- Is being cited by AI really enough to win a B2B customer?
- What the research says: how much time does your site really have to convince?
- What is cognitive-load testing and why does it matter for my site?
- Who are the buyers landing on your site and what are they really looking for?
- Why does a clear site also help AI systems cite you better?
- Three real-world scenarios: what changes for a B2B e-commerce store, a professional firm and a manufacturing company?
- Why is this being discussed now: is the Italian digital B2B market really at a turning point?
- How do you audit your site to reduce friction in the buyer journey?
- What mistakes are most common when optimizing a B2B site?
- How do I know if my site is really reducing friction for B2B buyers?
- So, is a cognitive audit of your site worth it for your B2B company?
- How we see it at SHM
- Frequently Asked Questions
- B2B buyers complete almost 70% of the purchase process before talking to a salesperson: your site has to convince on its own.
- Cognitive-load testing measures how much mental effort it costs visitors to understand, trust and act on your site — and it's the real bottleneck after AI visibility.
- A clear site helps both human buyers and AI systems understand your offer: less ambiguity, more accurate citations and more conversions.
- Reducing cognitive load doesn't mean simplifying your offer, but organizing information around how buyers assess risk.
71% of B2B buyers have already used generative AI tools to evaluate vendors, and in most cases they already have a shortlist before contacting anyone. Being found by AI is only the first step: if your site makes them work too hard, you lose them in 78 seconds.
Is being cited by AI really enough to win a B2B customer?
No. Visibility in AI engines is only the entrance: if your site doesn't turn that visit into trust and action, the buyer moves on to the next competitor.
Showing up on ChatGPT, Perplexity or Google AI Overviews has become the goal of many marketing teams. But there's a step that's often forgotten: after the AI has cited your brand, the buyer lands on your site. And there, on their own, they have to understand what you do, why you're credible and what to do next.
According to research by 6sense, buyers complete almost 70% of the purchase process before talking to a salesperson. 81% already have a preferred vendor when first contact happens. Your site is no longer a brochure: it's the real battleground of the decision.
In the SEO and AI visibility cluster we've already looked at how to get into the consideration set of AI systems. This article picks up from there and answers the next question: once the buyer arrives, what makes them stay — or flee?
| Your situation | The real problem | what to do |
|---|---|---|
| You're cited by AI but the traffic doesn't convert | Your site creates friction: the buyer can't find the answers they're looking for | Cognitive load audit + information architecture review |
| You have traffic but few sales leads | There's a single call to action ("Contact sales") that doesn't suit people still in the evaluation stage | Add intermediate steps: case studies, calculators, self-service demos |
| Buyers arrive but bounce within seconds | Confusing navigation or vague claims that don't answer specific questions | Restructure the content hierarchy around the buyer's questions |
| You're not cited by AI and you don't convert | Double problem: both visibility and site clarity need fixing | Start with site clarity: it helps both buyers and AI systems |
What the research says: how much time does your site really have to convince?
The average session on B2B sites lasts just 78 seconds, and it's shrinking. In that window, the buyer has to understand, trust and decide whether to stay.
The numbers from the original source on MarTech are clear: average session duration for B2B sites is 78 seconds according to Databox, and Contentsquare recorded a 7% year-over-year drop. Buyers increasingly decide on their own and spend less and less time on each site.
Only 35.7% of B2B buyers rate their experience on vendor sites as positive. That means nearly two-thirds of visitors find the site hard to use — and probably leave without contacting anyone.
What is cognitive-load testing and why does it matter for my site?
Cognitive-load testing — literally a "test of mental workload" — measures how much mental effort a visitor needs to understand what you offer, trust you and take action. It's an audit of your site from the buyer's point of view, not the designer's.
The idea is simple: every time a buyer has to stop and ask "What does this mean?" or "Where do I find this information?", they burn mental energy. The more energy it takes, the more likely they are to leave. Cognitive-load testing measures exactly these moments of friction.
It's not a generic usability test. It focuses on four specific areas that reflect how a B2B buyer evaluates a supplier.
- 1 · ComprehensionDoes the visitor instantly get what you do?Within a few seconds, it must be clear what you offer, who it's for and why it matters. Claims like 'leading solution' or 'AI-powered' answer none of these questions.
