The AI Visibility Index
We measured how visible 40 well-known brands are in AI answers by running each through the same live check: buyer-style questions put to Claude with web search, scored for whether the AI names the brand and how much share of voice it holds. The spread is wide, and fame does not guarantee visibility.
Free to cite with attribution and a link. Measured July 2026. Check your own brand free →
Key findings, July 2026
- The average AI visibility score across 40 well-known brands is 65/100, with a spread from 14 (Brooks) to 100 (On).
- Fame does not equal visibility. The widest same-category gap is On at 100 against Brooks at 14 in running shoes: a spread of 86 points on identical questions.
- 2 of the 40 brands score under 30, which means AI assistants rarely name them at all when buyers ask about their category.
Accounting software
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | QuickBooksquickbooks.intuit.com | 95/100 | 10/10 prompts |
| 2 | Xeroxero.com | 79/100 | 10/10 prompts |
| 3 | FreshBooksfreshbooks.com | 67/100 | 9/10 prompts |
| 4 | FreeAgentfreeagent.com | 62/100 | 8/10 prompts |
| 5 | Sagesage.com | 48/100 | 7/10 prompts |
CRM
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | HubSpothubspot.com | 88/100 | 9/10 prompts |
| 2 | Salesforcesalesforce.com | 70/100 | 9/10 prompts |
| 3 | Pipedrivepipedrive.com | 57/100 | 8/10 prompts |
| 4 | monday.commonday.com | 52/100 | 8/10 prompts |
| 5 | Zoho CRMzoho.com | 49/100 | 7/10 prompts |
Email marketing
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | Klaviyoklaviyo.com | 76/100 | 9/10 prompts |
| 2 | MailerLitemailerlite.com | 75/100 | 8/10 prompts |
| 3 | Brevobrevo.com | 62/100 | 8/9 prompts |
| 4 | Kitkit.com | 61/100 | 7/10 prompts |
| 5 | Mailchimpmailchimp.com | 56/100 | 8/10 prompts |
Project management
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | ClickUpclickup.com | 88/100 | 10/10 prompts |
| 2 | Notionnotion.com | 77/100 | 9/10 prompts |
| 3 | Asanaasana.com | 76/100 | 10/10 prompts |
| 4 | Linearlinear.app | 57/100 | 7/10 prompts |
| 5 | Basecampbasecamp.com | 19/100 | 3/10 prompts |
Website builders
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | Wixwix.com | 83/100 | 10/10 prompts |
| 2 | Squarespacesquarespace.com | 83/100 | 9/10 prompts |
| 3 | Framerframer.com | 70/100 | 9/10 prompts |
| 4 | Webflowwebflow.com | 68/100 | 9/10 prompts |
| 5 | WordPress.comwordpress.com | 30/100 | 5/10 prompts |
UK digital banking
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | Revolutrevolut.com | 89/100 | 10/10 prompts |
| 2 | Wisewise.com | 79/100 | 9/10 prompts |
| 3 | Tidetide.co | 67/100 | 8/10 prompts |
| 4 | Monzomonzo.com | 64/100 | 8/10 prompts |
| 5 | Starling Bankstarlingbank.com | 42/100 | 6/10 prompts |
Skincare
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | CeraVecerave.com | 96/100 | 10/10 prompts |
| 2 | La Roche-Posaylaroche-posay.co.uk | 85/100 | 10/10 prompts |
| 3 | The Ordinarytheordinary.com | 57/100 | 7/10 prompts |
| 4 | Glossierglossier.com | 36/100 | 4/10 prompts |
| 5 | Paula's Choicepaulaschoice.com | 32/100 | 4/10 prompts |
Running shoes
| # | Brand | AI visibility score | Named or retrieved in |
|---|---|---|---|
| 1 | Onon.com | 100/100 | 10/10 prompts |
| 2 | Asicsasics.com | 63/100 | 8/10 prompts |
| 3 | Nikenike.com | 58/100 | 8/10 prompts |
| 4 | Hokahoka.com | 54/100 | 8/10 prompts |
| 5 | Brooksbrooksrunning.com | 14/100 | 3/10 prompts |
Sector editions
Deeper cuts of the index, measured with the same method: 20 brands per sector.
