Digitalisation versus the Four-eyes Principle – Data Analytics in ICS

Why is an internal control system and the associated qualitative audit of a process in the export department important?

The answer to this question depends on who you ask. From a purely legal and regulatory perspective, compliance is necessary to avoid legal risks. Permits in particular represent a leap of faith on the part of the authority, which can also be quickly withdrawn if you do not prove to be trustworthy.

Holders of customs authorisations are obliged to monitor their customs processes and to ensure quality and compliance through internal control systems (ICS).

The customs administration is interested in quality assurance. This is explicitly asked about at various points in the self-assessment questionnaire in the EU Commission’s guidelines. For example, 1.3.2 of the EU questionnaire asks:

1. How and by whom is the customs tariff classification of goods decided?
2. What measures do you take in the area of quality assurance to ensure that the customs tariff classifications are correct (e.g. audits, plausibility checks, internal work instructions, regular training)?
3. Do you keep records of these quality assurance measures?
4. Do you regularly monitor the effectiveness of your quality assurance measures?
5. What resources do you use for customs tariff classification (e.g. databases with goods master data)?

Efficiency issues are another aspect of quality – good data quality minimises the manual workload. This not only saves valuable work resources, it also allows you to react much faster.

In the end, even customer service benefits from the fact that you can act quickly and in a legally secure manner when handling imports and exports. You will become a reliable, competitive player in the international economy.

And the quality of your data is also good for your balance sheet: You not only avoid paying fines, but also avoid the looming uncertainty of possible additional payments. You may also discover that some goods can actually be imported at a lower duty rate or are covered by a duty exemption.

In order to know how to build a control system, you first need to know which processes are affected and identify the key figures in these processes. In other words, what is the desired output and outcome? Let’s take master data relevant to foreign trade again as an example.

Quality is characterised here by predominantly correct and complete information. If you have stored high-quality master data, the dispatch department no longer has to ask the purchasing department whether this or that document is available. You will also no longer discover screws that have been classified as “parts of …”. The high quality of master data also enables the digitalisation and automation of processes. The software used for this requires good master data in order to work optimally.

If, for example, 2,000 articles are created each month, 2,000 correct weights, customs tariff numbers and goods descriptions should be associated with them as a result of the master data process.

How can the quality be checked and kept at a minimum acceptable level? While also being statistically valid, as well as having a balanced cost-benefit ratio?

If the classification is checked using the four-eyes principle, you can assume with relative certainty that the quality is very high. But at what price? As the number of items to be tariffed increases, one department can no longer handle it. This approach is also no longer in keeping with the times today. What would be an alternative approach? A sampling procedure according to DIN ISO 2859-1 AQL (Acceptance Quality Level) would be suitable here, for example, in which you only check a certain quantity again manually and can therefore draw a conclusion about the quality of the entire group of articles. The AQL method is useful for initial quality checks of the entire master data as well as for the subsequent continuous quality checks. Once the entire master data is checked, any need for action can be derived from it. With the continuous quality check, which takes place e.g. quarterly, the company can intervene in good time if inconsistent data has been entered into the system and correct it in good time.

In addition, not only the internal master data quality can be controlled, the delivered quality of internal and external shared service centres or service providers can also be monitored in an uncomplicated way.

Another digital solution would be the rule-based checking of fields in the ERP system and comparison; the use of artificial intelligence can also be useful for quality checks.

The processes surrounding the creation and modification of customs-relevant master data can now be handled much more efficiently than in with four-eyes principle. However, each company must examine for itself which option is sensible and efficient to implement. Such solutions certainly require a certain number of data sets. But it also goes beyond master data processes.

Customs clearance is a highly data-driven process. Most errors and delays in the outcome of a customs declaration, on the import or export side, occur because the data for such a declaration is wrong or incorrect. This must be effectively identified by means of ICS as well.

One possibility for smaller companies, for example, is a sampling procedure. In this case too, you check a certain number of operations for correctness in order to conclude on the error probability or quality of the entire group. Other approaches to internal control systems in the import procedure now include the automated comparison of tax assessment notices and ATLAS registration data with the available data from the accounting department.

A unique and digitally evaluable reference, e.g. via invoice number, must be provided for this. In this case, the digital control system can identify inconsistent registrations by mapping the registration and invoice data.

There are many processes in export and customs that can be made measurable. Processes and their measurable results are a prerequisite for an internal control system, whether it is manual or automated.

Internal control systems also mean regular reporting in order to carry out target-actual comparisons and to monitor developments in process quality. If you decide to use sampling or automated testing, it should be done over a longer period of time in order to get meaningful comparisons and results. The question is also what to do if the results of your quality audits from samples or other procedures suddenly become “worse”.

In this case, we recommend looking for the causes and taking measures, e.g. training new employees if necessary and validating processes again with service providers. Are there new product groups, countries or other factors that could require a change in quality auditing?

There are now much more efficient ways of setting up internal control systems than with the classic four-eyes principle often associated with ICS. It will be necessary in future to work even more with digital technologies in this area. The export department and those responsible for customs and exports will not only have to make themselves knowledgeable about the technical issues surrounding customs, but also about the digital options. Customs processes generate a lot of data that can be analysed, evaluated and interpreted.