- 2 · OrientationDoes the structure guide or confuse?Too many menu items, overlapping categories or ambiguous navigation force buyers to piece together the map of your offering on their own.
- 3 · CredibilityCan your claims be verified?A careful buyer wants numbers, sources, methodologies and real cases. If the data isn't next to the claims it supports, doubt takes over.
- 4 · ActionIs there a next step for those who aren't ready to buy yet?If every path leads only to 'Contact sales', buyers still in the evaluation phase have no useful exit, and they leave.
Who are the buyers landing on your site and what are they really looking for?
Not all B2B visitors are the same. Some arrive to validate what they already know; others need to understand everything fast because they were brought in late in the process. Your site has to work for both.
Picture a procurement manager who has already read industry reports, AI-generated comparisons and customer reviews. When they land on your site, they don't want to know who you are: they want to confirm you're credible. They look for specific data, implementation details, security, measurable results. If all they find are slogans, they move on.
Then there's the stakeholder who joins the buying group at the last minute, maybe the CFO called in for the final presentation. They have to reconstruct the whole story in a few minutes. If your site is disorganized, that's a huge job. If it's clear, they find the path from problem to solution on their own.
| Visitor type | What they look for | What blocks them | What convinces them |
|---|---|---|---|
| Buyer who has already done research | Validation: data, sources, methodologies | Vague claims, numbers without sources, demo as the only CTA | Chain of proof: claim → metric → source → customer case |
| Stakeholder who joined late | Quick understanding of the full offering | Fragmented navigation, disconnected pages | Clear path: problem → solution → proof → next step |
| Technical evaluator | Product architecture, security, integration | Technical information buried or missing | Dedicated sections with specs and technical FAQs |
| Senior decision maker | ROI, risk, references | Product content only, no business case | Case studies with numbers, verifiable testimonials |
Why does a clear site also help AI systems cite you better?
AI systems don't 'feel' cognitive fatigue, but they do run into ambiguity. A confusing site produces imprecise or incomplete citations; a clear site gets extracted and summarized more accurately.
When an AI system like ChatGPT or Perplexity analyzes your site, it has to understand what you do, what your product is called and what evidence supports your claims. If you describe the same product three different ways on three different pages, the AI doesn't know which version is authoritative, and it may produce a wrong or incomplete summary.
The practices that reduce cognitive load for human buyers are the same ones that make content easier for AI systems to read. As we discussed in our article on how ChatGPT cites primary sources, content structure is decisive for being cited accurately.
GEO (Generative Engine Optimization, i.e. optimization for generative AI engines) and cognitive-load optimization aren't two separate strategies: they support each other. GEO helps AI find your content; a clear site helps AI, and the buyer, understand what it means.
| Practice | Benefit for the human buyer | Benefit for AI systems |
|---|---|---|
| Headings that answer real questions | Finds the information they need right away | AI extracts precise answers to user queries |
| Consistent names for products and features | Doesn't get confused between different versions | AI correctly associates the product with its features |
| Claims placed next to the proof | Verifies credibility without searching elsewhere | AI cites data with the correct source |
| Scannable text (not just images) | Quickly reads the key information | AI can index and cite the content |
| Logical internal links between related topics | Navigates without getting lost | AI understands the relationships between concepts |
| Structured data (schema markup) | Richer experience in search results | AI identifies organization, products, authors |
Three real-world scenarios: what changes for a B2B e-commerce store, a professional firm and a manufacturing company?
Cognitive load shows up differently depending on the industry, but the logic is always the same: every obstacle between the buyer and the answer they're looking for costs conversions.
Illustrative examples: numbers are meant to show the reasoning, they are not real customer data.
| Company type | Typical problem on the site | Friction it creates | Priority action |
|---|---|---|---|
| B2B e-commerce (e.g. component distributor) | Catalog with 10,000 SKUs and no filters by application or industry | The buyer can't find the right product and calls or leaves | Navigation by use case + upgraded internal search engine |
| Professional firm (e.g. tax consulting) | Services page with a generic list, no concrete cases | Prospects can't tell whether the firm has experience in their specific industry | Anonymized case studies by industry + FAQs for each type of problem |
| Manufacturing company (e.g., components) | Product pages with technical specs only, no application context | The technical buyer understands the product, but the senior decision maker doesn't see the business value | Two levels: technical sheet + management summary with ROI and use cases |
Why is this being discussed now: is the Italian digital B2B market really at a turning point?
