For international expansion, localization is a must, and with dubbing preferred over subtitles by over half of consumers in Brazil, France, Italy, and Germany, this format is the way to go. However, traditional dubbing is expensive and time-consuming, and therefore not an option for topical or time-sensitive video or video distributed on YouTube, which is when media companies turn to AI dubbing. However, navigating the field and its many providers can be tricky, so we've laid out a framework for a successful localization project, focusing on the key decisions business leads must make.

Navigating the landscape

For any digital media or streaming company, a successful localization strategy begins with a crucial decision: which type of technology to use. When it comes to AI dubbing, there are a few options. The best voice generation platform for each company will vary depending on their chosen use case. There are lasting implications for the rest of the localization process, depending on the option picked.

The first – and most important – choice a company needs to make is whether to go for an end-to-end video dubbing service or to pick a fully automated do-it-yourself model. 

Fully automated voice generation services 

Fully automated voice generation services offer AI dubbing alone. Users upload their videos to the platform in question and receive either a translated audio track or a localized video in return. Although turnaround can be instantaneous, these services don’t offer built-in quality assurance or video editing. Instead, users have to edit transcripts and translations for accuracy themselves or source their own expert editors. For these reasons, these services are best suited to smaller volumes of video or small content tests where the output doesn’t need to be perfect. 

End-to-end video dubbing services

End-to-end video dubbing services take care of the whole localization process, from translation to quality assurance to post-production video editing. They lend themselves to use cases that require higher quality video output since they guarantee production-ready results. Since these services tend to include human-in-the-loop checks and account management, they’re also a more flexible option. They can support different input and output video formats, or tailor results for specific streaming or social use cases. 

End-to-end localization is better suited to larger volumes of content since the workflow behind these services is designed to scale with little input from the end customer. Each project has dedicated project and channel managers who post and optimize the content on social media and work with distributors to supply the correct formats for streaming platforms. A company that chooses an end-to-end service has far fewer ongoing demands made of them. They also benefit from access to specialist knowledge and advice that can support their ongoing localization strategy, such a monthly reporting.

When choosing between end-to-end solutions, there are still important things to look out for, such as checking what voice technology is used and whether the size of the machine learning team can support the amount of voice and language options available to an ongoing high standard. Asking these questions when evaluating potential AI dubbing tools will avoid any nasty surprises when it comes to the performance of the localization service. At Papercup, we use best practices such as bespoke synthetic voices, expert translator quality assurance, and continued machine learning investment to maintain the high brand standards for trusted brands like Bloomberg, Sky News, and Insider. 

Deciding on content type and localization language 

Once confident in provider choice, companies then face the challenge of understanding what content types will resonate where, and which regions or languages will perform best for their brand. To make a localization strategy a success, the target audience must be receptive to a company’s content and highly engaged with the chosen format. 

Luckily, there are ways to confidently decide on a strategy for both. 

Social and digital media use cases

For content owners distributing on social media, past performance metrics can provide the key indicators of future success. By looking at which regions content currently performs well in, companies can get a good idea of where there might be opportunities to better engage audiences in their native language. For example, Papercup client Sky News initially identified an opportunity to better engage its global audience when it noticed significant engagement from Spanish-speaking regions of Latin America. Realizing many existing news stories were of interest to LATAM populations, Sky News partnered with Papercup to introduce a dedicated Spanish-language channel that received 26 million views and 96,000 subscribers in the first 12 months, with a long average watch time. 

To maximize monetization potential, our media clients generally localize into languages that meet two key criteria. In regions where English-language content is already showing good traction, they look for high CPMs or large populations. High CPMs indicate strong advertiser demand and are the biggest factor in overall advertising revenue. Regions with large populations but slightly lower CPMs can bring in decent revenue due to the volume of views and healthy watch times outweighing the slightly lower cost per advertising impression. 

Streaming use cases 

Existing performance metrics can also help content owners distributing on streaming decide on where to focus. If these insights are lacking, owners looking to distribute globally can pair this approach with a low-risk localization pilot to confirm content demand, as our customer Fremantle does for Arabic-speaking countries.

Content owners distributing on streaming should also factor in the impact free ad-supported streaming has on monetization potential. By capitalizing on audiences in markets where free ad-supported TV is growing, they can maximize revenue across the content lifecycle and strike deals for different distribution windows. This helps diversify content strategy and unlock the potential of back catalog content. 

