Content Localization via Personas

Generate the same content piece with different regional personas applied — each persona’s cultural context influences tone, idioms, and messaging for culturally-adapted content variants

Scenario

You have one core message — a product announcement, a campaign brief, or a landing page — and you need it to resonate in 4–6 different markets. Direct translation loses nuance. A U.S.-centric blog post doesn’t land the same way in Germany, Japan, or Brazil.

This playbook uses Mavera personas as cultural lenses. You generate the same content multiple times, each with a different regional persona applied. The persona doesn’t just translate — it adapts idioms, adjusts formality, reframes value propositions, and shifts cultural references so the output feels native to each market.

Mavera-only. No translation API, no localization platform. Personas carry the cultural context; Generate produces the adapted content.

---

Architecture

---

What You Need

Requirement	Details
Mavera API key	Starts with mvra_live_. Get one at Developer Settings.
Workspace ID	From your dashboard URL (ws_...).
Regional persona IDs	Pre-built or custom personas representing each target market. Use GET /personas to browse, or create custom ones.
Core content brief	The message, product, and key points you want adapted.
Brand voice ID (optional)	Apply a consistent brand voice across all regional variants.
Credits	~200–600 depending on region count and content type. See Credits Estimate.
Python 3.8+ or Node.js 18+	requests for Python; native fetch for Node.

MAVERA_API_KEY=mvra_live_your_key_here
MAVERA_WORKSPACE_ID=ws_your_workspace_id
BRAND_VOICE_ID=bv_optional_voice_id

---

The Flow

1. Define regional personas

   List the markets you want to target. Use pre-built personas or create custom ones with POST /personas using the NORTH_STAR or ADVANCED pipeline. Each persona encodes cultural context, communication preferences, and decision-making patterns.

2. Prepare the core brief

   Write your content brief once — topic, key points, target audience description, and desired format. This brief stays constant across all regions.

3. Generate with each persona

   For each region, call POST /generations with the same app_id, input_data, and brand_voice_id — but inject the regional persona context into the input. The persona influences the AI’s cultural framing.

4. Compare and review

   Collect all variants. Compare tone, idioms, formality level, and value proposition framing across regions.

---

Stage 1 — Create Regional Personas

If you already have persona IDs for your target regions, skip to Stage 2. Otherwise, create custom personas for each market.

Python

import os
import time
import json
import requests

API_KEY = os.environ["MAVERA_API_KEY"]
WORKSPACE_ID = os.environ["MAVERA_WORKSPACE_ID"]
BRAND_VOICE_ID = os.environ.get("BRAND_VOICE_ID")
BASE = "https://app.mavera.io/api/v1"
HEADERS = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json",
}

REGIONS = [
    {
        "name": "US Tech-Savvy Consumer",
        "description": "American millennial, early adopter, values convenience and ROI. "
                       "Casual communication style, appreciates humor and directness.",
        "region": "US",
    },
    {
        "name": "UK B2B Decision Maker",
        "description": "British senior manager at a mid-size company. Values understatement, "
                       "evidence-based claims, and dry wit. Prefers formal-but-not-stiff tone.",
        "region": "UK",
    },
    {
        "name": "DACH Enterprise Buyer",
        "description": "German-speaking procurement lead. Values precision, data, certifications, "
                       "and thoroughness. Direct communication, low tolerance for hype.",
        "region": "DACH",
    },
    {
        "name": "Japan Enterprise IT",
        "description": "Japanese IT director at a large corporation. Values harmony, consensus, "
                       "and long-term reliability. Formal tone, indirect communication style.",
        "region": "Japan",
    },
    {
        "name": "Brazil Digital-First Millennial",
        "description": "Brazilian young professional, digitally native, values community and "
                       "personal connection. Warm, enthusiastic tone with storytelling.",
        "region": "Brazil",
    },
    {
        "name": "India Startup Founder",
        "description": "Indian tech founder scaling a startup. Values rapid results, cost "
                       "efficiency, and global ambition. Mix of formal and energetic tone.",
        "region": "India",
    },
]

persona_ids = {}

for region in REGIONS:
    resp = requests.post(
        f"{BASE}/personas",
        headers=HEADERS,
        json={
            "pipeline_type": "NORTH_STAR",
            "name": region["name"],
            "description": region["description"],
            "workspace_id": WORKSPACE_ID,
        },
    )
    resp.raise_for_status()
    persona = resp.json()
    persona_ids[region["region"]] = persona["id"]
    print(f"  {region['region']}: {persona['id']} ({persona['name']})")

print(f"\nCreated {len(persona_ids)} regional personas")

Custom persona creation costs 300 credits each. If you’re localizing regularly, create personas once and reuse their IDs. Store them in your config or database.

---

Stage 2 — Define the Core Brief

The brief is the same for every region. The persona changes the cultural lens, not the facts.

Python

CORE_BRIEF = {
    "app_id": "blog_post_generator",
    "input_data": {
        "topic": "Introducing Workflow Automation for Growing Teams",
        "target_audience": "Operations leaders at 50-500 person companies",
        "length": "800 words",
        "key_points": [
            "Manual processes cost teams 15+ hours per week",
            "Our automation builder requires zero coding",
            "200+ integrations with tools you already use",
            "Customers see 60% time savings in the first month",
        ],
    },
}

---

Stage 3 — Generate Regional Variants

For each persona, inject the cultural context into the generation call. The persona ID tells the Chat layer how to frame the content; the brief provides the facts.

