Unlocking Efficiency: The Power of Our Delivery Offering

Siri Bruskeland
Oda Product & Tech
Published in
7 min readAug 15, 2023

At Oda, our mission is to give you more space for life. We do this by letting you order your groceries online while we take care of the picking and home delivery, so that you can spend time on more important things.

We are here to stay, and in order to ensure a viable business model and a profitable logistics operation, there are especially two things that are important:

  • That we pick efficiently 📦
  • That we deliver efficiently 🚚

In this article, we will focus on delivery efficiency, which plays a significant role in our overall costs.

To measure our delivery efficiency, we use various metrics, with the most important one being SPORH: How many stops (customers) we deliver to per on-road hour. This metric reflects the productivity of our routes. We want to create as much value as possible when we’re out delivering. We consider the actual delivery of groceries to our customers as valuable, not the driving in between. Hence we want to have as many stops per on-road hour as possible.

But having a good metric is not enough. We still need to make sure that we actually deliver efficiently, and hence have a high SPORH. So how do we do that?

There are a lot of factors that impact our delivery efficiency. Route planning, driver assignment, service time prediction and the delivery tools our drivers use, are a few of them.

Another important factor is our delivery offering.

By delivery offering we mean where we deliver at which times.

Finding the optimal delivery offering involves a tradeoff between a customer offering that is as flexible as possible, and our delivery efficiency:

  • If our delivery offering is too flexible (i.e. we offer a lot of delivery slots throughout the whole day and week in both central and rural areas), we will be able to sell a lot, but it will be too costly to deliver
  • If our delivery offering is too rigid (i.e. we offer a very limited set of delivery slots) we will be able to deliver with super high efficiency, but we will not get enough sales

Hence we need to find the sweet spot where we sell as much as we can deliver while keeping a certain level of efficiency, and hence maximize our profit.

Illustration of flexible versus rigid delivery offering by Jørgen Hoff Amundsen

In order to find the sweet spot there are especially two things we need to consider:

  • Geography — Where do we deliver 🗺️
  • Time — When do we deliver ⏱️

We want to deliver in areas that has high enough population density to ensure a certain amount of customers without too much driving distance between the customers.

We also want short driving distance to and from our:

  • Fulfillment Centers (FCs) – where we pick and stack our orders
  • Distribution Points (DiPs) – the place our last mile vehicles starts from (last mile vehicles can also start directly from a FC)
Rough illustration of Oda’s delivery area in Norway by Jørgen Hoff Amundsen. Our FC (i.e. “Fulfillment Center”) is located in Lier, and we have several DiPs (i.e Distribution Points)

Once we have decided where to deliver we need to decide when to deliver in different areas.

We cannot offer the same set of delivery slots in our whole delivery area. Our customers that live in central areas, closer to our FC or DiPs, are offered more delivery slots than our customers living in rural areas with longer distance from our FC and DiPs. It would be challenging for us to reach the outskirts of our delivery area in time for our earliest delivery slots. In the same way we cannot offer our latest delivery slots in these areas because the drivers would not be able to get back to the FC or DiP before their shift ends.

As already mentioned, our delivery offering is also heavily dependent on the population density of the areas we deliver to, also when it comes to the delivery slots we offer. In less dense areas we get fewer orders. If these orders are spread out over the entire day we would need more drivers and vehicles to deliver than if they were all scheduled around the same time. This would cost us a lot, without giving us necessary revenue to cover the cost.

On the other hand, the more we grow (which we consistently have done in Norway over the last years), the higher order density we get for a given population density. This makes it possible for us to provide a more flexible delivery offering in the future, also in rural areas.

In general we try to nudge customers to choose delivery slots that make it possible for us to deliver efficiently. We do this by using:

  • Direct control mechanisms — only offering a specific set of delivery slots, and/or restrict the amount of orders we can deliver within each time slot
  • Indirect control mechanisms — by pricing the slots in a way that make the customers choose the slots that are beneficial for us

We normally use a combination of direct and indirect control mechanisms.

In more central areas, where population density is higher and customers are closer to our FC (i.e. “Fulfillment Center”), we offer a vast amount of delivery slots, while in more rural areas our delivery offering is more limited. Illustration by Jørgen Hoff Amundsen

We also use these control mechanisms to spread our demand more evenly throughout the week. Monday is by far our most popular delivery day in Norway, and it would be even more popular if we did not nudge the customers to choose other days. If we staffed our operation to handle a day that has much more volume compared to other days, we typically would have people and vehicles that would be unused rest of the week. This is not sustainable for either business or the environment.

The order volume we have on Monday’s are much higher compared to the rest of the week, even with control mechanisms in place. Without the control mechianisms the demand would be even more skewed

Now you know why it matters where we deliver and when we deliver.

Another thing that is important for our delivery efficiency is the duration of the slots we offer to our customers (e.g. two-hours-, three-hours-, four-hours delivery windows etc).

The wider slots (delivery windows) our customer chooses, the more flexibility we get when do our route planning. This is good from a business perspective, but also from a sustainability perspective since we drive less kilometres and need less vehicles.

I’ll try to illustrate this with an example. Let’s say that we (for simplicity) have four customers to be served by one route from one of our FCs. If all of the customers chooses slots with a duration of five hours, we can easily travel from customer A to B to C to D and get a quite efficient route.

If all of the customers chooses two hours slots instead, we might have to deliver to customer D after customer A in order to deliver on time to these customers, before we deliver to customer B and then C. This results in more time spent on the road and more kilometres driven.

We can look at the duration of our slots as Tetris pieces. Orders that needs to be delivered within a five-hours-slot gives us a lot of flexibility when we do route planning, just as a 1:1 Tetris piece is easy to allocate in the Tetris board. On the contrary, orders that needs to be delivered within one-hour-slots reduces the solution space for our route planning substantially, which makes it hard to ensure high delivery efficiency. It’s in fact so hard that we do not offer any one-hour-slots as of today. However, we might want to offer this in the future in areas with very high population density.

1-hour-slots lead to much more complexity when we do our route planning compared to wider slots. Illustration and idea by Matias Hermanrud Fjeld

It’s natural to believe that our customers would want as narrow delivery slots as possible to chose from throughout the whole day, regardless of where they live. But if we want to have a high delivery efficiency, this is not possible. Hence, our goal is to strike the right balance, offering delivery slots that meet our customers’ needs while maintaining high delivery efficiency.

We hope you now have learned a bit more about our approach to creating a good delivery offering.

We always appreciate questions and feedback, and if you’re interested in learning more about Oda and our delivery operations, check out How we went from zero insight to predict service time with machine learning model part 1 and part 2.

--

--