Most global audience documents look like this. Ages 28 to 35, lives in a city, values quality of life, enjoys trying new things. The persona fits any country, which is exactly why it guides neither content nor media. The team writes copy from it and produces the same brand monologue it always did.
The problem is not the effort. It is the unit of research. An audience is not a demographic slice of a country. It is a group of people showing up in a specific market with a specific job to do. The same product looks different to a procurement manager in Germany and an owner-operator in Nigeria. The same procurement manager also shows up as two different people on LinkedIn and on short-video platforms.
The parent guide, Global social media strategy, treats audience as one of four inputs and spends one section on identity. This article expands that section and does one thing, turning identity into a matrix a team can fill in and test. The four inputs and the one-page canvas stay in the parent guide. If the wider social landscape is new, start with the global social media guide.
Why One Persona Fails
Take a concrete case. An industrial parts brand enters two markets, Germany and Nigeria. The team writes one overseas persona, small and mid-sized manufacturer owner, focused on cost and efficiency. In Germany, engineers, procurement managers, and IT security leads all take part in the purchase, and they care about different things. In Nigeria, many small factories have one owner who decides alone, and he cares more about downtime and whether spare parts arrive the same day.
The same persona is useless in both places. Germany needs specifications, certifications, and compliance documents. Nigeria needs delivery speed, payment terms, and local service. The persona contains none of that because it describes people, not jobs.
Audience research should produce four things, who takes part in the decision, what job each person needs to complete, where they use social platforms, and what content moves them one step forward. Demographics can be background, not the conclusion.
B2B, Draw the Decision Chain
B2B audiences are usually a chain. Each role judges risk in its own way, and content has to meet each one.
| Role | What they care about | Where they show up | Content they need | Evidence source |
|---|---|---|---|---|
| Initiator | Whether the problem is solvable and worth a project | Searching symptoms, asking in communities, watching short tutorials | Problem diagnosis, similar cases, early options | Support tickets, first sales call notes |
| Influencer | Specifications, compatibility, technical risk | Industry communities, deep reviews, LinkedIn | Spec comparisons, test data, integration notes | Pre-sales technical questions, solution review notes |
| Decision maker | Budget, return, supply stability | Peer referrals, industry reports, LinkedIn | Business cases, ROI models, customer references | Win and loss interviews |
| User | Ease of use, daily maintenance | Tutorials, user communities, short video | How-to guides, FAQs, maintenance checklists | Training questions, support logs |
| Procurement and compliance | Payment terms, contracts, data and security | Website documents, email, vendor questionnaires | Compliance notes, SLAs, security whitepapers | RFP question lists, legal threads |
The most common misuse of this table is producing one content set per role. Most teams lack the capacity, and they do not need it. A more practical structure maps one core asset to each buying stage, a diagnostic guide for awareness, a specification comparison for evaluation, a case and pricing note for decision, and an onboarding checklist for delivery. The roles remind you to answer different people's questions inside the same asset.
In small companies, the owner may be initiator, decision maker, and user at once. Do not force five roles. Write down the two roles that most affect the deal and merge the rest. The test is who actually blocks deals in the sales process, not the org chart.
B2B content also differs by who will share it. Engineers share technical reviews; procurement usually does not. Owners share peer cases; engineers find them too general. Add a column for who is willing to share, because it decides whether content travels.
B2C, Draw Motives and Scenarios
B2C differences sit mostly in motivation. The same person may buy a coffee machine to save time every morning, to host friends, or to stock up during a sale as a gift. The motivation changes the content and the landing page.
| Motive | Trigger | Platform behavior | Content needed | Evidence source |
|---|---|---|---|---|
| Solve a concrete problem | Current solution fails or creates friction | Search symptoms, watch tutorials, ask communities | Problem explanation, before and after, steps | Support and after-sales logs, comment questions |
| Replace or upgrade | Old product ages out, new model appears | Compare prices, watch reviews, save options | Comparison tables, compatibility, trade-in | Site search, add-to-cart without purchase |
| Identity and aesthetics | Wants to express a lifestyle or furnish a space | Look for inspiration, save, share | Scene images, pairings, real user content | Saves and shares, UGC |
| Gifting and social | Holidays, birthdays, weddings | Search gift lists, read reviews | Gift sets, budget ranges, delivery times | Seasonal queries, support questions |
| Price and promotion | Major sale, discount seen | Compare, collect coupons, watch live | Price explanation, bundles, price protection | Ad clicks, coupon usage |
| After-sales and repeat | Product issue, consumable runs out | DMs, communities, search for parts | Troubleshooting, parts, care | DM tags, repeat purchase records |
The value of this table is that it replaces a group of people with a job. The team does not need to guess what a 28-year-old likes. It needs to answer what the user wants to finish at this moment, and what content helps them finish it.