Python

from openai import OpenAI

mavera = OpenAI(
    api_key=API_KEY,
    base_url=BASE,
)


def generate_localized(region_name, persona_id, brief, brand_voice_id=None):
    """Generate content adapted for a specific regional persona."""
    localized_input = {
        **brief["input_data"],
        "target_audience": f"{brief['input_data']['target_audience']} — "
                           f"adapted for {region_name} market cultural context",
    }

    payload = {
        "app_id": brief["app_id"],
        "title": f"Localized — {region_name}",
        "input_data": localized_input,
        "workspace_id": WORKSPACE_ID,
    }
    if brand_voice_id:
        payload["brand_voice_id"] = brand_voice_id

    resp = requests.post(f"{BASE}/generations", headers=HEADERS, json=payload)
    resp.raise_for_status()
    gen = resp.json()

    if gen.get("status") in ("PENDING", "RUNNING"):
        gen = wait_for_generation(gen["id"])

    return gen


def wait_for_generation(gen_id, max_wait=300):
    for _ in range(max_wait // 10):
        resp = requests.get(f"{BASE}/generations/{gen_id}", headers=HEADERS)
        data = resp.json()
        if data.get("status") == "COMPLETED":
            return data
        time.sleep(10)
    raise TimeoutError(f"Generation {gen_id} timed out")


# Also score each variant with the persona for cultural fit
def score_cultural_fit(content, persona_id, region_name):
    """Use Chat with the regional persona to score cultural appropriateness."""
    resp = mavera.responses.create(
        model="mavera-1",
        input=[
            {
                "role": "user",
                "content": (
                    f"You are evaluating content for the {region_name} market. "
                    "Score cultural fit 1-10. Flag any idioms, references, or tone "
                    "choices that feel foreign or off-putting. Be specific.\n\n"
                    f"---\n{content[:3000]}\n---"
                ),
            },
        ],
        extra_body={"persona_id": persona_id},
    )
    return resp.output[0].content[0].text


variants = {}
total_credits = 0

for region, pid in persona_ids.items():
    print(f"\nGenerating for {region}...")
    gen = generate_localized(region, pid, CORE_BRIEF, BRAND_VOICE_ID)
    credits = gen.get("usage", {}).get("credits_used", 0)
    total_credits += credits

    fit_score = score_cultural_fit(gen.get("output", ""), pid, region)

    variants[region] = {
        "output": gen.get("output", ""),
        "credits": credits,
        "cultural_fit": fit_score,
    }
    print(f"  ✓ {credits} credits | {len(gen.get('output', ''))} chars")
    print(f"  Cultural fit: {fit_score[:200]}...")

print(f"\n{'='*50}")
print(f"Generated {len(variants)} regional variants | {total_credits} total credits")

---

Example Output Differences

The same “60% time savings” claim lands differently across regions:

Region	Adapted Framing
US	”Save 15 hours a week. That’s a whole Tuesday back.”
UK	”Our customers typically recover 60% of time previously spent on manual tasks.”
DACH	”Measured across 200+ implementations: average process time reduction of 60.3%.”
Japan	”Many teams have reported that after careful implementation, their workflows became significantly more efficient.”
Brazil	”Imagine having an extra day each week to focus on what really matters to your team.”
India	”At this price point, the 60% efficiency gain delivers ROI within the first billing cycle.”

---

Variations

Multiple content types per region

Run the same persona across multiple generation apps — blog, email, social — for a full regional content kit:

APPS = ["blog_post_generator", "email_sequence_generator", "social_post_generator"]
for region, pid in persona_ids.items():
    for app_id in APPS:
        brief = {**CORE_BRIEF, "app_id": app_id}
        gen = generate_localized(region, pid, brief, BRAND_VOICE_ID)

Focus Group validation per region

After generating, run a Focus Group with the same regional persona to validate the content resonates:

fg_resp = requests.post(f"{BASE}/focus-groups", headers=HEADERS, json={
    "name": f"Localization Check — {region}",
    "sample_size": 15,
    "persona_ids": [persona_id],
    "workspace_id": WORKSPACE_ID,
    "questions": [
        {"question": "How well does this content resonate with your cultural context? (1-10)", "type": "SCALE", "order": 1},
        {"question": "What feels foreign or off-putting?", "type": "OPEN_ENDED", "order": 2},
    ],
})

Language-specific generation

Add a language instruction to the input data for non-English markets:

localized_input["language"] = "German"
localized_input["language_note"] = "Write in German. Use formal Sie address."

A/B within a region

Generate two variants per region — one conservative, one bold — and let the Focus Group pick:

for tone in ["conservative and formal", "bold and conversational"]:
    brief_variant = {**CORE_BRIEF}
    brief_variant["input_data"]["tone"] = tone
    gen = generate_localized(f"{region} ({tone})", pid, brief_variant)

Pre-built persona shortcut

Skip custom persona creation by using Mavera’s pre-built generational and professional personas (GET /personas), then adding regional context in the generation prompt.

---

Credits Estimate

Operation	Typical Cost	Notes
Custom persona creation (×6)	1,800 credits	300 each; one-time cost — reuse IDs
Generation per region	15–30 credits	Depends on content length
Cultural fit scoring per region	1–5 credits	Chat call with persona
Total (6 regions, new personas)	~1,900–2,010 credits	First run with persona creation
Total (6 regions, existing personas)	~96–210 credits	Subsequent runs

Create personas once and store the IDs. After the initial persona investment, localization runs cost only ~100–200 credits per content piece across 6 regions.

---

What’s Next

Content Series Generation

Extend each regional variant into a multi-part series

A/B Copy Production

Test different voices within the same region

Message Testing Matrix

5 messages × 5 personas — quantitative message fit

Personas

Pre-built and custom persona reference

Content Generation

Full API reference for generation apps

Credits & Budget

Track and manage credit usage