Yes. B2B digital commerce in Italy reached €316 billion in 2025, and buyers increasingly use AI to evaluate suppliers before any human contact. Your website has become the first, and often the only, salesperson.
According to the Osservatorio B2b Digital Commerce of the Politecnico di Milano, Italian digital B2B reached €316 billion in 2025, up 14% from the previous year. It accounts for 20% of all B2B transactions in product supply chains.
At the European level, according to Netcomm, the B2B e-commerce market is already worth twice as much as B2C: over €1,600 billion versus about €840 billion. Growth is 8.9% compared to 2024.
Meanwhile, according to The Starr Conspiracy, 71% of B2B buyers used generative AI tools during the evaluation phase in 2025. Buyers who use AI are 2.3 times more likely to finalize their vendor shortlist before even contacting anyone. As we explored in our analysis of how to interpret your AI visibility score, making the shortlist is only half the job.
How do you audit your website to reduce friction in the buyer journey?
You start with a structured assessment of the four critical areas — understanding, orientation, credibility, action — and fix issues in order of impact on the buyer.
You don't need an expensive tool to get started. Three things are enough: a structured analysis sheet, a few observation sessions with real (or potential) buyers, and the behavior data you already have in Google Analytics or similar tools.
- 1 · Step 1Map the buyer profiles that land on your siteIdentify at least 2-3 real profiles: who they are, what they already know when they arrive, what they need to understand before acting. Talk to your sales team.
- 2 · Step 2Analyze your existing behavior dataLook at pages with a high bounce rate, exit paths and your most visited pages. Often the problem is already visible in the data.
- 3 · Step 3Test comprehension in 5 secondsShow your homepage for 5 seconds to someone who doesn't know your company. Then ask: what does this company do? Who is it for? If they can't answer, the problem is clear.
- 4 · Step 4Check your chain of proofFor every important claim, ask: is there data? Is there a source? Is there a customer case? If any of these is missing, the claim won't be credible to a discerning buyer.
- 5 · Step 5Review your calls to action for every maturity levelCount how many different CTAs you offer. If the only option is 'Contact sales', you're shutting out anyone still in the evaluation phase.
- 6 · Step 6Prioritize and act in phasesDon't rewrite everything at once. Start with the homepage and your most visited pages. Measure the impact before moving on.
What mistakes are most common when optimizing a B2B site?
The most common mistake is confusing simplification with cutting content. A B2B buyer needs detail: the problem is how that detail is organized, not how much of it there is.
A second frequent mistake is optimizing only for AI visibility without working on conversion. As the analysis of tools to monitor AI visibility also explains, measuring citations is useful, but it tells you nothing about what happens after the click.
| Common mistake | Why it is a problem | How to fix it |
|---|---|---|
| Generic claims ('industry leader', 'innovative') | They don't answer any real buyer question | Replace with specific, verifiable statements: numbers, sources, cases |
| A single CTA for all visitors | It excludes those who are still evaluating and aren't ready to buy | Add intermediate steps: guides, calculators, self-service demos |
| Navigation based on the company's internal structure | The buyer doesn't know what your sales department is called | Reorganize around the buyer's problem, not your org chart |
| Technical information only in PDFs or images | AI can't read it; buyers can't find it easily | Convert to structured HTML text with clear headings |
| Different names for the same product on different pages | It confuses buyers and creates ambiguity for AI systems | Adopt consistent naming across the whole site |
| Oversimplifying so as not to 'scare people off' | Expert buyers can't find the detail they're looking for and lose trust | Structure on two levels: a summary for management, detail for technical readers |
How do I know if my site is really reducing friction for B2B buyers?
There are clear indicators that show whether your site is working well or losing buyers along the way. You don't need complex tools: the data you already have is enough.
