From Guesswork to Growth: Using Data to Drive Your Global Strategy

Full transcript

Britt (host): I am so thrilled today to have Rebecca Grier joining us. Rebecca has a PhD from the University of Cincinnati in human factors and experimental psychology, and she leads our research here at Lokalise. She's been with Lokalise for a little over a year and a half. She's here today to present a little bit about what she knows — this won't be everything, since it's only covering one hour, but we're really excited to hear from her.

Rebecca: Thank you so much, Britt. I'm so excited to talk to everyone here today about From Guesswork to Growth: Using Data to Drive Your Global Strategy. On this first slide, we have a lot of companies you all know — Disney, Starbucks, Walmart, Coke, eBay. One thing they all have in common is that they tried to expand globally and stumbled in doing so. Today we're going to talk about how to make global expansion successful, since it's what we all want — it increases revenue and the number of people using our products. We'll talk first about why global expansions fail (spoiler: it's research — failure to do appropriate research ahead of time), then about Lokalise's approach to research, some concrete tips for conducting research, and I'll open it up for questions at the end.

Why global expansions fail: the eBay case study

There's a great resource called The Culture Map, a book by Erin Meyer, who now has a website where you can map countries on seven dimensions that are primary for cultural differences in business. Although eBay had successfully launched in Australia, Western Europe, and Canada by 2002, they took the same approach in China — buying a successful local site and rebranding it as eBay — and it didn't work. Today, eBay's efforts in China are really only about helping Chinese stores sell internationally.

The main reason was not understanding China as a marketplace. On the trust dimension, there's a huge disparity between how the US, Western Europe, and Australia work versus China. China is a relationship-based trust culture, while the US is a task-based trust culture: in the US, if we complete a task together successfully, I start trusting you. In China, people need a relationship and to communicate and know something about each other before trust is established. Around the time eBay launched in China, Alibaba launched a competing site called Taobao with instant messaging so buyers and sellers could communicate — which was very attractive because it allowed that relationship-based trust to build, something not possible on eBay's standard site.

The second major difference is communication style. China is a high-context communication culture; the US is low-context — more direct, saying what we mean. In China there are certain rituals and ways of saying things, with much more context and many more words used. This showed up in site design too: eBay kept a terse, just-the-facts style, while Taobao included every piece of information about a product in multiple different ways, which resonated more with high-context communication.

There were other cultural distinctions eBay missed. In China, a lot of purchases happen through haggling in small-shop marketplaces — you ask a price, get a counter-offer, and go back and forth. eBay's fixed-bid concept didn't allow for that; Taobao's instant messaging did, so it felt more familiar to Chinese shoppers. eBay also bought up online advertising, assuming competitors would advertise online, but not much of China was online yet in 2002 — they did the bulk of their own advertising on buses and taxis. Taobao instead advertised on television, which nearly everyone had access to and which was seen as more trustworthy than transportation ads, building much more brand awareness.

Lastly, in 2002 China still had strong superstitions about secondhand goods — the belief that if someone was selling something, misfortune had happened in their family and would transfer with the item. So buying secondhand was something done only out of desperation, unlike in the US where you might say "I got a deal" or "I'm being better for the environment." That core part of eBay's marketplace just didn't work in China at the time.

TAM, SAM, and SOM

So without cultural understanding, why did eBay enter the Chinese market at all? The same reason that draws a lot of companies in: at the time, with over a billion people, China had over five times the population of the US. That's the total addressable market (TAM) — it exists and can be used as a metric, but it's never fully attainable, especially without understanding the culture you're entering.

What you can actually address is the serviceable addressable market (SAM), or product-market fit: the number of people in a country who actually have a need your product can serve, given its current design (or design changes you're willing to make). Even that number is optimistic, because what matters most is your serviceable obtainable market (SOM) — the realistic share of the population you can capture given your resources, the competition, and your knowledge of the culture. The only way to identify the SAM and the SOM is by doing research to understand a country, its needs, its context of use, and how it views the world.

