carbon tax. It's often very, very difficult for individual researchers to calculate precisely which types of behavior are greener than others. There are all sorts of variables and wild cards to consider. There's always the possibility a study might miss something.
But if fossil fuels were taxed in proportion to the environmental damage they cause, then the market would quickly sort out many of these questions. Behaviors and business models that used less fuel and emitted less carbon would also be relatively cheaper. The price signal would help reveal which activities were truly greener. That's one reason why so many economists like the idea of a price on carbon.