For many logistics operators, route optimisation has traditionally been viewed as a question of distance, traffic, and sequencing. But for delivery networks optimised with AI and routing technology, the biggest source of inefficiency is not on the road anymore. It is at the stop.

Every delivery network evolves over time. New depots are added, different vehicle types are introduced, service offerings change, and customer expectations become more complex. As these operations grow, layers of assumptions are built into the plan. However, average stop times, buffer periods, and manual workarounds become accepted as normal, even when they don’t actually reflect the actual length of time it takes to make a delivery in the field, skewing the rest of the delivery process.

Andrew Tavener, Head of Marketing, Descartes believes this is where AI has a more practical role to play. While technology has made routing on the road more efficient, AI provides a fantastic opportunity to examine real operational stop time data, helping businesses understand how long jobs actually take, which variables affect performance, and where hidden inefficiencies are costing time, money and customer satisfaction.

Looking Beyond the Road

Route optimisation is often thought of as getting from A to B. No two deliveries are exactly the same and depending on the goods being delivered, things can be more complex. For example, if you’re delivering white goods (e.g., large, heavy electrical home appliances), the product may need to be installed. Operators have to be more conscious of the time slot a customer has requested and the type of location they are delivering to. Is it a commercial property? Is it a flat? Is it a house? All these factors must be considered when determining how long it will take to complete the delivery.

It’s also important to consider the delivery crews skill set and driver tenure. Even the time of day on a specific street or road can affect the speed of delivery. However, another important thing to consider is what happens when the driver gets to the delivery location, as significant time can be spent at the stop rather than on the road.

Many operators still believe that offering an approximate delivery time is enough, but it doesn’t consider all of the real-world parameters and, when something goes wrong, customers are left feeling frustrated. The commercial risk of this is damaging when you consider younger consumers, with 21% of under-35 buyers saying they would stop ordering from a retailer after a negative delivery experience. This is an area where there can be significant compounding value from the use of AI.

One practical use of AI in logistics operations is learning from real service-time data and feeding this into predictive models that recommend how long a stop should take based on the specific job, driver, vehicle, location and operational context. Optimisation tools can then use these insights to recommend where jobs should be reassigned, particularly where one operative is likely to complete a task more efficiently than another based on learned performance patterns rather than theoretical averages. Here, AI can deliver immediate operational value.

By consuming and learning from real-world data, logistics teams can make faster, more accurate decisions, reduce unnecessary delays, and improve the reliability of route plans. Ultimately, the goal is not just to create faster routes, but to create achievable routes that reflect what actually happens at every stop.  

The Hidden Cost of Inaccurate Stop Times

Essentially, if route optimisation is accurate on the road, but is not accurate from a stop and service time perspective, it is not as effective as it could be.

A practical example can be seen in an operation where the same average stop time is applied across different drivers, job types or vehicle setups. If one driver is relatively new to the business and another has years of experience, should they really be allocated the same service time for the same task? The same applies to vehicle type, equipment, access requirements, and site complexity. One driver may have specialist equipment available, while another may not. All of these variables influence how long their stop time will take.

The impact is felt by both customers and the business. If a two-hour delivery window is missed, a customer may be unavailable, the delivery may need to be repeated, and the customer experience is immediately damaged. For the fleet operator, inaccurate stop times can quickly compound across a route, creating overtime, operational dead time, greater reliance on third parties, and a higher cost per stop.

The same issue can also impact revenue. If deliveries are not balanced effectively across routes, operators may miss the opportunity to add an extra slot per driver per day. When stop times are managed effectively, a business’s On Time in Full (OTIF), profitability, and Customer Satisfaction (CSAT) all improve. But that only happens when stop time is treated as a core part of route optimisation, rather than a secondary detail.

Why AI Still Needs Human Oversight

Because operational nuance and judgement remain important, fleets can leverage AI to recommend the actions to take rather than letting them make changes autonomously in the background. Human operators have knowledge and insights that AI doesn’t; for example, they may know a customer specific requirement or a current one-off delivery risk. It’s crucial to retain this type of human knowledge before any recommendations are acted upon.

What’s more, AI is only as reliable as the data it receives. This means that the likes of incomplete stop time records or inaccurate delivery data all undermine the quality of the recommendations an AI tool can give. Therefore, when significant risk is involved, AI must be handled correctly. It should also be iterative, with operational teams reviewing the results and correcting inaccurate assumptions or how it is applied over time. Ultimately, the strongest results for businesses will come when experienced operators, robust processes, and AI work together.

Conclusion

It’s clear to see that logistics operators must look past the fastest route and focus on what happens at every stop. Utilising operational data enhanced with AI, combined with keeping human expertise in the loop, can help businesses truly understand the impact of each delivery location and customer requirement by developing more accurate plans, reduce wasted time and improve customer satisfaction. In turn, this will unlock sustainable efficiency gains to help logistics businesses in the long run.