Winning without anyone clicking: how to exist in the age of visit-free answers

London, Tuesday morning, 8:43am. Outside the window, along City Road EC1V, the thin Shoreditch drizzle soaks the red brick of old industrial warehouses turned startup incubators. There's that particular smell this part of the city does so well: wet tarmac, scorched takeaway coffee, and the electric heating just kicking in across the offices.

I've got the Mac open on a scratched wooden table. I'm looking for a strategic partner — a firm specialising in the design and development of innovative packaging solutions — someone capable of working across product lines destined for both the Campania market and the British one. Instead of opening Google and scrolling through the usual parade of paid ads and blue links, I type the query directly into Perplexity. I ask for a comparative analysis, rigorous, grounded in real export case studies from the sector.

The machine takes four seconds. Three clean paragraphs. Three names, three specialist firms, with an explanation of why their approach suits that kind of cross-border market. I read, assess, memorise the names, jot them down in my notebook. Then I close the tab.

Session over. I haven't clicked a single link. I haven't visited a single website.

As far as those three agencies' analytics systems are concerned, I don't exist. No session recorded in GA4. No heatmap. No pixel. And yet, in that moment, those three firms made it onto my mental shortlist. In a fortnight, when I pick up the phone or send an email, I'll contact them — and only them.

This isn't some futuristic edge case. It's happening billions of times a day.

The click is dead. Long live the click.

Let's be precise about this: zero-click search is any search session in which the user's informational or commercial intent is satisfied directly within the search engine or AI assistant interface, without the user navigating to an external website.

It didn't start yesterday. Google had already laid the groundwork back in 2014 with featured snippets and People Also Ask boxes. But that was an outpost. Today, it's the main territory.

Data from SparkToro — a 2024 analysis of global search samples — shows that over 60% of Google searches resolve without a click to an external site. On mobile, that figure tops 65%. With the rise of AI-native interfaces like Perplexity, ChatGPT Search, and Google AI Overviews, this is no longer a trend: it's the load-bearing structure of online search.

Three levels make up this landscape:

Traditional Featured Snippets and PAA boxes.

Google pulls a block of text from a page and displays it above the organic results. The user gets the straight answer — a price, a technical spec — and stops searching.

Google AI Overviews.

This is no longer simple extraction. Gemini synthesises in real time by cross-referencing multiple sources, inserts minimal grounding links, and produces a response so comprehensive it renders further reading unnecessary for most queries.

LLM-native interfaces.

ChatGPT Search, Claude, Perplexity. Here, a results page doesn't exist. There's only a conversation. The brand either makes it into the response text or disappears entirely.

The paradox that SMEs need to get their heads around is this: the click is dying as a discovery metric — what we used to call the top of funnel — but it's holding its value, and amplifying it, as a pure conversion metric.

We'll see a steady decline in visits to company websites. Anyone watching only the GA4 counter will see what looks like a disaster. The reality is different: the user who lands on your site after reading an AI response isn't a distracted visitor. They're someone who's already been through the pre-evaluation stage, knows your specs, knows your positioning, and arrives with high transactional intent.

The click no longer serves to introduce you. It serves to sign the contract.

What actually happens when ChatGPT cites you — or doesn't

A few weeks ago I ran a test with Berto. I met him at the café near the Caserta bypass, espresso on the counter, and showed him what happens when you query Perplexity about a business like his. I typed on his behalf: "best industrial packaging suppliers in the province of Caserta."

His name didn't come up. A competitor with half his years in the trade — did.

Berto looked at me with that expression I know well: head tilted slightly, hands resting on the counter. "Umbè, I've spent thousands of euros to be on the first page of Google, and now you're telling me people read the AI's answer and don't even bother checking who we are?"

The answer is blunt: yes, Berto. But the solution isn't to stop investing. It's to understand the rules of the new game.

LLMs don't choose which brands to cite based on algorithmic favouritism or pay-per-click advertising spend. They use RAG (Retrieval-Augmented Generation) systems that query web indices to anchor responses to real, verifiable data. The factors that determine selection are five:

1. Structural authority of third-party sources.

If your website is the only place on the web where you claim to be an expert, that data carries almost no weight for the AI. If the same claim can be traced across news outlets, industry blogs, technical forums — the weight rises exponentially.

2. Informational density and content granularity.

Pages with generic copy, stuffed with hollow adjectives and lacking real technical specifics, get passed over. The AI is looking for clear definitions, numbers, processes described step by step.