One product can map to several motives. Do not trim the table to one for neatness. Put the two strongest motives in the matrix and leave the rest as watch items. Decide which is strongest from where current orders and inquiries come from, not from intuition.
The Market, Audience, Platform Matrix
Put the roles and motives into one table, split by market. Here is a template.
| Market and language | Audience | Buying job | Platform scenario | Content needed | Language and culture notes | Evidence grade | Hypothesis to test |
|---|---|---|---|---|---|---|---|
| Germany, German | Industrial parts procurement | Find a certified supplier with stable lead times | Learning and comparison, industry forums, supplier sites, LinkedIn | Certification list, lead-time notes, spec comparison | Formal address, units and standards, data protection notes | B, from 3 pre-sales conversations | Whether certification or lead time matters more |
| Indonesia, Bahasa | First-time buyer of small home appliances | Buy within budget with service coverage | Inspiration and comparison, short video, marketplace reviews, DMs | Scenario demos, price ranges, warranty and service points | Local language, installments and cash on delivery, holiday timing | C, platform comments only | Whether service coverage affects purchase |
Every cell has a rule.
Split market and language. The same audience in Germany and France may differ in language, platform mix, and payment habits. Do not merge them into Europe.
Write the audience as a role or a motive, not as age and gender. Use the five B2B roles or the six B2C motives, and pick the one that most affects the deal.
Write the buying job as one sentence. If you cannot, the audience is not researched yet.
Classify the platform scenario by intent, inspiration, learning, comparison, community, or support. You can add where you observed it, but do not let platform names become the classification.
Write content needs as what the user must see to move forward, not what the brand wants to publish.
Make language and culture notes concrete, forms of address, units, holidays, payment, trust signals. Do not write respect local culture.
Grade evidence A, B, or C. A means multiple first-party sources agree. B means one first-party source or stable platform observation. C means a third-party report or a handful of comments. C is only a hypothesis.
Write the hypothesis in one sentence that a content test or sales feedback can disprove.
Keep the number of rows executable. Three markets times two audience types is six rows, each with content and measurement, which is already a lot. Too many rows means nobody updates the table.
Three Evidence Layers
The most common mistake is treating a third-party report as a conclusion. A report can suggest a direction, but it cannot tell you what a German engineer will ask in a review meeting.
| Layer | Sources | What it answers | What it cannot answer | Collection habit |
|---|---|---|---|---|
| First-party | CRM, sales call notes, support tickets, site search, GSC queries, reviews | Real jobs, real objections, real wording | People you never reach | Export monthly, tag by role and motive |
| Platform | Comments, community threads, creator content, ad comments, saves and shares | Publicly expressed needs and emotions | The silent majority | Sample weekly, keep quotes and context |
| Third-party | Industry reports, platform insights, survey data | Market size and trend hypotheses | Why your specific customers buy | Use as hypothesis only, note sample and date |
First-party evidence is the most valuable because it carries the outcome of a deal. The three questions that stop a sales call are often the three missing pieces of content. High-frequency words in support tickets are often the terms that site search and ads should cover. GSC query reports show the sentences people use to find you, and those sentences can go straight into titles and FAQs. For turning queries into a keyword library, see the keyword research and search intent guide. For metric definitions and reporting, see the SEO data and tools guide.
Platform evidence needs protection from algorithm amplification. The loudest comment is not the majority, and a creator's content does not represent the average user. Record source, date, and engagement when sampling, and do not store only the conclusion.
Third-party evidence depends on sample and market. A US B2B survey does not transfer directly to Southeast Asia, and a 2024 report may not describe the 2026 platform landscape. Cite the source and date so the team knows how much weight to give it.
One practical rule. Every claim that enters the matrix needs at least two independent sources. A single source stays at grade C and goes into the hypothesis column.