AI visibility is measured with dedicated tools, as we saw in the article on local visibility in AI search for SMEs. Site quality is measured with behavior and conversion metrics. Keeping them separate is a mistake: they should be read together.
| Indicator | What it measures | Positive signal | Red flag |
|---|---|---|---|
| Average session duration | How much time buyers spend on the site | Grows on key pages | Under 60 seconds on product pages |
| Bounce rate | How many visitors leave without interacting | Drops after clarity improvements | Above 70% on B2B landing pages |
| Pages per session | How many pages a buyer visits on average | Increases after the navigation restructure | Fewer than 2 pages per session |
| Conversion rate on intermediate CTAs | How many people download guides, use calculators, request demos | Grows after adding intermediate steps | Zero interactions with evaluation content |
| Quality of incoming leads | Are the contacts already informed? | The sales team receives more qualified leads | Leads don't know what the company does |
| Accurate AI citations | Does AI describe your offering correctly? | Citations reflect the naming used on your site | AI uses names or descriptions different from yours |
So, is a cognitive audit of your site worth it for your B2B company?
It's worth it if you have traffic but few conversions, your buyers arrive already informed and are looking for validation, or your sales team receives poorly qualified leads. In these cases, friction on the site is almost certainly the bottleneck.
A middle ground: you're still building AI visibility and traffic is low. Work on both in parallel: a clear site helps both buyers and AI systems understand your offering, so it's not wasted time.
Not yet if your sales cycle is handled entirely through direct relationships and your site is just a business card. But be careful: with 71% of buyers using AI to evaluate vendors, this scenario is becoming increasingly rare.
How we see it at SHM
For us, clarity isn't an aesthetic detail: it's the highest form of digital expertise. A site that makes buyers work hard isn't a design problem — it's a business problem. And it's often the problem nobody sees, because everyone is only looking at visibility metrics.
In the projects we work on, we often see companies that have invested well in SEO and AI presence, but lose buyers in the next phase because the site doesn't answer the right questions at the right time. The cognitive audit is the starting point: not to simplify your offering, but to organize it around how real buyers make decisions.
If you want to understand where your site loses buyers, we can start with a structured analysis of your information architecture and user experience. Our web design and SEO for AI services work on exactly this: visibility and conversion as a single system, not two separate projects. If you'd rather start from the inside, our AI training for companies program helps teams understand how AI systems read and evaluate your site's content.
Frequently Asked Questions
Cognitive-load testing is a structured assessment of how much mental effort a visitor needs to understand what you offer, trust you and take action on your site. In practice, it combines observation sessions with real buyers, analysis of behavior data (bounce rate, exit paths, session duration) and quick comprehension tests like the '5-second test'. It doesn't require expensive tools: you need a method and the willingness to look at the site from the buyer's point of view, not the company's.
No, and this is the most common misconception. An experienced B2B buyer needs detail: technical specifications, methodologies, results data. The problem isn't the amount of information, but how it's organized. The goal is to structure the detail around how buyers assess risk, not to remove it. Often you work on two levels: a summary for the senior decision maker and the technical detail for whoever has to evaluate the implementation.
Far more than you'd think. AI systems don't "feel" cognitive strain, but they do run into ambiguity: if you describe the same product in three different ways on three different pages, the AI can't tell which version is authoritative and may produce inaccurate or incomplete citations. The same practices that help human buyers (clear headings, consistent names, claims placed next to proof, scannable text) also make your content easier for AI systems to extract and cite correctly.
The signals are probably already in your data: high bounce rates on product pages, short sessions (under 60-90 seconds), few pages viewed per session, and almost no interaction with evaluation content like guides or demos. Another indirect signal is lead quality: if your sales team gets contacts who don't know what your company does, your website isn't doing its pre-qualification job.
The most effective starting point is your homepage and your most visited product or service pages. Begin with the "5-second test": show the page to someone who doesn't know your company and ask what it does, who it's for and what they should do next. If they can't answer, you've found your first area to fix. Then check your chain of proof: does every important claim have data, a source and a real case study next to it?
It depends on the scale of the changes and your traffic volume. Updates to the homepage and key pages can show effects on behavior metrics (bounce rate, session duration) within 2-4 weeks. The impact on conversions and lead quality typically takes 1-3 months, the time new visitors need to complete their evaluation cycle. For long B2B sales cycles, plan your measurement over a 3-6 month horizon.
Being found by AI gets you through the door: the game is won or lost in the next 78 seconds, on your website.
Related articles
Discover more articles exploring similar topics, selected to offer you a more complete and stimulating perspective. Each piece of content is carefully chosen to enrich your experience.