3. Consolidated topical authority.

The system analyses the vertical consistency of the entire site architecture. Anyone who writes about logistics one day and recipes the next loses credibility on both fronts.

4. Frequency of contextual mention.

How often does the brand name appear alongside a specific keyword in qualified digital discussions? Not automated spam — citations in semantically rich contexts.

5. Semantic structure of the code.

Proper headers, clean paragraphs, explicit answers placed directly beneath direct questions — all of this makes the algorithm's job easier.

A textbook case: HubSpot. Ask ChatGPT or Perplexity anything related to CRM, lead management, or inbound marketing. HubSpot appears almost every time. Their blog is structured so thoroughly in semantic terms that AI models use it as an internal encyclopaedia. The economic result has nothing to do with immediate clicks: it produces devastating brand recall. Thousands of decision-makers absorb the HubSpot name as a synonym for authority simply by seeing it appear in their AI assistant's responses — without having read a single one of their articles in full in the preceding weeks.

The brand nobody clicks on, but everyone knows

There's a dimension of digital presence I call passive visibility. It generates no immediate traffic. It builds brand equity progressively and relentlessly. It rests on four pillars.

1. Citation Authority — Being cited as a source

This kicks in when an LLM uses your content to validate a response, inserting the brand name as its source. You're no longer a result in a list: you're the official source validated by artificial intelligence.

Semrush applies this strategy systematically: it periodically releases massive reports containing raw data on online search. These studies aren't designed to sell software in the short term. They're built to be cited. Thousands of bloggers and journalists pick up those figures. When LLMs scan the web on digital marketing queries, they encounter this density of citations and put forward Semrush as the statistical authority. Pervasive presence in AI responses, regardless of clicks.

2. Featured Snippet Ownership — Owning the answer block

The featured snippet is Google's position zero: the box that appears above all organic results, with the answer extracted from a single web page.

The technique for capturing it requires a precise structure: open the section by echoing the exact words of the query, deliver the core answer within the first 40–60 words using short, coordinated sentences. Then the depth.

The Mayo Clinic dominates featured snippets on an enormous number of medical queries. The user gets the answer without clicking. And yet the Mayo Clinic sits consistently at the top of healthcare brand trust rankings. The public learns to trust a brand by watching it resolve their everyday doubts, day after day, without ever needing to visit the site.

3. Knowledge Panel Presence — The entity card

The Knowledge Panel is the identity card Google displays on the right-hand side of the page when a specific entity is searched. It's fed by the Google Knowledge Graph — a semantic database that connects information extracted from Wikipedia, Wikidata, official profiles, and government sources.

I see a recurring problem amongst Campania-based SMEs here. A great many long-established firms have fragmented Knowledge Panels — or worse, ones containing outdated information: incorrect opening hours, addresses of old premises, missing current logos. This inconsistency in NAP data (Name, Address, Phone) creates a semantic conflict in the Knowledge Graph. The algorithm loses confidence in the entity and stops surfacing it in local searches or AI summaries.

4. PAA Domination — Owning the related questions

"People Also Ask" boxes are a dynamic map of how users reason around a topic. When a brand manages to position its content across four or five sub-questions related to the same primary query, it achieves a visual encirclement effect. The user doesn't click, but reads the brand name associated with every variation of their problem.

The zero-click strategy in practice: five concrete actions

Action 1 — Audit your current citability

Before writing a single line or touching the site's code, you need to understand what AI models actually know about you right now. Forget traditional search. Open Perplexity, ChatGPT with web access, Gemini. Run two types of test:

Brand query: "Who is [company name] and what exactly do they do?" Analyse the result. Is the information accurate? Are the sources cited up to date? Can the AI identify your unique value proposition, or does it produce a vague, generic description?Geolocated category query: commercial intent prompts similar to those your clients would use. "Best corrugated cardboard packaging manufacturers between Naples and Caserta for the food sector?" Does your name appear? If not, analyse who's on the list and which web sources the AI used to surface those names.

I ran this exercise with Berto — at that café near the bypass, phone screen propped on the counter between two coffees. The result was a slap in the face he hasn't forgotten. The competitor who appeared in his place had half his company history but three years of structured trade articles, a mention on a regional craft portal, and customer reviews written with semantic keywords. For the algorithm, the competitor existed. Berto didn't.

The same pattern repeats itself in London. A financial consultant specialising in tax planning for startups, based in EC1V, who produces no editorial reports packed with data and market analysis becomes invisible on ChatGPT Search when founders query the system to work out how to optimise their seed capital. The geography changes. The problem is identical.

