AI visibility across markets and languages: one scorecard

A marketing lead of a brand that sells in several countries asks this before the first board report. Each regional team holds a number, and the board wants one. This guide shows why the same prompt answers differently per market, and how to add markets and languages. It then combines the markets into one scorecard that keeps the spread.

The example is our own project, in the 28 days to 30 October 2026. In that window it asked its 47 prompts in Ireland from 3 to 14 October, and Worldwide from 15 to 30 October. For the five numbers of the report itself and their order, read our guide on client reports. For how many prompts each market needs, read our guide on prompt counts.

Why does AI search visibility across regions and languages differ?

AI search visibility across regions and languages differs because each market is a separate run, with its own answers and its own cited pages. A market is the country a prompt runs in, or Worldwide. Each market keeps its own numbers, and no market borrows answers from another.

The cited sources of our own project differ between its two markets. In the 28 days to 30 October 2026, on 7 AI platforms, the Irish answers ran from 3 to 14 October. The Worldwide answers ran from 15 to 30 October. The two markets ran on different days, so time explains part of each difference. In that window, google.com was the sixth most cited domain in the Irish answers. It took 397 citations in 3,695 answers. In the Worldwide answers of the same window, google.com was not among the ten most cited domains. The domain hubspot.com took a place in that ten, with 463 citations in 5,258 answers. It was absent from the Irish ten.

A language is a second split. The same question in German, asked in Germany, is a second prompt, with its own answers and its own trend. The market sets the country the question comes from. The language is the wording itself. FullReach AI asks each prompt in the language you write it in, and translates nothing, as our methodology states.

How do you add markets and languages?

Write each prompt in the language of the buyer, and pick the market it runs from. On FullReach AI, the choice is 97 countries or Worldwide. Three rules apply:

1. Write the question as a local buyer asks it. Do not translate an English prompt.
2. Pick the markets in the Markets field of the Add a prompt dialog. For a prompt that runs already, use Markets in its row menu on All Prompts.
3. A prompt counts once for each market it runs in. Every plan lists this feature: any number of markets and languages.

Our guide on prompt counts gives the count per market and language, and the formula that multiplies them.

What does the same prompt return in two markets?

The same prompt named a monitored brand in a different share of answers in each market. The table reads six prompts of our project in the 28 days to 30 October 2026, on 7 AI platforms. Ireland answered from 3 to 14 October, and Worldwide answered from 15 to 30 October. The two markets ran on different days, so time explains part of each gap. Read the table as an example of the spread, not as a measurement of Ireland against the world.

In the same window, one prompt differs by 17 points between the markets, and two differ by 2 points or less.

The order of the rivals differed too. The Competitors page of our project, filtered to one market at a time, gives this order. In the Irish answers from 3 to 14 October 2026, Scrunch stood fourth of six monitored brands, at 16.0% visibility. AthenaHQ stood fifth, at 14.1%. In the Worldwide answers from 15 to 30 October 2026, AthenaHQ stood fourth, at 13.6%, ahead of Scrunch at 13.2%. In the same two runs, our own brand appeared in no Irish answer and in 14 Worldwide answers, 0.3%. A pooled number shows 0.2%, and it hides that one market gave every one of the 14.

How do you combine the markets into one scorecard?

Keep one row per market and language, add one pooled row, and print the spread beside the pooled number. Our methodology keeps results from different markets apart. The pooled row is a reading aid beside the market rows, not a merged market. It weights each market by its analyzed answers, so a market with more answers moves it more. In the 28 days to 30 October 2026, our project held 5,258 Worldwide answers and 3,695 Irish answers. The pooled row leaned 59 to 41 toward Worldwide.

Four rules apply to the scorecard:

1. State each market's base, the analyzed answers behind its number. Two markets with different answer counts look alike as percentages.
2. Print the spread as the lowest and the highest market, with their names. "0.2%, from 0.0% in Ireland to 0.3% Worldwide" says more than 0.2% alone.
3. Read each market's change on the month before with the method in our guide on client reports.
4. Keep the rank per market. A pooled rank hides a rival that is fourth in one market and fifth in another.

The failed attempts stay out of every base. In the 28 days to 30 October 2026, 3,558 Worldwide attempts and 1,139 Irish attempts returned no answer. A market whose platforms refuse more requests has fewer answers, not lower visibility.

What does the scorecard template hold?

The template is one table, with one row per market and language and one pooled row at the end. Our own project fills it below, for the 28 days to 30 October 2026, on 7 AI platforms. Ireland answered from 3 to 14 October, and Worldwide answered from 15 to 30 October.

The same 47 prompts ran in both markets, so the pooled row counts 47 prompts, not 94. Its 8,953 answers are the sum of the two market rows. The pooled row carries the spread in its visibility cell, so a reader sees both markets at once. The language column stays, because a second language in the same country is a second row.

A board reads the pooled row. A regional team reads its own row and opens the stored answers behind it. To fill the table each month without copying numbers, read each market by API, or ask the MCP server in plain words. Our guide on API and MCP shows three worked requests and the Looker Studio connector. Our guide on which AI search engines matter names the platforms to track first, by audience.

Each row fills only from the daily answers of that market. A market that you add later starts with no history. Its row stays empty until its first answer, and its trend line starts on that day.

What to do next

1. List your markets and languages. Count one prompt per market and language.
2. Add the markets to your prompts, each written in the buyer's language.
3. Build the scorecard with one row per market, and print the spread beside the pooled row.