If Only You Spoke Esperanto

The premise of Esperanto is extremely compelling: invent a simple universal language that would be easy for every person on earth to learn so we can all communicate with each other flawlessly. No more “lost in translation.” In 1921, The New York Times quoted a research by the French Chamber of Commerce that found Esperanto to be the best language for business.

Ĉu vi parolas Esperante?

Chances are the answer is a resounding Ne.

Over 125 years later, global commerce makes up over 60% of the world’s economic output, yet the number of people who can converse in Esperanto is less than 100,000. Despite being a relatively easy to learn language, very few people have made the effort. Other attempts at constructing an international auxiliary language such as Ido and Interlingua gained even far less traction.

In the technology-enabled commerce world, new Esperantos are repeatedly created in the form of “universal” protocols and standards, with similar results. RoesttaNet and EDIFACT are two of the better known examples. Even when it comes to more narrowly defined vocabulary such as the classification of goods and products, we find multiple competing standards (e.g., UNSPSC, CPV, [email protected]).

Should we give up on the attempts to create universal languages and protocols? No, but until such a language becomes a reality, we need other means to facilitate communication between buyers and suppliers that use different systems, protocols, formats, languages, and units of measurement.

A big reason EDI and other supplier connectivity methods have such low market penetration is the effort and cost associated with data mapping, format conversion, and translation of trade terms from buyer to supplier systems and vice versa (e.g. from the supplier ERP/Accounting system to the buyer ERP/Accounting system).

The variations and permutation for each field can be extensive. For example, a part can be called part, item, unit, P/N, etc. Date field can refer to shipping date/transaction date/payment date/order date/etc. and appear in different formats. Additionally, the need to sync catalog numbers and units of measurements demands a huge on-going investment from all parties. Multiply this by the number of fields, document types, and trading partners and you get a pretty messy picture and a significant challenge on your hand.

Defining, coding, testing, and maintaining one-to-one maps for each buyer-supplier connection is costly, ineffective, rigid, and error-prone. That’s one reason EDI doesn’t scale.
Attempts to remove these barriers have led to the emergence of canonical data models used by B2B integration gateways. The idea is that once buyers and suppliers map their systems to the canonicals used by the B2B gateway, they can then communicate with any other company that has made the same effort.

While this is a step forward, it has hardly made a dent in the ability to scale B2B supplier connectivity. For one, the mapping and translation efforts are still rather substantial, typically months for each supplier. Since not all buyers use the same B2B gateway, suppliers would still have to map, code, test, and maintain multiple connections.

So what can we do to scale B2B supplier collaboration? Until we have a truly universal protocol, the only hope for low-cost, rapid, and scalable supplier connectivity is automating the mapping and translation process. Such automation cannot be achieved by brute force. It requires smart machine learning algorithms that can abstract the different data structures and enable a universal representation while eliminating the need for explicit one-to-one mapping. This is not a simple task, and it’s why Nipendo was recently awarded a patent for mediating digital transactions between systems.

Once the mapping and translation barriers are removed, a new world of possibilities opens up, such as automatic association of all related transactions to create process orchestration business rules that are agnostic to the format and the terminology used by the buyer or supplier system.