McDonald's: research as a competitive advantage

Given how hard this is, is anyone doing it well? McDonald's is doing it amazingly well. It's in almost every country in the world, and almost everyone knows a McDonald's — Big Mac, chicken nuggets, fries, and a Coke are on the menu everywhere, because McDonald's knows that consistency is its biggest asset. But McDonald's is also smart enough to know not everyone wants a Big Mac all the time, so it researches the countries it enters and continues researching to understand local tastes.

Examples: in Brazil, McDonald's sells pão de queijo (cheesy bread), a ubiquitous local food. In many European countries, McDonald's sells beer, which would be unheard of in the US given different cultural attitudes toward alcohol. In the Philippines, McDonald's serves McSpaghetti. McDonald's adapts its menu across the globe, and sometimes uses these local items to test recipes that might later travel elsewhere. User research is McDonald's secret sauce — it's well known not just because it serves the same core items everywhere, but because it adapts and has something for everyone, wherever they are.

How the brain processes information — and why it matters for research

Since research is so important to market success, especially global expansion, here's a bit about Lokalise's approach. Rebecca's research approach is rooted in cognition — if you've read Kahneman's Thinking Fast and Slow, this will be familiar. At a high level: we have a sensation system that picks up information from the real world and technology. Those sensations pass through a filter in long-term memory (Kahneman's "System 1"), which decides whether information should be ignored, acted on immediately, or sent on for more processing. If more processing is needed, it goes to working memory ("System 2"), where we perceive, comprehend, decide, and act.

This filter exists because everything processed in working memory requires cognitive energy. If every sensation had to be fully processed, we'd be drained constantly and unable to think abstractly or creatively. The filter doesn't just decide what to pass on — it bundles sensations into patterns and sends those patterns on, which reduces the energy our brains need. That filter uses pattern recognition, and those patterns come in three varieties: universal patterns (like Gestalt principles, or math — three plus three equals six, everywhere), cultural patterns (unique to certain regions — a TV tower recognizable to Berliners, hurling equipment recognizable in Ireland, football goalposts in the US, a lucky cat in many Asian cultures, or languages themselves), and individual patterns (highly personal, though "expert" patterns — like an air-traffic-control display, a chess position, a lab panel, or an advanced equation — are a good stand-in for how idiosyncratic this category can be).

These cultural patterns matter because they're a component of usability. In the 1990s, the International Standards Organization defined usability as a product's ability to help specified users accomplish specified goals with efficiency, effectiveness, and satisfaction in a specified context of use. A product cannot be usable for every user on the planet — it has to be usable for some subsection, and one of the key adaptations for broadening that subsection is culture.

The three phases of research

Most product development lifecycles go through three phases: understanding needs, exploring solutions, and delivering solutions while measuring impact. These map to three phases of research: formative (identifying user requirements and the problem space — what matters to this group of people and why), directive (once you understand needs, generating and testing multiple options against business goals, technical constraints, and human capabilities to make trade-off decisions), and evaluative (once you've designed and developed a solution, testing that it actually works for the people using it, and monitoring after delivery). Every research phase should be treated as iterative — you might learn something that sends you back earlier in the process.

At each phase there are questions to ask, mapped to the six components of usability. In the formative stage: what are users' goals, how are goals met today, what do users want to improve (not just what we assume), and how do users define effective, efficient, or satisfying? In the directive phase: how will a potential solution change what users do today, is that change a good value, and if so, how could it go wrong and how likely is that? In the evaluative phase: is the design usable, and once delivered, why are we meeting or missing our KPIs, and how can we expand them?

Research techniques and ethics

There are many techniques for collecting data across these phases — beta tests, surveys, analytics, behavioral research (usability testing, eye tracking, Wizard-of-Oz testing), interviews, field studies, and thought experiments like heuristic evaluations or cognitive walkthroughs. Different techniques suit different phases: interviews and field studies are great for the formative phase but less valuable later; surveys are great for the directive phase (understanding why KPIs aren't met) but not great early on, before you know enough to write good questions.

At Lokalise, the main way we get people to participate in research is the Lokalise Lab — a panel where people answer a few survey questions, and if their profile matches a research need, we reach out, often with a gift card as thanks.

