Transforming passenger data into actionable intelligence at the border

With regards to data collection, a country’s border offers one of the richest touchpoints for Governments looking to monitor their territory. Beyond the border, these touchpoints are few and far between, with both the quality and quantity of the data collected reducing significantly the further away you move from the border.

In the simplest sense, this data provides insight into who is traveling to the country, when and for how long. But the scope for analysis and examination extends far beyond this – yet many governments are still failing to fully grasp the opportunity with both hands.

Whether it be for more effective profiling and immediate decision-making by law enforcement or for trends analysis and long-term policy development, when utilized intelligently this traveler data can transform the way a country policies its borders.

The State of Data at Borders Today
In many countries Border Agencies are still overly, if not completely, reliant on manual data collection. Passengers are handed paper forms on the plane, to be completed before arrival and presented to Immigration. In cases where a Visa on arrival is required, the same or similar data must be provided by the traveller, albeit in a different format, via a separate paper form. The issue with this approach is that it is wildly inefficient, ineffective and inaccurate.

The Seychelles, a country that has recently undergone a major digital transformation, no longer issues paper forms to in-bound passengers. By moving data collection online, the Border Agents can now border enter a person in less than forty-five seconds. Previously, it would have taken 3-4 hours and 4-5 Immigration Officers, just to manually enter the data from the 300 paper forms into the online database.
Why this matters is twofold: manual data entry leaves room for mistakes, and Border Agents focus on administrative tasks instead of the specialist profiling that they are trained to do. Illegible forms can render entire datasets redundant and undermine the purpose of collecting such detailed data in the first place.

Additionally, this passive collection of data leaves much to be desired when it comes to security and risk-assessment. The general consensus in the sector is that data, when used effectively, gives governments the ability to shift from a reactive to a proactive approach to their borders. Digitalised borders allow smart analysis to prevent and predict issues – with Border Agents intervening before a problem arises in-country. Digital systems identify known unknowns and unknown unknowns before arrival or at the border, supported by the intelligent application and analysis of data and passenger information.

All of this enhances security, and while completely avoiding risk is never realistic, minimizing the risk of a country unknowingly allowing a high-risk traveller to enter should always remain a priority.

Navigating API-PNR requirements
Since February 2021, all UN member states have been mandated to implement API-PNR systems. The goal of such a policy is to enhance data collection at the border, and improve data sharing across borders in a bid to tackle transnational crime and fight international terrorism.

Two years later, many countries are still not compliant. These countries face immense pressure from the UN and ICAO to introduce new systems to facilitate responsible and effective data collection. The information, collected directly from the airlines, provides advanced knowledge about who is on the plane, such as how they paid for their ticket, who they are seated next to and how much luggage they checked in. Such information can be used by Customs, Immigration and other law enforcement to build a more detailed picture of the individual and make smarter decisions as a result.

API-PNR data can offer an extremely strong foundation to develop insights about travelers but when applied on its own, has some severe limitations:

  1. API-PNR data is demonstrably limited as a targeting tool, focusing on known high-risk individuals and not unknown risks (i.e. those not on watchlists).
  2. PNR data is not always consistent which can compromise the quality or completeness of the data. This will inevitably have analytical implications.
  3. API-PNR is primarily a technology for the air transport sector, leaving maritime or land borders vulnerable.
  4. Interactive API (which vets individuals during the check-in process) issues a binary board/not board directive to the check-in agent. Due to the short timeframe for vetting, the profiling methods used are relatively basic and can be inaccurate.

As threats become increasingly sophisticated, border agents are best served by a combination of multiple datasets – each offering their own unique layer of information to deliver truly complete visibility at the border.

Transforming data at the border in 5 steps
Today, innovative technology companies are disrupting the border security market by offering powerful, intelligent and automated solutions.

As a government looking to improve existing tools or even to implement a solution from scratch, there are a number of things that need to be considered.

  1. Moving from collection to analysis
    It is fundamental to shift thinking away from the passive “collection” to the proactive “analysis” of data at the border. When done right, smart data analysis enables more effective profiling and risk-assessment followed by real-time decision-making. As a result, governments can expect to see a reduction in high-risk travelers at the border and successful mitigation of security incidents both at the ports of entry and in-country.

A more effective approach is a collaboration between data scientists and border security agents – leveraging the skills of both to develop a powerful risk engine. Border agents can use their expertise to determine the data / indicators that would be necessary and relevant to their roles. The data scientists then determine where it can be obtained and together determine how best to use the data to gain insights.

  1. Use multiple data sources
    Using multiple live data sources together elevates capabilities 100-fold. Using API, PNR, Visa, travel authorization and permit data together can significantly enhance efficacy and accuracy when profiling. With the right technology, biometric and biographic data is combined into a single view of the traveler, including supplementary data sources such as coded seat maps, airport data, airline data for a comprehensive overview of each individual.

Underpinning all of this functionality, however, is robust identity resolution, to reliably link an individual across all data sources. While this can pose a challenge, the collection of facial biometrics from the traveler provides the building blocks for accurate and secure identity verification using artificial intelligence.

  1. Automate alerts and profiling for effective short-term action
    Automation is key for combing through vast amounts of data efficiently, and alerts such as watchlist matches can be automatically flagged for further (manual) assessment. Profiling rules for known risk profiles based on input from border security subject matter experts can enhance the screening process. These rules must be tied to the country and their specific threats. Predictive profiling using AI for complex profiles that cannot be easily articulated performs the same task. However, authorities don’t need to specify the combination of data attributes that constitute a risk profile, the AI does this with sufficient examples of high risk travelers.

All of this allows the border agent to focus on manual risk assessment of just the 1% of high-risk travelers, ensuring the responsible use of AI in what is a highly-sensitive sector.

  1. Visualizing the data
    Border Agents shouldn’t have to be data scientists in order to access and use the data that they have access to. Next-generation border systems will provide tools for border security agents to mine the data, using their experience and intuition. From user-friendly interfaces, to intuitive design, the platform itself is developed with the end-user in mind. By balancing ease-of-use and power, the system can be up and in use quickly and limited training is required.
  2. Constant improvements and updates
    Performance management is key when it comes to handling constantly evolving and highly-complex environments. In order to be effective, there is a need to establish a feedback process for users of the platform. This will enable the data scientists to understand how accurate the system is, in practice, and assess ways to improve. The border agents can then identify practical interventions based on how well the watchlists are performing; whether there are too many false positives or too few hits; or whether the thresholds for biometric matching need to be adjusted.

There is lots of talk of the “digital border” throughout the sector and many new providers are stepping up to the plate to deliver the next-generation of border management solutions, directly to Governments. While eliminating the need for paper or increasing efficiency are undoubtedly key objectives for many, perhaps the most powerful outcome of digitisation is the access to unprecedented amounts of data. When collected, analyzed and applied correctly, this opens up a world of opportunities for border agents and policy-makers looking to tackle modern day threats head-on.

By Renaud Irminger, CEO, Travizory