By laying the groundwork for scalable processes and reducing operational uncertainties to consistently deliver optimal results, robust mathematical frameworks, such as optimisation, are key enablers of the On-Demand economy.
Meeting customer expectations in the On-Demand economy
As the novelty of the digital age is slowly wearing off, more and more of us have become accustomed to having an endless supply of information and entertainment at our fingertips. With this familiarity has come an increasing expectation of immediate access to other types of services, from sectors as diverse as healthcare to public transport. Technologies, such as 4G, have also brought with them a radical shift in consumer behaviours towards online service applications rather than independent purchases. Customers can now connect with an effectively limitless amount of goods and services, even on the move, and this trend is driven by a preference towards rapidly deliverable, mobile centred, service providers – services On-Demand.
Technology companies such as Amazon, Uber, and Deliveroo have pioneered the high standards expected by consumers today, and with the upcoming launch of 5G services across many mobile providers, these will almost certainly continue to increase. To remain competitive, all service industries must now adapt to this changing landscape, and in doing so they face new and complex operational challenges. In the On-Demand economy, customer experience is directly affected by mobile communication, transparency, and personalisation, delivered via careful application design and advertising. But the loyalty of customers remains fundamentally tied to the key metrics of reliability, punctuality and cost. Customer reviews claiming slow delivery, or poor value for money, have the potential to cause significant reputational damage and reduced sales in the On-Demand marketplace.
Balancing supply and demand
Companies experiencing significant success in this environment have one common denominator: sophisticated, scalable methods of real-time supply and demand balancing. Meeting day-to-day fluctuations in demand already requires flexible service delivery but coordinating operations in the most time efficient manner possible presents an additional challenge, which is far from trivial to address. Fortunately, the processes required to orchestrate flexible service delivery, such as scheduling and dispatching resources, can now be automated to provide optimal results; otherwise a business risks operational waste, the forfeit of potential profits, and the dreaded impacts of customer dissatisfaction.
Food for thought – an example
Take an On-Demand food delivery service, such as Deliveroo, as an example. After an order is placed on the app, the restaurant is notified, and an expected pick-up time established. As this time is neared, a delivery driver is selected who must collect the food as close to the pick-up time as possible and deliver it in the fastest possible time – failing this, the food could arrive cold or spoilt. If there are few orders, and plenty of drivers, this problem is simple enough; but as peak hours approach, many drivers will be carrying out drop-offs as food becomes ready. Ensuring the right delivery driver is selected for each order, becomes the difference between a satisfied customer or negative review, so there is significant incentive to getting this right, every time.
As with most resource dispatch problems, this scenario will always have an optimal solution. If the pick-up, drop-off, and driver locations are known in real-time, then traffic weighted travel times can be calculated; reducing this complicated choice to a mathematical question with a definitive answer. Procedural complexities such as a mixed vehicle fleet or driver working hours agreements, also play a role in determining operational efficiency and may, therefore, require incorporating into these calculations. Advanced dispatch tools capable of solving these mathematical problems become essential as the scale of operations grows.
Watch the full video here
The technology that offers the capability to continually optimise the scheduling and dispatching of resources to cost-effectively meet demand is Mathematical Optimisation (also known as Mathematical Programming). Optimisation comprises a field of problems in which an objective can be defined (a quantity you want to maximise or minimise e.g. average delivery time, in the example above) and feasible solutions to the problem can be ranked according to their objective value. So, for food delivery, the possible configurations of driver assignments to deliveries (our feasible solutions) will be ranked by the average delivery time for the expectant customers, or a similar objective focused on achieving the best possible customer experience.
Mathematical Optimisation has been an established scientific discipline since the 1950s, but it’s full potential to real-world applications has only been realised in the last few decades thanks to the phenomenal progress made by the developers of commercial optimisation solvers, such as Gurobi. The incredible advances realised by computing power in recent years has enabled such solvers to become financially viable options for businesses to solve large-scale optimisation problems in close to real-time, once problems are correctly formulated in mathematical terms.
Why invest in a bespoke optimisation solution?
Off the shelf tools are readily available and produce solutions to generic operational problems. For many companies, though, it is the differentiating features in their service that attract and drive continued customer use. Sustainably delivering a personalised experience, up to the ever-present high consumer expectations, now requires unrivalled operating efficiencies; falling short of this, a service risks being outcompeted.
No two companies are the same and neither are the challenges they face in On-Demand service delivery. Possessing a tailor-made optimisation solution, fit to your company’s unique offering and operational challenges, can provide a much-needed and sought-after competitive edge.