Total Cost of Trip is the business travel industry’s unicorn. Everyone knows what it is, but no-one has ever seen it.
Usually, companies simply add the employee’s hotel fees to the air fare and the car rental, and then balance this cost against the commercial value of the trip. But what about meals, visa costs, ground transport, meeting room fees, fuel and other ancillary costs? Why aren’t these incorporated into the cost analysis?
The reason is typically very simple – the data.
Calculating the Total Cost of Trip inherently requires data from multiple sources to be intelligently combined together. But the data that it relies on is typically undermined by inaccuracies or even not available. Companies are often beholden to external providers for air and hotel data – which is often rife with errors, gaps or duplications – or they rely on employees to complete their expense reports correctly.
On top of this, the three systems (minimum) that the travel department relies on – the travel management company, the corporate credit card and the company’s expenses management system – will rarely use common employee IDs, and indeed may have several variants of the same employee’s name within the same data set.
Many companies consider this an unsolvable problem, but it needn’t be.
Advances in technology and data science mean that core information can be pulled together to show true Total Cost of Trip information, and even cross-referenced with HR and ERP systems to ensure that an individual’s spend is commensurate with their authority level.
Right tool for the job
It used to be assumed that simple rule-based algorithms would be sufficient to solve this. But in fact, this deterministic, linear approach is not sufficient for travel departments’ purposes.
As we have seen, considering the sources of data that travel departments rely on, a great deal of modelling and cleansing is necessary before any analysis can take place. Missing information needs to be back-filled, duplications need to be removed and employees need to be consistently identified across data sources, regardless of any differences in systems.
To correct this manually is a prohibitively arduous process. It would take too long to do and would be a never-ending battle as more and more erroneous data is added.
Travel departments therefore need to use intelligent technology to triangulate across all data sources, no matter how they are structured, and confidently identify each employee and trip across the systems and automatically build the total cost of each trip. Moreover, the technology needs to adapt to new types of error or data conflict as they occur and learn how to deal with new data sources as they are introduced.
This degree of automation and intelligence capability requires machine learning.
But this analysis is about more than simply producing a single number at the end of every trip that includes all elements of cost. The degree of granular insight that sits behind this output means that important questions can be answered for the business.
The obvious insights of how much has been spent and where are easy. More useful information includes detailed analysis of whether spend is going through a travel management company, and whether certain individuals or departments are less or more likely to book their travel correctly.
Knowing where your spend is going and stopping unsanctioned spend is vital for negotiating better deals with travel management companies, who need reassurance of specific volumes of business before granting rebates.
Similarly, understanding this data will allow travel departments to identify any inconsistencies between suppliers’ headline costs and the eventual real costs. Hotels may quote certain room rates, but the employee ends up having to spend considerably more on necessary extras that were not included, such as meals or Wi-Fi. Meanwhile another hotel may have higher headline costs, and ordinarily be rarely used because of it, but in actual fact includes more extras within that rate.
On an employee level, real-time insight from machine learning allows travel companies to manage an employee’s spend during a trip and change behaviour before it is too late. An employee’s travel costs can also be proactively balanced with the value of the projects the travel relates to, or with the employee’s role and authority level within the company.
Determining the accurate and reliable total cost of every trip is currently an unachievable dream for most. The data simply is not available or trustworthy.
With machine learning however, the data can be aggregated and made consistent and dependable.
This leads not only to a new appreciation of travel costs, but most importantly, the opportunity to make deliberate choices that highlight efficiencies and reduce spend.