responsible gaming

3 questions for gaming operators before investing in machine translation for player support

5 min

London, UK

The online gaming industry has reached all corners of the planet, with new players for casual and real-money games coming from Peru to the Philippines. This is due to ease of access via mobile phones, growing discretionary spending, and a change in consumer habits during the pandemic.

The industry is raking in billions every year from emerging markets – but these players cannot be taken for granted. As they become more discerning, emerging market players will drive towards operators that provide a smooth, localized experience.

Fail to localize and run the risk of players flocking to more authentic competing brands.

But what does it mean to localize?

  • Local languages. Players in Nigeria who speak Yoruba natively should never be told to speak English to get support – this doesn’t respect player preferences nor the local market.
  • Omnichannel. Players in Vietnam who don’t own a laptop should not have to use desktop webchat – these players should be engaged via their preferred mobile messaging channels, such as SMS or Zalo.
  • 24/7 instant support. Players in Colombia should never be told to wait hours or to send an email because support agents who speak their language are yet to start a shift.

So is the only option to invest in a costly 24/7 contact centre operation with multilingual agents? No – but there are 3 questions to consider before spending on this problem.

1. Is machine translation really the magic solution?

Proto is often requested by gaming operators to apply machine translation to 100% of their multilingual player support queries. At first glance, this sounds ideal: Thai players could be supported by English-speaking agents from Malta, without any wait times.

In reality, emerging countries have diverse and rare languages that are under-resourced – this means that translation options like Microsoft Azure and Google Translate will have critical errors with certain language pairings. This is further complicated by industries such as gaming, which have unique slang terminology (e.g. 流水 meaning both water current and amount of the offer) and special terms (e.g. sports team names like Crystal Palace).  

While agents will do their best to decipher translated player queries, your agents are only human and may get overwhelmed. Likewise, players who receive translated agent replies could be easily confused and give up altogether on your brand.

While you could customize translation models directly or with partners like Proto, this effort will take time and may never reach the level of accuracy desired for all languages.

Therefore, machine translation should be the tool of last resort, ideally for less than 10% of queries to ensure player satisfaction and brand reputation. So, what about the 90%?

2. Are chatbots (and NLP) fully utilized?

Machine translation is not the only tool for the localization problem. Gaming operators are increasingly adopting chatbots for player support. Proto’s competitive survey of the industry shows these chatbots are still limited to European languages and basic FAQ handling.  

This is a major under-utilization of the potential of chatbots for localization due, in part, to a lack of focus of most chatbot vendors on the underlying Natural Language Processing (NLP) technology that powers the bots. But a handful of vendors do have their own NLP engines, such as Freshworks (FreddyAI) and Proto (HermesAI). Proto is unique in the industry with its dedicated R&D focus on emerging market languages, such as Tagalog, Kinyarwanda, and Yoruba.

Applying localized NLP-powered chatbots to player support allows native-language understanding and instant replies for multilingual player queries. This removes those destructive machine translation steps in odd language pairings. The outcome for agents and players is less stress and confusion, with improving NLP performance over time. For operators, this can be achieved for more or less the same spend as the translation model customization effort.

Once the localization problem is under control with an NLP-first solution, the chatbots can be upgraded for advanced support functions, like player KYC (authentication & tagging) and proactive cross-selling (API integrations).

3. Has responsible gaming been considered?

What is responsible gaming doing here? Player support queries are a critical resource for detecting misconduct and problematic behaviour. This is especially important in emerging markets, which have high and unfortunate volumes of underaged and loan betting.

As regulators in these markets evolve, localized customer support tools with strong player protection processes will be mandated for operators. This is seen in the financial industry in these countries, where regulators are using localized Proto chatbots across financial operators to help consumer protection catch-up with rapid advances in fintech.

Gaming operators aiming to hold long-term market share in these countries should be applying a responsible gaming lens to their localization effort, and searching for vendors that maintain strong alignment with local regulatory policy – this will ensure a lower cost of compliance in the future.

TLDR? Direct 90% of multilingual player queries to local language chatbots before using machine translation between players and agents. Here’s a map of Proto’s AICX solution in a national gaming ecosystem – machine translation is a last resort:
Learn more about Proto's AICX solution for gaming.

Curtis Matlock is CEO of Proto. He leads globally distributed teams to achieve social impact and growth objectives in emerging markets. To reach Curtis, please write to him at