Mastering ETAs: The Key to Resilient and Efficient Supply Chains

An ETA, or estimated time of arrival, is a term widely used in the supply chain industry predicting when a shipment will arrive at its destination. Since the rise of real-time visibility platforms, achieving accurate ETAs has become the holy grail for supply chain organizations given the critical role it plays in boosting operational efficiency, satisfying customer expectations, and avoiding production line halts or store stockouts.

Even if calculating ETAs seems simple at first glance, there are many challenges to overcome to ensure high enough accuracy and reliability for supply chain users to benefit. Here are my thoughts on the benefits ETAs bring, the associated challenges, and how to solve them with a blend of AI capabilities.

1. The benefits of ETAs

ETAs bring plenty of benefits across a supply chain organization. There are three main use cases in which ETAs are helpful in the supply chain context.

Firstly, ETAs help organize logistics operations at loading and delivery sites. Here, the ETA is used in daily operations to manage logistic teams' workload and make sure all loadings/deliveries go smoothly. For instance, it is widely used at retail warehouses and stores where the availability of docks and forklift operators must be ensured, in order to minimize dwell times (i.e. the total amount of time a truck spends on site).

Secondly, ETAs help manage inventory. In the manufacturing industry (e.g.; automotive, petrochemical, CPG), ETAs are looked at by material planners to make sure that critical supplies arrive timely to factories. This ensures the right level of inventory is available at all times. Then, in the event of a supplies shortage, ETAs help transport managers promptly anticipate and react accordingly, by accelerating deliveries or by booking an alternative mode of transport, for instance.

Finally, ETAs help improve customer satisfaction when shared with customer service teams, or directly with end customers. It is a piece of information telling them when the delivery of their purchase will take place. The best example is parcel deliveries in the B2C world for which ETAs are common. ETAs bring the same value in the B2B world, both for parcel and any other outbound deliveries (e.g. containers, finished vehicles, truck loads, etc.)

2. The challenges of delivering ETAs...

Increasing ETA reliability is a fascinating topic. Theoretically, there is an infinite number of real-world variables that can impact an ETA. For instance, variables impacting over-the-road shipments can include vehicle type, traffic, driving regulations, carrier logistic network or even border-crossing times.
So what can be done? It requires a 3-step approach.

First, it is necessary to find out which variables matter the most by analyzing data collected from a carrier network. Differences in the potential impact variables have for each transport mode can be observed. For example, truck load shipments are highly affected by driver regulations and truck swaps, whereas ocean shipments are impacted by changing vessel schedules and the risk of container rollovers at transshipment ports. When it comes to parcel shipments, usually operations within hubs are the biggest driver of ETA variability.

Second, we source additional relevant data for each of these variables. As we are constantly analyzing new variables to understand their value in terms of accuracy gains, we also never stop looking for new data sources, which are usually diverse. For example, a mobile app for truck drivers or the use of IoT devices.

Lastly, the final (but by no means small) challenge is to make use of the data collected to actually compute an ETA. For this, predictive AI capabilities combining machine learning with statistical lead time models is key. Users can benefit from the stability of the latter to predict recurring patterns, such as driver breaks or regular traffic trends, but also from the flexibility of machine learning models that are able to adapt predictions in response to immediate traffic changes or any other unexpected disruptions.

Such a combined approach, makes it possible to provide ETAs for 100% of shipments on every mode of transport, and guarantee up to 90% delay prediction accuracy!

‍

3. ... and why we need to go beyond!

90% is not 100%, it's true. Even with significant investment in this approach, reaching 100% is not technically possible due to cases where the bare minimum of data required is not available, or because data conditions change continously. The good news is that when these cases arise, recent AI advances will help fill these blind spots.

Firstly, a predictive AI layer is able to detect tracking anomalies and can help users know what is actually preventing a given shipment's ETA from being computed. Examples of anomalies detected include errors in transport plans, incorrect trucks or containerships assigned to  shipments, or lost connectivity with GPS tracking devices.

Secondly, a generative AI layer can help filling in missing data (e.g., lost GPS positions) based on route history patterns, reaching minimal data requirements and compute a reliable ETA, without relying on input from other parties.

Finally, in case the previous is not sufficient, an AI-agent is able to take over and automate requests to carriers and third parties to collect missing data on the fly, which reduces human intervention significantly.

We believe that all of these additional layers combined will help the industry moving towards 99% delay prediction accuracy!
image.png

‍

4. Conclusion

Calculating an accurate ETA isn't easy. It requires a continuous focus on data quality, dedicated teams of data scientists, and significant investment. It also requires a collaborative approach between provider, client and stakeholders. Shippeo has pioneered the above methodology and made significant progress over the past years on all areas, significantly boosting the value that Shippeo's ETA brings to supply chain organizations.

At Shippeo, we have been working on solving these problems for the past decade and now provide ETAs for more than 90 million shipments each year for hundreds of industry leaders across the world. If you’re interested in finding out more about how Shippeo can help you streamline your supply chain operations with reliable ETAs, get in touch with one of our visibility experts here.