Audience and market research 

Just because content has been successful in its home market does not mean a direct translation of that content will be successful in another market. The first season of the US office tanked until it was adapted (localized) to suit the humor of the new target American market. And while the animated film Inside Out was an international success, its global appeal was partly due to localization efforts to make it suitable for different audiences. In Japan, broccoli was switched to green peppers in a scene about ‘disgust’ because broccoli is a big favorite for Japanese children. Leaving the scene as it was would’ve caused alienation and confusion. In several countries, ice hockey was switched to soccer, which is a more familiar sport.

It’s clear, then, that an understanding of the target market and its audience is critical to the success of any localization project. Choosing an AI dubbing provider that has a real team of translators checking for accuracy and applying brand and local-language phrasing can make the difference between a successful localizaation project and a failed one.

Assessing content for streaming

To evaluate audience response, companies can either test content with a localization pilot or carry out market research using audience intelligence and social listening tools to prioritize content that’s a fit for the target market.

For companies evaluating FAST, certain genres such as documentaries, cooking channels, and news are evergreen high performers – although the relative popularity of these FAST genres varies by region. 

While content that performs well in the US or UK won’t necessarily be a great fit for other audiences, it’s worth considering if existing content that’s underperforming could resonate strongly elsewhere. For example, FuseTV, which originated in the US, has seen success after localizing for LATAM Spanish audiences. 

Assessing content for social

What works for brands on social media varies, and demand for certain content or topics can changes frequently. To make a confident decision about social channel strategy, it’s therefore critical to pilot social content rather than just relying on industry trends alone to validate which areas are worth investing in a full localization strategy 

Luckily, the barrier to entry when localizing for social channels is low. This is helped especially by new features such as YouTube’s Multi-language audio tracks, which allow content owners to test dubbing tracks for new audiences without the need to create a separate YouTube channel. 

After seeing success localizing for a new Spanish-language audience using Papercup, content distributor Fremantle is now using AI dubbing to carry out localization pilots for new areas. Fremantle is testing Arabic-speaking regions for two years to assess growth in traction and audience demand for its ‘Got Talent’ content, de-risking its international expansion. 

Localization metrics

Finally, at the onset of any localization project, it’s important to decide on metrics of success. While these can vary depending on business objectives, there are a few measures we suggest all of our clients track to assess the impact of localization. 

It may be that for larger projects, there is more than one metric allocated to a specific team or part of the process. This way, on review it’s easier to see where the roadblocks were and how best to tackle them.

Metrics for social

Since the aim of localizing for social channels is to tap new audiences or increase engagement among those previously watching in non-native languages, success metrics for social localization should prove increased reach and improved watch times. Ultimately, this is all in service of increasing advertising views, which will improve monetization. 

Some metrics to track include: 

Video views: If these increase significantly post-localization, especially for target regions, this suggests a new global audience has been unlocked.

Watch time: A higher watch time for localized vs. non-localized content shows viewers in target regions are more engaged – which is important for monetization, as higher watch times mean more advertisements can be shown. 

% of content localized: By tracking what percentage of all content intended for localization has been completed, companies can assess localization efficiency and address roadblocks quicker. This metric might be tracked as part of a strategy with phased objectives – first to localize a bank of content (where this metric would come into play) and then to deploy that content as part of a wider content strategy.  

At Papercup, our clients tend to first test demand, then build audiences, and then look to capitalize on increased engagement and reach through monetization opportunities. 

Metrics for streaming

When evaluating streaming success, companies can either track raw metrics such as the number of views or advertising impressions for a given region, or they can take a more holistic approach and assess overall supported language revenue. 

This involves breaking down monetization into input from each localized region, to understand the worth or potential of a given project. For Free ad-supported streaming, that can be as simple as assessing CPM and the number of impressions for each region. When brokering agreements with more traditional streamers, distribution deals will play a larger role in overall content strategy. 

==

Building a localization strategy can be complicated, but with planning and the right framework, it doesn’t have to be. Research is key, as is prioritizing languages with the greatest revenue potential, understanding the localization methods available, and choosing the right localization partner. 

At Papercup, we have years of experience localizating for media and streaming leaders such as Cineverse, Bloomberg, and Fremantle. To learn more about how we support businesses along their localization journey, book a consultation with one of our experts.