> ## Documentation Index > Fetch the complete documentation index at: https://docs.heralded.ai/llms.txt > Use this file to discover all available pages before exploring further. # Segments > Measure the same brand in another market, language or buyer's voice. A segment is a market, a language and a persona, read with its own prompts. Buyers in Germany asking in German see different answers from buyers in the United States asking in English, and a segment measures each on its own terms. Every brand starts with one segment, **Default**. Manage segments on your brand's **Segments** settings page. How many a brand can have depends on the plan, and Default counts as one; see [Plans, trials and limits](/concepts/usage#brands-and-segments). The comparison table shows one page at a time. Its sort applies to every segment, with Default first and unmeasured figures last. Choosing a sort returns you to the first page. ## What a segment sets * **Market**, the country the engines are asked from and where search volume comes from. Setup fills in the country it detects. Choose **Global ยท search volume from the United States** to measure without a country. * **Language**, the language the prompts are written in and the answers come back in. * **Persona**, who the engines are asked as. The general buyer is the baseline. Every segment is read on the six standard engines, and on [Claude](/concepts/engines#claude) if your workspace has the add-on. ## Keyword data by market and language You can choose any supported market with any supported language. Keyword research, including discovery and difficulty, depends on the published coverage for that pair. Search volume comes from Google Ads, which supports every language in segment setup. Romania has keyword research in Romanian, but not in English. Both get search volume from Google Ads. You can still track English prompts in Romania and measure their AI answers. When a pair has limited keyword data, a small info icon appears beside the language in segment setup before you save, beside the prompt count in onboarding, and beside the relevant heading or volume in Discover and the prompt table. Hover over or tap it to read the coverage. For English in Romania, it says "Keyword research not available for English in Romania. Search volume comes from Google Ads." A pair with full coverage has no icon. **Draft again** appears only for an unfinished draft where another try may help. Each search volume carries its measured market and language. Hover over a volume to see them. Global uses the United States for search volume, even when its prompts are in another language. ## Prompts per segment Each segment has its own prompts, and each one uses a [prompt slot](/concepts/usage#prompts). When you create a segment, Heralded rewrites the prompts it brings in from elsewhere once, in the persona's voice and the segment's language. Prompts you type are tracked as you typed them, and wording never changes after that. ## Segments and your figures Home, Competitors and Sources cover all your segments together. The **Segments** table and a segment's pane state each segment's score and figures over your latest weekly report, read from that segment's answers, with the [change](/methodology/change) every other page states. Once a brand has two segments, Prompts can show one segment or all of them, and Actions shows one segment at a time. A new segment adds its own [cells](/methodology/how-we-measure#prompts-engines-and-cells), and they sit out of [change](/methodology/change) until they have a report to compare with. Changing a segment's market withholds change until the window no longer spans it, three more reports on the default window, because the answers now come from somewhere else. ## Read segments through the API `GET /v1/brands/{brand_id}/segments` returns your segment configuration, prompt counts, personas and plan limit. Its `comparison` holds the complete table with each segment's figures over the last week. Use `sort=mentioned`, `recommended`, `prompts` or `change`; Default stays first and unmeasured figures stay last. Retired segments are excluded unless you set `include_retired=true`. `GET /v1/brands/{brand_id}/segments/{segment_id}` returns one segment's figures, trend, engine rows, prompt evidence and setup. Add `/engines/{engine_id}` for a selected engine's figures and prompt answers. A segment or engine outside this brand's selection returns 404. An engine row's `figure` and the engine screen's **Mentioned** and **Recommended** cards use the same figures as `/v1/brands/{brand_id}/scores?segment={label}&group_by=engine` over `1w`. Each figure includes its value, range and measurement state. Prompt counts describe the latest prompt evidence separately. See [Figures](/figures) for the populations and ranges. These reads return a scan pin in `meta.query.as_of`. Pass it as `as_of` to keep measured facts at that scan, including daily answers where available. Configuration and tracked prompt counts stay live. MCP `get_segments` returns the collection in `segments`; pass `segment_id` for `segment`, or both `segment_id` and `engine_id` for `engine`. [API keys](/api-keys) and [assistant access](/assistants) follow your access to the brand. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.