Simultaneous multilingual audit: 4 European markets in one audit

Picture this: your client sells in four European countries. France, the UK, Spain, Germany. Each market has its own competitors, its own keywords, its own AI search habits. To measure visibility in ChatGPT, Claude, Gemini and Perplexity across these four markets on a classic platform, you launch four separate audits. Four dashboards to compare manually. Four PDFs to merge. Four credit invoices. Half a day lost gluing things back together.

With the Agency+ plan, you launch one single audit. Four markets in parallel, one unified dashboard, one Premium PDF that aggregates the four languages in distinct sections. Here's how it works, and why it's a clean win for European agencies.

The problem: language fragmentation

Most GEO/AEO platforms treat each language as an independent audit. That's technically coherent (every AI model answers differently per language), but operationally it's a nightmare for anyone serving a European brand:

• Repeated configuration: you recreate prompts in each language, reselect models, reset native/web modes.
• Manual comparison: to know whether your client performs better in Germany than in Spain, you have to export each audit, paste into a spreadsheet, harmonise metrics.
• Scattered deliverables: four PDFs, four client portal links. The marketing director juggles four reports to figure out where to invest first.
• Cumulative costs: each audit has its fixed orchestration cost. Four audits cost more than four times one audit.

The solution: one audit, four markets, cross-market dashboard

The Agency+ plan activates the multi_language_audits feature. You configure the target languages (FR, EN, ES, DE — more soon) in the launch wizard, and the platform:

1. Localises prompts per market — not a simple translation. If your French prompts ask "which is the best French running shoes brand?", the German prompt isn't "welche ist die beste französische Marke für Laufschuhe?", it's "welche ist die beste deutsche Laufschuh-Marke?". The target brand stays the same, but the query matches what a local consumer would actually ask.
2. Queries models in each language in parallel — Celery routes requests to the right linguistic versions of models. No mix-up.
3. Scores every response via the AGS engine (five AI judges that calibrate against each other), with an inter-judge reliability coefficient published per market. See our AGS article for details.
4. Aggregates results in a cross-market dashboard with a colour-coded heatmap: row per model, column per language, green for strong presence, red for absence. You spot the language × model pair that needs attention in two seconds.

Four features exclusive to the Agency+ plan

1. Cross-market heatmap

A matrix view that crosses your languages with the AI models queried. If your brand is cited by ChatGPT in French but ignored by Gemini in German, you see it immediately. Aligned with the AI Share of Voice metric, segmented by market.

2. AI recommendations contextualised per market

For every language that underperforms on the 6-dimension GEO scoring, an Anthropic Claude Haiku model analyses the results for that specific market and produces a prioritised action plan. Not a generic "optimise your Schema.org". A concrete plan: "in German, your brand is cited but without context; here are the three pages to enrich in priority, with the FAQPage markup to add, and the E-E-A-T signals to reinforce for a German audience that values certifications."

3. Multi-section Premium PDF per language

One PDF document, with a dedicated section per language. Common cover, global summary on the first page, then four distinct chapters (one per market) with their scores, recommendations, detected competitors. The marketing director receives a single deliverable, defensible in front of a steering committee.

4. Contextual localisation, not translation

Prompts aren't translated word-for-word. They're rewritten per market, accounting for local context: national competitors, common expressions, dominant search intents. Our prompt_localizer service orchestrates this adaptation, validated by the six MCP tools dedicated to prompt management (Sprint B+).

Real case: a European luxury brand

A Paris agency supports a French fashion house distributing in 12 EU countries. Previously, it delivered one audit per strategic market (FR/EN/ES/DE) — four audits per quarter, around thirty minutes of configuration and thirty minutes of analysis each. Four hours of "invisible" billable work per cycle.

With the Agency+ plan and multilingual audit: one hour of unified configuration, the wizard shares prompts across markets (one default, adjustable per language), Celery orchestrates the four executions in parallel, and the unified PDF is generated in one go. Savings: three hours per cycle, or twelve hours per year. On an agency budget at €80 excl. VAT per hour, the return on investment on the Agency+ plan (€599 excl. VAT/month) is immediate.

More importantly: cross-market consistency. The marketing director sees immediately that the brand is strong in France and the UK, but underperforms in Germany on certain models. This information would be drowned in four separate dashboards.

How to enable multilingual

Three prerequisites:

1. Active Agency+ plan (€599 excl. VAT/month — see the pricing page for details).
2. A client profile filled with target markets (languages are inferred automatically, adjustable in the wizard).
3. Sufficient credit budget — multilingual audit consumes N times the credits of a single-language audit (where N = number of languages), but the precise estimate is shown at launch, adjustable before validation.

At audit launch, the wizard shows a language selector. Tick the ones you want (up to 4 simultaneously by default, more possible on request). Prompts are localised automatically, but you can edit them manually before execution (case by case, language by language).

During execution, the real-time dashboard shows progress per market. Once finished, you access the heatmap, the cards per language (dropdown filter), and the exportable CSV table. The Premium PDF generates in the background via Celery — typically available within 5-10 minutes.

Honesty: what multilingual does not do (yet)

• No response translation between languages: if your client wants to know what ChatGPT French answers translated into German, you'll have to do it manually. Multilingual measures visibility per language, not narrative consistency across languages.
• Limited to four official languages today: FR, EN, ES, DE. Italian, Portuguese, Dutch and Polish are on the roadmap (Sprint H). For other languages, contact the team.
• The human golden set for AGS calibration is in progress across the four languages — scoring quality is guaranteed but may be refined over the coming months.

Going further

• AGS: the multi-judge anti-drift scoring engine
• Agency mode: managing a team of GEO consultants
• AI Share of Voice — the cross-market metric
• E-E-A-T — authority signals per market
• Agency+ plan details and comparison with other plans

Article reviewed and updated: May 2026 — verified against current behaviors of ChatGPT, Claude, Gemini and Perplexity.