Moving to tips for actually conducting research: the most important one is that research is all about planning — as Benjamin Franklin said, "If you fail to plan, you are planning to fail." Before the details of planning sessions, a note on ethics: throughout history, humans have been cruel to humans in the name of research, so there are laws and ethics worldwide — they vary, but five universals apply everywhere. Informed consent: people know ahead of time what they'll be asked to do, the risks, and how the data will be used. Voluntary participation: nobody can be pressured into research, and anyone can withdraw at any time without consequence. Respect for participants' time and energy, typically through compensation (a gift card, early software access, a report, or swag). Confidentiality: participants' names (and sometimes even the company) must be kept out of the research so people give open, honest feedback — breaking this trust can make it much harder to get valid information in future research. And it must be legal — laws like GDPR require that however you told participants you'd use the data is the only way you can use it.

Planning and running a research session

Most research sessions run about an hour — shorter is easier to recruit for, but harder to get meaningful information from; longer than an hour makes recruiting harder. Within that hour, only about forty minutes go to actual research objectives — five minutes at the start and five at the end are needed as buffer for lateness, video issues, talkative participants, or interruptions, plus time for informed consent, welcoming the participant, and getting permission to record, and time at the end for a note-taker's follow-up questions and for thanking the participant.

In that forty minutes, plan for about four questions (three is often even better). The first is a warm-up question — related to the topic, but not the core of what you want, since it takes participants a little time to get comfortable talking about themselves, to a stranger, and on a recording. The end-of-session buffer is crucial too: once you ask if they have any questions, or say the session is done, people relax — and that's often where the best insights, the real nuggets and quotes, come from.

Why only about ten minutes per question? Because in every session, the participant should be talking more than you are — talking yourself only repeats what you already know, while listening can teach you something new. Active listening means asking follow-ups: the favorite go-to is simply "anything else?", followed by five seconds of silence — people tend to fill silence, and repeating that question gets you to a point where they genuinely have nothing more to add. "Why" is another powerful follow-up, tied to the "five whys" technique from accident investigation: you often don't get to someone's real reason for needing or doing something until you've asked why several times (nobody buys a drill because they need a drill — they need a hole, and that hole serves some other goal, and so on).

The first, main question matters too. Keep it neutral and real — ask about things people have actually done, not hypotheticals or predictions about the future, and make clear any response is a good one (compare "wasn't Guardians of the Galaxy the best Marvel movie?" to the neutral "what do you think of Marvel movies?"). Keep it simple and clear — short, plain language, since customers aren't fluent in your product's internal terminology. And aim to surface assumptions rather than present them: ask why people did things, help them reflect on real experiences, and dig into why.

Subtle changes to questions change the quality of your data. Four weak questions people might ask when considering a new market: "Would you use this product?", "Is this pricing fair in your country?", "Do you usually work from home?", "Is this efficient?" Better versions surface the same information without an assumption baked in: "Tell me about a time you faced a problem related to this," "Tell me about the last time you paid for something similar — was that a fair price, and why did you pay it?", "Tell me where you work," and "Tell me what you'd want to change about this process, and why."

Summary and Q&A

In summary: global expansions fail mostly because companies assume TAM instead of thinking about SOM, and getting to SOM requires adapting to the local market — which requires understanding how cultural and individual patterns shape user experience. Lokalise's approach uses three iterative, interlocking phases — formative, directive, and evaluative — each with its own questions, and the research method is chosen to fit the goal. When actually running research: mind the laws and ethics of working with humans, practice active listening, keep questions neutral and real, keep them simple and clear, and make sure to surface assumptions.

Q: What KPIs do you think work best for showing localization impact?
It varies by what you're selling, but some measure of usability and completed sales — how many people started versus completed the sales process, whether there's a return or stopped usage, and understanding why. NPS is also worth watching if it increases.

Q: How do you approach estimating the SOM once you've done some user research in a potential new market?
Honestly, it depends — that's the researcher's favorite phrase. It comes down to thinking about how many people would have that problem, your competitors and their market share, what makes them successful, how similar your product is to theirs, and then combining that with your own resources for go-to-market and how successful you think those efforts would realistically be.

The session closed with a reminder that a short survey would follow, and that the next Lokalise webinar would be in September, after a break in August for the busy European holiday season.