In-house platform technology contributes to PostNL's ambition to be the favourite delivery service. We recently won a Computable Award for a machine learning algorithm with predictive capabilities that gives recipients more control over the delivery time of their parcel. Bram Grijzen, the Platform Owner, explains why this is so important and how the platform works.
Scale is and remains essential to PostNL's ability to provide our service at the right price. We also need the technical capabilities required to continue to provide our customers – both senders and recipients – with optimum service. As Platform Owner, I am primarily focused on the development of our platform technology functionality.
Modern technology is really treacherous: it's great to be able to quickly and easily make something new, but if you're not careful you end up with a haphazard mess and you neglect the ability to innovate. You create a monster that you can't build on any further, and in the longer term you actually lose value.
Our in-house logistics platform enables us to quickly connect and disconnect new options and partners to and from our network and, ultimately, improve and expand our services.
The platform provides the orchestration between sender, deliverer and recipient. Data and intelligence determine the optimal routing and completion of a delivery. A judge module automatically decides what needs to be done with the order, how and when the delivery must be made, and which carrier will then be most appropriate. Obviously, our ultimate objective is to give customers as much control as possible over the delivery of their parcel.
The machine learning algorithm, which predicts whether the sender will supply a parcel to PostNL on the same day that the advance shipping notice was received, is an important element of the platform. The prediction enables us to estimate the expected delivery time of a parcel very early in the process, which gives the recipient more time to change the delivery time.
The recipient then benefits from greater control over the delivery time, which in turn increases our hit rate, i.e. the proportion of parcels that are delivered. This reduces the logistics costs and increases customer satisfaction – a win-win situation.
The data science department developed the model in Python, which was then converted into Java code and imported into our platform. The award-winning solution takes just half a millisecond to perform each prediction, which really is lightning fast.
The data and intelligence in our logistics platform provides for the optimal routing and delivery of consignments. Combining external data with our transactional data enables us to make parcel orders – webshop advance shipping notices – environment-specific: the system knows the destination of the parcel and the context/preferences of the recipient.
Recipients will shortly be able to specify a safe place where parcels can be left in their absence. Obviously, we cannot do this when the sender has paid for a service with signature for delivery. The system has the capability to make the appropriate considerations for this.
The maximum dimensions of a parcel, for example, are of relevance to the deliverer: some parcels are too large to delivered by bike. The event management platform generates a message when a parcel falls outside the tolerances specified for the deliverer and the new information is used to draw up an alternative dispatch plan.
PostNL is fairly tech-savvy for a large company. We were the first to switch completely to the cloud, which forms the basis for further developments. We are taking steps in clearing legacy and developing new digital capabilities such as orchestration and routing.
We make maximum use of low-code-technology and combine this in our platform with, for example, event streaming and machine learning. Our volumes and performance are unparalleled. We also measure everything that we do, and we know precisely what added value is provided by which technology option. This information offers our developers an immediate insight into the impact and furthers their awareness of the context in which they are working.
I am convinced that software development pivots on value creation, and that intrinsic motivation is the driving force behind innovation – which is why I always welcome an opportunity to drink a virtual cup of coffee with impassioned engineers. When talking about ‘digital at the core’, we may be the perfect match!
Do you want to know more?
Contact Bram Grijzen:
+31(0)6 13 91 72 24