One Campaign. Every Language.

Marketing's Own Tower of Babel

MyHeritage opened its Optimove Connect 2025 session with a myth. In the story of the Tower of Babel, a united human race speaks one language and starts building a tower to the sky. A wary God scatters them and confounds their speech, and humanity has been stuck with many tongues ever since. 

“God did not think of us marketers. A single language would work best for us.” 

Yinon Glasner, MyHeritage 

For most brands, that's a punchline. For MyHeritage, it's the central operating challenge. 

The Challenge: One Brand, 25+ Languages, and a Workload That Didn't Scale

MyHeritage is global unlike few consumer brands. Where competitors run five, fifteen, even eighteen regional brands, MyHeritage runs one. Its website's language picker offers 40 languages, and its campaigns go out in 25 or more. Roughly half of its users, including paying subscribers, are not English speakers. 

That global footprint isn't incidental to the product; it's the engine of it. A user in the U.S. or Canada very likely has relatives in Western or Eastern Europe. MyHeritage's presence in those territories is exactly what lets them find each other. Every market the brand reaches strengthens the network effect that makes the product work. 

Which makes the marketing problem unavoidable: speaking well only in English means quietly switching off half the audience. 

Before Optimove, the team ran its multilingual program on in-house, heavily code-based tooling: custom templating and triggering logic, a translation-management system called Crowdin. Plus, it had a user database that stored each user's language. It worked, but building an email almost took a developer’s skill. Maintaining it all was, in Glasner's words, a disaster. 

As the team moved to Optimove, the old way of handling languages didn't scale. Every campaign had to be duplicated: a separate template for each language, a separate campaign for each marketing action, and a separate target group for every language inside it. A single promotion could mean up to 20 templates, and the team was running as many as 60 campaigns a day. 

“Close your eyes and imagine 20 templates for a single campaign, when you have 60 campaigns a day. Horrible.” 

Yinon Glasner, MyHeritage 

Then came the part nobody budgets for: maintenance. Once a team has built 20 versions of everything, it must keep all 20 in sync forever. Localization adds its own traps, too: some languages use grammatical gender, so a template isn't just per language; it's per language and gender. Except the gender field is often unknown because the brand doesn't always collect it at registration. 

Multiply campaigns by templates by target groups, and the system collapses under its own weight. 

“You get a huge bowl of nice spaghetti. I love spaghetti, but I can't work inside a bowl of it.” 

Yinon Glasner, MyHeritage 

As a result, the team narrowed the scope to English plus a couple of top markets, German and Swedish. And in most other markets, English only or nothing at all. The complexity wasn't just a workload problem. It was silently capping the brand's reach. 

The Insight: The Bottleneck Was Duplication, Not Translation

The fix wasn't going to come from translating faster. It was going to come from never having to duplicate a template, a campaign, or a target group for a language again. That reframe is what MyHeritage and Optimove designed. 

The Solution: Four Towers That Replaced Twenty Templates

The fix came from working with the platform, not just on it. MyHeritage and Optimove co-designed the architecture back and forth, with mutual design and mutual QA. Finally, one requirement was met: a single template per campaign that works across all 25 languages, populated dynamically based on each user's language. Key elements were as follows: 

• One multilingual target group. Instead of one target group per language, a single group contains speakers of every language at once.
• Language as a customer attribute. Each user's language is passed to Optimove in the daily batch and through real-time events, so the system always knows which language to render.
• A new kind of template. Templates no longer contain hardcoded, human-readable text. They contain keys, variables that call the right content from the translation-management system; the template looks strange to a human and perfectly clear to the machine.
• The integration plumbing. APIs and a live connection to Crowdin let each key pull the correct translation for each user at send time.
• A content key replaces the sentence. Crowdin holds the English, the Danish, the German, and the rest, and the right words assemble themselves per recipient.

“Hardcoded templates should go to the trash bin. They are history.” 

Yinon Glasner, MyHeritage 

To keep that manageable at scale, the team built the guardrails the work actually requires: previewing any template in any language where a translation exists, firing test sends across 15, 20, or 25 languages at once, and reading a per-language status dashboard, green for fully translated, or flagged when a language is only 50 percent done, 25 percent done, or brand new. 

“I can now build a template and send tests in 25 languages with a snap of a button.” 

Yinon Glasner, MyHeritage 

The result: a single 50%-off sale that once required 20 templates and a fistful of target groups now ships as one template to one multilingual group. 

“One campaign, one template, one target group: all languages covered in the same template. And it works.” 

Yinon Glasner, MyHeritage 

The Results: From a Hard Ceiling to Room to Grow

The payoff was, in Glasner's words, extreme efficiency. The contrast is the story. Before, the team could realistically serve English plus a few priority markets and leave everything else on the table. After, they went back through campaigns they had launched in English only, translated them, and reactivated them against target groups spanning many languages at once. 

Glasner was candid that this isn't the finish line; there is more to migrate and more to improve. But the ceiling moved. A program that had no path to scale now has one 

“This would never have worked unless we did what we did with Optimove. Our ability to scale was non-existent.” 

Yinon Glasner 

Director of User Engagement, MyHeritage 

Seven Lessons for Multilingual Marketing

Glasner distilled seven lessons that travel well beyond family history marketing: 

1. Separate content from design. Store copy in a translation management system and reference it from templates with keys, so one template serves every language; the moment text lives inside the template, you've signed up for endless duplication. 

2. Make language a data field, not a campaign. Pass each user's language as a customer attribute, in batch and in real time, so personalization is driven by data the platform reads at send time rather than a parallel campaign per locale. 

3. Collapse duplication into dynamics. One campaign, one template, one target group beats a pile of copies, and the bigger win is maintenance, not creation: anything you duplicate, you maintain forever. 

4. Build QA for scale before you need it. Preview in any language, multilingual test sends, and a per-language status view are what make 25 languages safe to ship; scale without visibility is just risk. 

5. Plan for localization's edge cases. Grammatical gender, missing or unknown attributes, and right-to-left scripts multiply complexity quietly, so design for them up front rather than in production. 

6. Count the hidden cost of not solving it. When multilingual execution is too hard, teams default to English and quietly abandon markets without ever deciding to; complexity doesn't just slow you down, it shrinks your addressable audience. 

7. Co-design with your platform. The breakthrough came from treating the vendor as a design partner and iterating together, since the hardest scaling problems tend to get solved jointly rather than configured alone. 

“It's not the endgame, but we're getting much closer. And now we can scale.” 

Yinon Glasner, MyHeritage