In startups, localization managers and their part-time replacements are often challenged by the topic of quality. The prevailing opinion seems to be that managing quality in a systematic way is reserved for bigger localization departments with larger budgets, more mature language service providers (LSPs) and sophisticated in-house tools. However, it doesn’t have to be this way. By starting early and following a strategy tailored to its current growth stage and localization maturity level, your startup can gradually master the difficult art of optimizing localization return on investment (ROI) while minimizing quality risks.
Collect data on localization outcomes
For an early-stage startup team just beginning to shape a half-baked idea into a product and push this product out to market, it’s only natural that getting the job done comes first, while quality takes second or third place. Most startup product teams at this stage can’t afford a dedicated localization manager, and many might not even have a single person on the team familiar with localization in practice. Localization, if it happens at all, is a part-time job for a product manager or a marketer. Processes tend to be nonexistent. Ad hoc decisions rule the day, no tools are in place, and no dedicated LSP might yet be chosen. CSA Research’s localization maturity model has a great term for this stage: chaotic localization.
If you are involved with a team like this as a consultant or LSP, the most important quality management activity you could do is get them to immediately start gathering data that’s cleanly split per country and language. You need:
■ Product metrics (AARRR: acquisition, activation, retention, revenue, referral).
■ User experience (UX) metrics (event-level analytics that can showcase usability issues).
■ Content performance (reach, engagement and conversion rates).
■ Local customer feedback (solicited via surveys or unsolicited via support channels).
Any localization investments in a for-profit organization almost exclusively serve to grow the company’s global business — but it’s also extremely important for an early-stage startup. Localization quality at this stage can and should be reframed in terms of business outcomes. If you managed to achieve the right level of quality in your localization for a particular country or language, you should see an uptick in your product, content and customer feedback key performance indicators (KPIs) for that localization. Benchmarking those metrics across locales, including your source locale, gives you insights about which of your languages might suffer from quality issues. However, often it’s tricky to explain why that might be happening. To design solid solutions, you’ll need to gather more data — and it might be difficult to do that at this stage.
By the time a startup has reached product/market fit and is starting to grow aggressively in several key markets, the amount of localization-related tasks quickly gets too big for a part-time nonprofessional to handle. A dedicated localization person might be hired at this time with a mandate to bring order to chaos. At roughly the same time, early in-country sales and marketing team members might get onboarded in top-priority geographies, and the startup finally gets a physical presence in different parts of the world. This stage often sees the first LSPs engaging with a startup, as well as the first in-country reviews, at least in some shape or form. The localization maturity model calls this repeatable localization, but in reality, this might easily require 6-18 months to achieve.
If you’re the newly hired localization manager or consultant for such a startup team, you’re likely to be overwhelmed with the many types of messes you’ll have to clean up. Pushing for internationalization from the start, centralizing all linguistic assets, deploying a translation management system, and onboarding a single LSP all take a great deal of energy, so managing quality systematically might seem like a far-flung proposition. However, by leveraging your extended team — your language partners and your in-country colleagues — you can actually do things that were unachievable in the previous maturity level. Here are a few tactics that won’t break your budget and do not require any extra manpower from your startup.
Your LSP should measure quality
You might not know it, but chances are good that if you’ve picked a sufficiently mature and reliable LSP, they already have an internal process to assess the quality of the translated content they deliver for your localization programs. In the best scenario, your language partner might even engage independent third-party evaluators so as to constantly keep their core translation teams in good shape.
If your vendor management strategy is centered around cherry-picking freelancers off a marketplace, you might be out of luck: it’s unlikely they can provide you any internal quality evaluation data, simply because they might not be doing much in this regard (and that is OK and expected). If that’s your case, this first step might not be directly applicable to you. However, the next two steps definitely are.
Initially, focus on understanding the existing quality evaluation processes on your language partner side and getting uninterrupted access to their quality data and related analytics. Once you have that, start working with your language partner to gently align their definition of quality toward your startup’s needs.