Language and Culture Is Not Translation
In most matrices, language and culture gets one line that says needs localization, which is the same as writing nothing. It affects at least four things.
Word choice. The same category has different common terms in different markets. A literal product name may not be searched or understood. Build a market glossary with local customer wording, competitor terms, and support vocabulary, and write from it.
Tone and address. German distinguishes formal and informal address, Japanese has honorific levels, and Spanish varies by region. Getting it wrong is rarely illegal, but it signals that the brand does not understand the market. Write the tone into the matrix and have a local colleague or native editor confirm it.
Taboos and sensitivities. Religion, history, gender, body, and alcohol have different boundaries in different markets. Do not build a taboo list from memory. Have a local colleague review it and write confirmed boundaries into the content guidelines.
Holidays and spending rhythm. Ramadan, Diwali, Black Friday, Singles Day, and local back-to-school seasons all change attention and purchase timing. Plan content around the local calendar instead of translating headquarters' calendar.
None of this is solved by translation. Translation handles the literal meaning. Localization handles whether the sentence works in that market. The language and culture column is about the latter.
Platform Behavior by Intent
Platform names are not a classification. On the same platform, a user may look for inspiration, compare prices, or seek support. Classify by intent to decide format and measurement.
| Intent | What the user wants | Common formats | Signals to watch | Common misread |
|---|---|---|---|---|
| Inspiration | Discover something new, find scenes and pairings | Short video, scene images, collections | Saves, shares, return visits after saving | Watching views only |
| Learning | Understand a problem, learn a task | Tutorials, long video, step-by-step posts | Completion, question comments, time on page | Treating likes as learning |
| Comparison | Compare price, specs, and reviews | Comparison tables, reviews, offer details | Price questions in comments, DMs, add-to-cart | Using engagement to judge purchase intent |
| Community and belonging | Find peers, express identity | Community topics, UGC, live interaction | Comment quality, remixes | Broadcasting only |
| Support and after-sales | Solve a problem, check an order | DM replies, FAQs, tutorials | Response time, repeat questions | Treating support as a content burden |
In the matrix, the platform scenario can read learning, YouTube long reviews and industry forums, or comparison, marketplace reviews and short-video comparisons. The platform is where you observe. Intent is the classification.
The same content changes across intents. Inspiration can be short and visual. Learning needs steps and detail. Comparison needs numbers and conditions. Support needs accuracy and action. Copying one video into every scenario loses effectiveness. For amplifying organic reach, see the organic growth guide.
Validation, Two-Week Cycles
A completed matrix is still a set of hypotheses. Validate with small content tests and comment feedback, not one large research project.
Every two weeks, pick one audience cell and publish three to five pieces, changing one variable, such as three openings for the same topic or two kinds of proof for the same benefit. Watch four signals, qualified comments, DM inquiries, saves and shares, and new questions in comments. Views and likes are reference only.
Code the comments. Sort them into questions, objections, use cases, price feedback, and irrelevant, and review them weekly with sales and support. The category with the most questions and objections becomes the next content topic. This loop is more useful than a quarterly report because it sits closer to revenue.
Judge after two cycles. If a cell produces no qualified engagement, suspect the audience or platform scenario before blaming the content. Retest another cell or downgrade the hypothesis. If engagement is good but inquiries are few, check the path from content to landing page. If inquiries are many but deals are few, the problem may be pricing, delivery, or trust signals, not social.
Small samples give direction, not statistical significance. Three to five pieces over two weeks is enough to decide whether a direction deserves another cycle, not enough to prove a market works. Do not turn test results into industry conclusions.
Common Misconceptions
One overseas persona can cover every market. Markets, languages, and platform scenarios differ, so personas must split by market.
Demographics are enough. Age, city, and income do not guide content. Jobs and scenarios do.
Third-party reports are conclusions. Reports provide hypotheses. First-party evidence provides conclusions.
Platform names define the work. Platforms are locations. Intent decides format.
Translation equals localization. Literal accuracy is not local relevance. Tone, wording, and holidays need separate work.
Build the matrix once. Audiences and platforms change. Review quarterly and update when new evidence arrives.
Related Reading
- Global social media strategy and platform selection, the four inputs and the one-page canvas.
- Overseas market research guide, research methods and first-party study design.
- Social media content system, turning audience needs into pillars and topic libraries.
- Paid social advertising, connecting the audience matrix to targeting and tests.