Lean delivery in a physical world
Releasing value early, amplifying learning and reducing risk in the supply chain space
Gousto offers 40 delicious meals to customers across the UK each week — our ambition of becoming the nation’s most loved way to eat dinner means that it’s easy to see us just as a food company — however, speak to anybody at Gousto and they’ll tell you we’re a tech company. Tech makes up nearly 50% of our London office. I personally see Gousto as a supply chain company. Effectively, our core business is providing food to our customers in a way that suits them, with no waste. In order to achieve that, we must build a top class supply chain.
At the same time, Gousto is also a Lean business. We value experimentation massively, we strive to release value early, learn and iterate as much as we can. In the Tech industry there are a lot of standards and examples in the e-commerce space that we can rely on, e.g. A/B, multi-variate and painted door testing. However, in the context of a physical supply chain process, we hear of fewer of these examples, as testing a new capability in a factory or a new concept with a small reference set of customers can be quite difficult. This is due to constraints that are intrinsic to the physical world. For example, if we want to deliver some value for our customers that require a change in our supply chain, this may involve buying new physical machines or lengthy contractual changes with our suppliers. As you can imagine, with such constraints it can be very difficult to release value early, learn continuously and iterate rapidly, so we need to think differently.
Here I’m going to tell you a story that explains how we managed to think Lean in a physical world (in fact, Lean is coming from Toyota manufacturing, but here I’m talking about Lean in software engineering dealing with physical rather than digital).
Gousto already offers a very competitive delivery choice within the recipe box market, by offering a market-leading, three day lead time to our customers — in a nutshell, if you order your box today, we’ll deliver it to your door in three days. However, we like to think big, and if we want to compete with supermarkets in the future, we need to do more. So a few months back, one of the teams were presented with the challenge of reducing our lead time to one day in order to be able to offer next day delivery to our customers.
We quickly realised how this would not only massively impact our operations and logistics at the factory, but also require huge engineering effort to adapt our platform. From the initial estimates, we were looking at several months worth of work, which meant it was going to be impossible to offer next day delivery to our customers by the end of 2019 as we had hoped.
“Simply changing the order in which we delivered the different pieces of work enabled us to release huge value to our customers earlier, while testing our internal processes and satisfying our stakeholders sooner than we ever imagined”
As an engineering team that is passionate about Lean, we challenged ourselves to come up with some way of achieving our goal earlier. We broke down the scope of the project and we looked at all the tasks that we had to complete to make our platform ready to support next day delivery. What we achieved by doing that was impressive: simply changing the order in which we delivered the different pieces of work enabled us to release huge value to our customers earlier, while testing our internal processes and satisfying our stakeholders sooner than we ever imagined.
Specifically, instead of aiming for next day delivery being available on our website and apps for our customers as a first milestone, we decided that we would make it available for our internal customer care agents. Every day they have to help customers whose boxes have failed to be delivered, luckily not in huge numbers. Normally they can offer a replacement box; however that won’t reach them for another three days. This can be a frustrating experience for our customers. With our MVP (minimum viable product) we added next day delivery as a new capability to our customer agents portal.
This unlocked tremendous value as we can now effectively send a replacement box within one day, improving the customer experience.
Even more importantly, this solution allowed us to be in control of the number of boxes that we would deliver on the following day. At the same time, it gave an immense opportunity to our operations and logistics teams to test end to end the impact of such a big change and learn from it, reducing the risk of releasing the feature to customers. After a few weeks of tests from the customer agents portal, we felt confident enough to run an A/B test and enable a small percentage of our customers to order next day delivery boxes from our website only on specific days. This is helping us to learn about the impact of the new capability on conversion and retention. We created operational success metrics, easily measured every day, so that, at the end of every week, we can decide whether to scale to either more days or more boxes per day. When we don’t meet the criteria, we learn something and we try again the following week.
This initiative demonstrated how cross-functional collaboration between multiple teams, like operations, logistics, projects, customer care, digital product and software engineering can lead to great results.
A note on tech debt
From an engineering perspective, we accepted some compromise. In particular, we had to touch areas of our codebase that were written when Gousto was smaller and had a different set of challenges. We saw this project as a good opportunity to improve the stability and maintainability of our platform as well as make it more scalable. However, we accepted to build the new capability without reducing the tech debt upfront, so that we could release value to our customers earlier. At the same time, we agreed to ring-fence some time after go-live to pay off that debt. In my mind, this is exactly what tech debt is about. Accruing a debt to get earlier value, exactly like when you buy a house and get a mortgage; the mortgage is debt and the house is value.
I want to reassure all the engineers who are reading this article, this is actually happening, the team are currently working on reducing that tech debt :)
By empowering our cross-functional team to work towards outcomes, engineers were able to really cut-down the time to value for the first iteration. In doing so they got not just early feedback but a real feeling of making impact and ownership of the whole piece.
By rethinking our approach and thinking Lean, we managed to push the go-live date forward by more than three months and we’ve already tested the new capability reducing the risk of the release.
The main achievements are:
- We’re increasing delivery convenience for our customers
- We learned massively about the impact of such a big change and we’re building the confidence to go live with more customers, more boxes and more days
- We got more trust from our stakeholders from other parts of the business (read my very own blog post on the topic if you’re interested!)
Do you have any experience of applying the Lean principles in the supply chain world? Please share your insights in the comments below.
If you wish to know more about what we’re up to at Gousto, please check out our tech blog!