.png)
Webflow scored 2,000 websites on how ready they are for AI answer engines. The median came back at 2 out of 5. If you're starting from zero, that's good news: the fixes are ordinary, and they work fast.
On a recent 1Mind webinar, AEO That Converts and Delivers Results, Guy Yalif (Chief Evangelist at Webflow) and Rahul Jain (CEO of Noble) walked through what's actually stopping brands from showing up in AI answers. It was hosted by Jonathan Kvarfordt at 1Mind.
This is a guide for teams at the starting line. No AEO program yet, no baseline, no idea whether ChatGPT mentions you at all. That describes most companies, including plenty of large ones, and the work to fix it is more boring than the conversation around it suggests.
Four things to fix, in the order worth fixing them.
Webflow analyzed more than 2,000 US company websites, asked answer engines the questions a real buyer would ask, and scored each site from 1 to 5 on readiness. Three numbers from that study are worth carrying into the rest of this post.
Worth knowing why it's worth the effort at all. Guy shared that people arriving at Webflow's site from an LLM convert around 6x better than unbranded organic search. 1Mind's research across more than 500,000 human-to-AI conversations also found 27% of website visitors were already in active evaluation - both stated verbally during the webinar.
The problem: 62% of sites have broken internal links and 60% are missing basic SEO metadata.
Guy said the results surprised him. The blockers weren't new AI problems; they were old SEO problems nobody got to. His explanation for why: fixing them has always required an engineer and a marketer in a room at the same time, and that almost never happens.
Then there's schema, which is metadata telling a model what it's looking at. This is a product page. This is a question, and here is its answer. Guy's figure was that only 12% of websites use it, dropping to 2% in B2B.
Webflow tested this on six product pages. They added an FAQ block at the bottom and schema behind it, and changed nothing else. Within two weeks those six pages accounted for 57% of Webflow's incremental AI citations, out of a site with roughly a quarter of a million pages.
Do this:
Two shifts matter here. Freshness now counts in a way it didn't for search, and the unit of work is a question rather than a keyword. Buyers are asking full questions in full sentences, and your content either answers them or it doesn't.
Guy's shortcut for finding those questions: run an LLM over your Gong calls and your sales and support email. Most teams are guessing at what prospects ask while sitting on a recording of the answer.
One caution he was firm about, because a lot of current advice says the opposite. Don't chop your content into machine-shaped fragments. Models read pages written for people perfectly well. Write for the human who lands on your site, structure it clearly, and stop there.
Do this:
We have our own in-depth piece on how to generate and build your prompt list, so you’re actually measuring the right prompts - not just taking a best guess at what your customers are searching to find you.
The problem: 73% of companies are mentioned in less than a quarter of the places AI cites for their category.
This is the biggest gap in the study, and the one most teams have no plan for at all. Answer engines look for consensus. If several independent sources say the same thing about you, that reads as true. If nobody says anything about you, there's nothing to agree on.
There are two halves, and they need different work.
Onsite, Webflow found that backing your claims with data, sources and author credentials had the single strongest correlation with mention rate of anything they measured. And 55% of companies lack author bylines on 85% or more of the pages that should have them. Cheap to fix, highest leverage thing on your own domain.
Offsite is where the volume is. When AI mentions a brand, 85% of the sources it cites are third-party, not the brand's own site. Rahul's point on the webinar was that this is the hardest number for marketers to accept, because twenty years of SEO trained everyone to start and finish on property they own.
Two corrections he makes constantly, both of which save teams from wasted effort:
1. Domain rating isn't the filter. What predicts whether a source gets used is citation frequency: how often that specific article appears across the prompts you care about. Rahul routinely sees articles with a domain rating of 30 or 40 cited ahead of ones at 80 or 90.
2. Tier-one publishers usually aren't the cited ones. Many large publishers block AI crawlers outright, and the article that actually answers "best CRM for mid-market" is far more likely to sit on a niche industry site than in a national newspaper. PR relationships built over decades are often pointed at the wrong targets.
On volume, Noble's own research across 79 brands and 599 mentions over 18 weeks found that brands with fewer than about four live mentions rarely held their gains. Brands at four or more held a lift more than twice as often. One or two placements produce a bump that fades.
And presence is only half of it. When your pricing or positioning changes, updating your own site no longer finishes the job. The description of you sitting in a two-year-old roundup is still being read. Profound's analysis of WHOOP found 7.9% of AI claims about the brand were false or inaccurate across millions of daily queries.
Do this:
This is the work Noble does. Noble places brands in the third-party articles AI already cites for the prompts your buyers search but where you aren't mentioned, and corrects the mentions already out there so they reflect current positioning. Every placement is written to your brand guidelines and reviewed by you.
The problem: most teams are still trying to measure this without clear search metrics.
Rahul's advice for anyone starting out is to get a baseline before you change anything. If you come back and find you show up in 5% of the prompts you care about, that's not a disaster. It's a starting line you can measure from.
Track mention rate, citation rate, share of voice, and accuracy. Do this consistently for a month (your visibility will change dramatically week to week) and then start to implement changes. You can begin to optimize for certain blog types being cited on your website or decide to set aside more budget for brand mentions - this is the part where you can begin to actually create a strategy rooted in your own data.
If you don't have a measurement tool yet, we have a roundup of our favorite ones that make the shortlist here.
The closing question of the webinar was what you'd bet on with 90 days and no extra headcount.
Guy's answer, in order:
Rahul agreed on sequencing and added the part no optimization fixes. Product-market fit still matters. Happy customers still matter. There's no amount of AEO that compensates for a product nobody wants to write about.
What both would ignore: the instinct to treat this as a new discipline needing a new team. Guy was blunt about it. Your AEO team is your SEO team. Your AEO agency is your SEO agency. The overlap is heavy and the differences are additive.
The full conversation covers more than we fit here, including Guy's walkthrough of what schema looks like on a live page.
Watch the full 1Mind webinar with Guy Yalif and Rahul Jain:
Starting from zero and want to know which sources AI cites in your category? See where you're being cited.
About Noble
Noble helps brands show up accurately and often in AI search. Working with thousands of publishers, Noble places brands in the third-party articles AI cites, and updates the mentions already out there so they reflect current positioning. Pricing is success-based: brands set a budget and pay only for mentions that go live.
GET DISCOVERED ON