Being able to put some numbers on the table will help localization become a first-class citizen in data-driven management initiatives that might be starting now across your startup. It will also allow you to begin demystifying potential problems you might have discovered earlier when looking at outcome-level metrics per language and country. Could this conversion rate issue on our checkout page in Portuguese be caused, at least partially, by inadequate localization quality? You’ll now at a minimum have some data points to help you validate or invalidate such a hypothesis.
While on an aggressive growth trajectory, a startup tends to produce massive amounts of new content, including new product features, website materials, marketing collateral and customer support articles. Not all of them have the same impact on your company’s business. Furthermore, different geographies usually have a different expected return placed on them by your startups’ business leadership, so not all countries and languages are equally important.
Work with your business, product and marketing colleagues to jointly agree on clear priorities between combinations of different languages and different content types. Explain that in order to achieve your highest localization ROI, you will set up distinct localization quality levels in a way that reflects those priorities. Consider choosing between simple options like premium quality, standard quality and understandable quality. Discuss them with your language partner to learn how they will implement those quality levels in the actual production process.
Even if the only thing you actually do is achieve agreement internally and then clearly relay the expected quality levels to your language partner, this one-time effort alone is bound to have a huge impact on your localization quality. Freeing up scarce resources from low-impact localization to focus on high-impact localization can move the needle for your startup’s global business.
Having in-country marketing and sales professionals provide feedback on localized products and content is standard practice, and for good reason — these people have a unique insight into the local market and the company’s business goals that can be hard to replicate with a single-person localization team.
However, once a localization manager starts to bring in-country review into the picture, it quickly escalates into a tough fight. Typical battles might include countless hours of negotiation, lack of common understanding about goals, missed deadlines, confusion from the language partner and frustration for regional teams when their feedback is not properly taken into account. A lot of blood, sweat and tears routinely go into this process.
But rest assured that your efforts won’t be spent in vain. Lean on your LSP to get in-country teams to provide clear, if somewhat unstructured, feedback that significantly improves localization quality. Here are some tips to make this process more effective:
■ Focus their work on feedback that will tangibly affect business outcomes.
■ Limit their involvement to premium quality level content (at least initially).
■ Offer technology with a great UX that doesn’t require computer-aided translation expertise.
At this point, your growth-stage startup is likely to have been expanding aggressively for some time now, adding more countries, more languages and more content at a steady rate. You might have had the luxury to hire an extra person or two onto your localization team — or you might still be alone, but with a larger budget, core translation management technology in place and good working relationships with your language partner. How can you take your localization quality management game to the next level?
Consider bringing quality management in-house
While it felt great to heavily lean on your LSP’s shoulders in the previous growth phase, in many cases it’s difficult for their teams to live and breathe your product and market priorities to the same extent that you probably do. This has the risk to negatively impact quality during the growth stage, where your LSP’s key job is to support your aggressive scaling with their production and project management capabilities. At this point they might need a helping hand from your team to guide them toward the right quality expectations for the ever-growing amount of language pairs, content types and sophisticated quality levels.
Having the right technology starts to be an important benefit now, so push your language partner hard if all they offer is sending you a weekly Excel spreadsheet with a few charts and numbers. Ask for modern, standards-based, comparable quality metrics (like MQM-DQF), flexible quality profiles, real-time KPI dashboards, quick access to historical quality evaluations per content type and language, and the ability to deep-dive into individual quality issues if the need arises. If you can’t get your LSP to deploy a suitable solution, think about acquiring one yourself.
If your team structure allows it, consider making one of your team members a part-time quality management lead. At this point, it’s tempting to start running your own quality evaluations. However, hiring even part-time language quality professionals for two to three top languages might still be cost-prohibitive in a startup. If your LSP doesn’t offer third-party quality evaluation of their translations as part of their service package or if you are not satisfied with their quality management expertise, look for a marketplace that facilitates quick, on-demand access to localization quality evaluations done by language and domain experts or even turn-key quality management services offered by trusted providers.
Explore different approaches to measuring quality
By this point, you’re probably aware of an analytical approach to measuring localization quality. This is where a trained evaluator, usually a linguist:
■ Reviews translation for any potential issues that are important for a given content type.
