# 7 Ways To Get Your Business Mentioned In Claude

Based on approximately 95,000 search result records captured across 3,000 unique prompts in Claude from July 2026.

This Searchable study separates the parts of Claude visibility that brands can influence through live retrieval from the parts shaped by Claude's learned memory.

## Play 01: Sort your buyer questions into retrieval and memory prompts

Claude handles buyer questions in two different ways. Retrieval-led questions cause Claude to look for current sources; memory-led questions are often answered without reading a live page.

Observed retrieval frequency:

- "best CRM 2026": 9 times in 10
- "cheapest CRM": 7 times in 10
- "best CRM": 1 in 3

Observed retrieval frequency for memory-led questions:

- "Zendesk vs Intercom": 1 in 7
- "is Zendesk worth it": 1 in 11
- "how to pick a CRM": 1 in 34

Action: write down twenty buyer questions. Put freshness, price, and category prompts in the retrieval group. Put comparisons, worth-it, and how-to prompts in the memory group. Use the ratio as your content budget split.

## Play 02: Index your business in Brave Search, Claude's closest-matching index

The pages Claude read overlapped with 63% of Brave's top-ten results and 34% of Google's top-ten results for the same queries on the same day. This indicates index behaviour; Claude has not disclosed a search provider.

Action:

1. Search Brave for `site:yourdomain.com`.
2. If your pages do not appear, submit your homepage through Brave's URL submission process.
3. Confirm that robots.txt, firewall rules, and bot protection do not block crawlers.
4. Keep sitemaps current and internal links crawlable.

## Play 03: Move your page-two rankings into Brave's top ten

Claude's source pool behaves like a gate. Results in positions 1-10 were read; position 11 and below were rarely read. Once a source was read, exact rank inside the top ten mattered much less:

- Ranks 1-3: 39.4% of read sources were cited
- Ranks 4-6: 37.2%
- Ranks 7-10: 36.8%

Action: prioritise queries where your pages rank 11-30. Moving a page into the top ten matters more for Claude visibility than polishing a result that is already inside the pool.

## Play 04: Earn coverage in national press and high-converting comparison sites

Share of read sources that appeared in Claude's final answer, compared with ChatGPT:

| Source type | Claude | ChatGPT |
| --- | ---: | ---: |
| National press | 52.0% | 19.2% |
| Review sites | 49.1% | 44.8% |
| Smaller comparison blogs | 42.6% | 28.8% |
| Directories | 30.3% | 29.3% |
| Reddit and forums | 8.3% | 12.7% |

Action: prioritise editorial press, review sites, and smaller comparison publishers. Avoid concentrating the plan on one large publisher: no single site supplied more than 9% of Claude's cited sources in this sample.

## Play 05: Rebalance your content toward third-party mentions

A brand's own site was used in 31.6% of Claude answers, compared with 60.9% of ChatGPT answers.

Action: keep publishing content that serves search, product education, and conversion, but put the next marginal investment into credible third-party editorial mentions. This is a weighting argument, not a reason to stop publishing on your own site.

### August 2026 note: Claude-generated content can carry machine-readable marks

Anthropic says Claude models launched on or after August 2, 2026 support marking at launch. Text from supported models carries an imperceptible embedded watermark, while supported SVG, PNG, and JPG files receive signed C2PA provenance metadata. The mark can travel with copied text and may persist through some editing.

Treat the mark as provenance context, not a quality or ranking signal. Anthropic says detection only indicates that Claude may have processed the material; it does not prove Claude was the original author. Models released before August 2 are still being updated for marking support.

Source: [How Claude marks AI-generated content](https://support.claude.com/en/articles/16266773-how-claude-marks-ai-generated-content), Anthropic Help Center, updated August 10, 2026.

## Play 06: Measure Claude's unprompted recall of your brand

Claude often supplies a shortlist even when the buyer names no brand. In this sample, 86% of brands Claude supplied in an unprompted shortlist appeared in the final answer.

Examples:

- "cheapest applicant tracking system for agencies" produced Manatal, Zoho Recruit, and Recruit CRM.
- "which ergonomic chair is worth it" produced Herman Miller Aeron and Steelcase Leap V2.
- "Zendesk pricing and alternatives" produced Freshdesk, Intercom, and Zoho Desk.

Action: write ten unbranded buyer questions, run each five times in a fresh Claude chat, and count your brand's mentions out of fifty. Repeat quarterly and compare the result with competitors.

## Play 07: Optimise every location's business profile for Claude

Local questions use a listings service rather than conventional web search. In testing, 97.7% of local prompts triggered retrieval, while Claude read only 0.1 conventional web pages per local query.

Action: audit opening hours, categories, descriptions, and location data for every physical location. Use explicit towns, cities, and postcodes when benchmarking; an unresolved location can produce misleading results.

## Study method

Approximately 95,000 search-result records captured across 3,000 unique prompts in Claude from July 2026. Search-provider behaviour is inferred from index overlap; Claude did not disclose a provider.

Source: [Searchable Claude visibility study](https://www.searchable.com/claude-visibility-report)