Most of the audits we run are for small companies, the kind of brand a model has barely heard of. So we got curious about the other end. What happens when you point the exact same audit at names everyone knows? We picked five: Notion, HubSpot, Mailchimp, Webflow, and 1Password. Big brands, real budgets, sites that load clean. Then we ran our standard audit on each one and read what the engines actually said.
The short version is that being famous does not save you. Across the five, the engines left these brands out of nearly half the questions their own buyers ask. And the reason was not a broken robots.txt or a slow page. It was something quieter and harder to fix.
How we ran it
This was our normal audit, nothing special, run five times. Same method a paying customer gets.
- ✓Five brands: Notion, HubSpot, Mailchimp, Webflow, 1Password. One clear category each: workspace, CRM, email marketing, website builder, password manager.
- ✓Twelve buyer-intent prompts per brand, spread across the journey from broad awareness to head-to-head comparison, the questions a real buyer types.
- ✓Three web-connected engines per prompt: ChatGPT (GPT-4o), Claude (Sonnet 4.5), and Gemini (2.5 Flash), each grounded with live web search.
- ✓That is 36 answers per brand and 180 in total. For each answer we recorded whether the brand was named, in what position, which competitors showed up, and whether the engine cited the brand's own site.
- ✓One snapshot, taken in July 2026. Answers move over time and run to run, so read this as a probe, not a permanent scoreboard.
Finding one: household names, missing from half the answers
Here is the headline table. The score is our composite AI Visibility Score out of 100. The absent column counts how many of the 36 answers named the brand not at all. The last column is how often the engine cited the brand's own website when it did answer.
| Brand | AI Visibility Score | Absent from (of 36) | Own site cited |
|---|---|---|---|
| Mailchimp | 72 | 10 | 50% |
| 1Password | 59 | 14 | 17% |
| HubSpot | 59 | 13 | 3% |
| Webflow | 46 | 26 | 8% |
| Notion | 41 | 24 | 0% |
Add the absent column across all five and you get 87 missing appearances out of 180, which is 48%. Notion, a brand that spent years becoming a verb, was left out of two-thirds of the workspace questions we asked. Webflow was absent from 26 of 36. Even Mailchimp, the strongest of the group, was missing from more than a quarter of email-marketing answers.
We pulled each brand into its own teardown, with the specific prompts, the rivals that beat it, and the fixes the audit generated: Notion, HubSpot, Mailchimp, Webflow, and 1Password.
Being a household name gets you brand equity with humans. It does not automatically get you named in the answer a model gives a buyer who never types your name.
Finding two: the plumbing was fine. The citations were not.
The obvious guess is that these sites have a technical problem, a blocked crawler or a page that renders empty. They do not. We checked all five, and not one of them blocks AI crawlers in robots.txt. The plumbing is clean. That is the interesting part, because it means the gap is somewhere else.
Look at the last column of that table again. The engines almost never cited the brand's own domain. Notion sat at zero. HubSpot at three percent. When a model answered a question these companies should own, it reached for someone else's page to back up the claim. On the prompt for the best AI-powered workspace, ChatGPT named ClickUp, Coda, Wrike, and two tools most people have never heard of, and cited clickup.com and coda.io. Notion was nowhere in the answer, and notion.com was nowhere in the citations.
0%
Of Notion's 36 grounded answers cited notion.com. The engines answered workspace questions using competitors' pages instead.
This is the pattern we see over and over, now confirmed on brands with every advantage. Crawlability gets you in the door. Citations are what get you into the answer, and citations come from the rest of the web talking about you, not from anything you can toggle on your own site in an afternoon.
Finding three: smaller rivals out-showed the famous ones
For several of these brands, a competitor appeared in more answers than the brand itself did. Not a close call either. The rival was simply more present in the engines' responses.
| Brand | Answers it appeared in (of 36) | A rival the engines named more |
|---|---|---|
| Webflow | 10 | Wix (29 answers) |
| Notion | 12 | Asana (15) |
| 1Password | 22 | LastPass (25) |
| HubSpot | 23 | Salesforce (26) |
| Mailchimp | 26 | Constant Contact (13) |
Webflow showed up in 10 of 36 website-builder answers. Wix showed up in 29. If you only watched AI answers, you would conclude Wix is the default and Webflow is the challenger, which is close to the opposite of how the design world talks about them.
The same shape repeats. LastPass, a brand that has taken real reputational hits, out-appeared 1Password. Salesforce buried HubSpot in CRM answers. These are not verdicts on product quality. They are a read on which names the models have absorbed as the safe answer, and that ranking does not match the human one.
What the audit told each of them to fix
The fix list our audit generated was almost identical across all five, and the top item was the same every time: earn more inbound citations. Not add an llms.txt, not stuff keywords. Get the rest of the web, the comparison posts, the review sites, the communities buyers trust, to name you, because that is the raw material these engines read back.
It is worth sitting with the fact that this advice landed on Notion and HubSpot, companies with entire marketing departments. The citation layer is genuinely hard to move, which is exactly why it is where the visibility lives. If it were easy, everyone would already be winning it.
What this means if you are not a household name
If a brand as known as Notion can be absent from two-thirds of its own category's answers, your name recognition is not going to carry you either. That sounds grim, but it cuts the other way too. The thing separating you from the leader in AI answers is not their brand equity, which you cannot buy. It is citations and presence in the sources these models read, which you can earn, one comparison page and one honest community post at a time.
So the playbook does not change based on your size. Confirm the engines can reach you, which for most sites they already can. Then spend the real effort on becoming the name the rest of the web keeps mentioning, and check your work across all three engines rather than eyeballing one. The brands in this study have the budget and the plumbing. What they are missing is the part money does not directly buy.
One honest limit before you quote these numbers at a dinner party. This is five brands and 180 answers, taken on one July afternoon. Model answers drift over time and vary run to run, something we measured directly in our earlier study. A bigger sample would sharpen the exact percentages. It would not change the direction, which is hard to miss: these names are far less present in AI answers than their reputations would suggest, and the reason is citations, not code.