■ Annotates each issue with a category. Is it a terminology issue or a style issue?
■ Assigns a severity level to it. Major mistakes usually happen on your product’s signup screen or checkout page, while preferential-level might be a typo in a footnote in your 350-page product manual.
■ Suggests an improved translation or a fix for the quality issue.
■ Ultimately arrives at a single score, usually numeric, that serves as a rough indicator of the overall quality of a localized piece of content.
Analytical quality evaluations are widely used and are helpful for many things: making quality scores more objective, taking a deep look into issue root causes and training linguists to avoid those issues in future. But they are also fairly expensive and could take a long time to run. If this is the only type of quality evaluation your language partner is offering, you might be limiting your ROI optimization options and hurting your global content delivery schedules. Consider introducing a holistic quality evaluation process that doesn’t pinpoint all quality issues in detail, but still gives a feel for the perceived quality, which is often what drives end-user experience the most.
A holistic evaluation can be as simple as rating every translated segment or even entire documents on a couple of core quality attributes, such as adequacy (how accurate the translation is versus the source) and fluency (how easy the translation is to read and how “native” it feels in the target language). Talk to your language partner to see if their tools and workflows are flexible enough to support this type of quality measurement and seamlessly combine it with analytical evaluations. This is essential in order to have a big-picture view of quality without breaking the bank and the schedule. You might even be lucky enough to involve some of your in-country colleagues in this type of quality evaluation if the user experience of the tool is nice enough.
Cherish in-country feedback
By now, you should have a somewhat workable process of in-country review that both your regional teams and your LSP should be moderately comfortable with, if somewhat frustrated.
Sometimes, savvy localization managers inadvertently take this frustration to a whole new level in their desire to push for quality measurement. Specifically, they try to task their in-country teams with something they have never done before: quantifying the localization quality analytically at the same time as they adapt localized content to specifics of their market. Asking a regional marketing manager to decide if her stylistic correction is preferential or critical is akin to asking a Spanish corrida bull if he wants to run toward the red flag: you know that both will push as hard as they physically can!
A similar situation might occur with in-country feedback coming through your customer support channels: while most support teams can quickly tell that a Japanese customer is complaining, they rarely have the capacity to pinpoint whether a specific localization quality issue might be behind this particular complaint.
In both cases, while actual feedback and improved translations come from the region, quality-related classification and analysis must be done by localization professionals. The solution is simple if your language partner or perhaps independent vendors can offer you the technology and staff that will enable you to:
■ Capture all in-country feedback, from in-country review and from support channels, in a single place, preferably next to localization quality metrics.
■ Review it regularly to see whether it uncovers any real localization quality problems; if it does, ask your language partner to categorize them as any other quality issue.
■ Remember to keep these issues coming from the market easily findable and treat them with utmost importance: make sure they are well-known to all involved linguists and other language partner teams, and take great care to avoid making those mistakes ever again.
Scaleup, not a startup
What used to be a bouncy, chaotic startup roller coaster might not feel like one anymore: growth starts to become less steep and things generally level out. A startup becomes a scaleup and finally earns the right to be called a predictable business, while hopefully still preserving at least some of the original spirit.
Your localization team might now have several people with clearly defined roles, and your localization scope and processes are likely stable as well. This is a good time to work on optimizing your ROI, and quality management provides a great avenue to help you with that.
By this point, the challenges to tackle get pretty specific so here are a few ideas to get you started:
■ Fully outsource your in-country review process to your language partner.
■ Combine localized software testing results with linguistic quality metrics.
■ Fully dedicate one of your team members to quality management.
■ Experiment with crowdsourced quality evaluation and cross-language usability tests.
■ Correlate product usage data with linguistic quality data to get to segment-level ROI.
Whatever you do, remember that your ultimate goal is to help your startup optimize its localization ROI while keeping quality risks at bay. Use the best methodologies, processes and tools that you have available. Play on the teams’ strengths, and always remember that quality is not a goal in itself, but rather a means to an end.