Methodology, honestly stated
Every brand was measured in the same week with Discoverable's standard snapshot check: a generated set of buyer-style prompts for its category, answered by Claude with live web search, scored for mentions, sources and share of voice against the competitors the AI itself named. Scores are 0 to 100 and comparable within this edition of the index only.
What “sources” counted as, in this edition: a source counted when an engine retrieved it while answering, whether or not the answer went on to attribute it, and the “named or retrieved in” column counts prompts where the brand was named in the answer or one of its pages was retrieved. From 8 August 2026 our live product counts only what the answer itself attributed, and from 5 August 2026 engines no longer see the brand name while answering. A score you run today is therefore a different quantity from the numbers in this table rather than a higher or lower one. The next edition will be regenerated under the current method and labelled as such.
Limits worth knowing: this edition measures one engine (Claude, web-grounded), snapshot depth, at a single point in time. AI answers are probabilistic, so individual scores carry noise; treat gaps of a few points as ties. The index is a measurement of visibility in AI answers, not a judgment of product quality, and inclusion is not an endorsement. Brands listed can contest a figure at hello@bediscoverable.ai; we correct it in the next full regeneration rather than re-measuring one row into this table, because a row measured under today's method is a different quantity from the rest of the edition and every finding above is computed across all of them.
The pattern behind the numbers matches our study on how often AI gets brands wrong: assistants reward brands whose third-party footprint is consistent and current, not merely famous ones. How that works is covered in how AI assistants choose brands to recommend.
How to cite this index
Free to use in articles, decks and research with attribution and a link. Suggested citation:
Discoverable AI Visibility Index, July 2026 edition. Be Discoverable Ltd. bediscoverable.ai/ai-visibility-index
Need a custom cut of the data for your audience, or a quote on why the gaps exist? Email hello@bediscoverable.ai and we will turn it around fast.
About the index
How is the AI Visibility Index measured?
Each brand is run through Discoverable's standard snapshot check: a set of buyer-style questions for its category is put to Claude with live web search, and we score whether the brand is named or cited in the answers, how prominently, and what share of voice it holds against the competitors the AI itself names. Every brand in the index is measured the same way in the same week. This edition was measured in July 2026, under the methodology in force then. Our live product has changed twice since: from 5 August 2026 engines no longer see the brand name while answering, and from 8 August 2026 a source counts only when the answer itself attributed it rather than when the engine merely retrieved it, with engines given a higher search ceiling at the same time. A score you run today is therefore a different quantity from the numbers in this table rather than a higher or lower one, and the two should not be compared in either direction. The next edition will be regenerated under the current method and labelled as such.
Why do household names sometimes score low?
AI answers are not brand-awareness surveys. An assistant answering 'best accounting software for a small business' builds a shortlist from the sources it retrieves, and a famous brand that is absent from those sources loses to a smaller rival that dominates them. Low scores for big names are the most common surprise in our data, and the reason this is worth measuring at all.
Is this an endorsement or ranking of product quality?
No. The index measures one thing only: how visible each brand is in AI-generated answers for its category at the time of measurement. It says nothing about whether the products are good. A brand can be excellent and nearly invisible to AI, which is precisely the problem Discoverable exists to fix.
My brand is listed and I think the data is wrong. Can you re-run it?
Tell us either way: email hello@bediscoverable.ai and we will check it against the next edition, which is regenerated in full under one method so every row stays comparable with every other. We do not re-measure single brands into a published table, because a row measured under today's method is a different quantity from the July numbers around it. You can run the current check on your own brand in about a minute with the free checker, no login needed, and that result is directly comparable with nothing in this table.
How often is the index updated?
We plan to re-measure periodically as engines and their sources change. Each edition states its measurement date, and scores are only comparable within the same edition.