For a more structured methodology on building presence in localised searches, the article on GEO strategy for local Campania and English SMEs covers territory-specific case studies. And if you want to understand how to structure content so it's legible to AI algorithms from the writing stage itself, the article on GEO copywriting is the natural companion to this one.

Action 2 — Build AI-citable content, not just SEO-friendly content

There's a profound gap between old-school SEO and what I call AI-citability.

Traditional SEO often produced repetitive copy, optimised for search engines but a slog to actually read. AI-citability demands high informational density, logical clarity, and rigorous semantic architecture. Not keywords sprinkled through the text: real answers to real questions.

Four operational rules:

Explicit ontological definitions.

Open sections with clear formulations: "[Object/Service] is the system that enables...". RAG algorithms love direct definitions because they make it easier to extract clean snippets for reinsertion into the synthesis.Question–Straight Answer structure. Organise paragraphs using H3 subheadings as direct questions: "How much does it cost to install a 50 kW industrial photovoltaic system?". Provide the precise answer in the first line, with exact figures or ranges.Quantify with real data. Avoid "we're a leading company with many years of experience." Replace it with "Since 2012 we have designed and installed 147 industrial filtration systems across Campania, reducing average emissions by 34%." Generative models favour quantifiable information.Schema.org structured data. Implement FAQPage, Product, and LocalBusiness blocks in JSON-LD format in your code. It doesn't mathematically guarantee rich snippet display, but it significantly raises the probability that the algorithm will understand the semantic relationships between your site's data.

The article on how to get found by ChatGPT and AI search engines goes deeper into the technical indexing mechanisms with practical examples applicable even without an in-house development team.

Action 3 — Earn mentions on authoritative third-party sources

We need to accept a hard truth: what a brand writes on its own website carries only partial credibility for an LLM. It's a self-declaration. To shift the needle, you need third parties.

In the Italian context — and for Campania-based businesses in particular — this means developing digital PR relationships aimed at securing dedicated articles, interviews, or citations in regional or national outlets with a strong digital archive history: Il Mattino, Corriere del Mezzogiorno — or in vertical sector portals, category directories, and regional digital guides.

In the UK market, if the business operates in London, the target shifts towards Evening Standard, Time Out London, and professional directories such as Bark.com, Yell.com, and FreeIndex.

Imagine a joinery called "Rifugio Romano" — a fictional name for a real pattern I've seen repeat itself in the Caserta area at least four or five times over the past two years. If a journalist writes a feature on the finest bespoke woodwork in the province of Caserta and describes that joinery's craft process in detail, that article enters the global textual corpus indexed by LLMs instantly. The next time a user asks Perplexity where to order solid wood bespoke furniture in southern Italy, the algorithm will cross-reference that journalistic source and use that citation to validate the inclusion of "Rifugio Romano" in the response.

This mechanism has, however, a dramatic flip side: the permanence of bad press. LLMs don't filter mentions based on emotional tone. They register the density of available text. If the most authoritative mentions associated with your brand relate to poor service or controversy, the AI will absorb that narrative and reproduce it in its responses.

The United Airlines case from 2017 — a passenger forcibly removed from a flight — generated such a volume of articles, posts, and videos that the phrase "United Airlines passenger dragged" became embedded in the training data of the major language models. Years later, on queries that mention that brand in contexts of safety or customer management, the machine has a high statistical probability of evoking that episode. Bad digital press, in the age of LLMs, acquires an indelible permanence — at zero clicks.

Action 4 — Optimise your Knowledge Panel and structured data

Three immediate operational steps:

A complete and consistent Google Business Profile.

Make sure the registered company name matches exactly what appears on the website and social media. Rich descriptions with concrete copy, service categories mapped with precision, opening hours kept scrupulously up to date.

Schema.org in JSON-LD.

In the site's head, a LocalBusiness block with precise geographical coordinates, links to official social profiles via the sameAs property, and VAT number. Fill it in as though it's going to be read by a machine — because it is.

Absolute NAP consistency.

The company name, address, and phone number must be identical everywhere they appear: from commercial email signatures to the Companies House listing, right through to local directories. Discrepancies fragment the algorithm's confidence in the business entity.

Action 5 — Monitor your AI presence over time

Managing a zero-click strategy without measurement is like driving at night on a mountain hairpin bend with no headlights. Traditional KPIs lose their meaning: a first-page keyword ranking doesn't count if the user never sees that page. You need to monitor your presence in AI responses.

Three practical tools accessible without an enterprise budget:

Monthly manual control queries.

Define a fixed set of 15 high-value commercial queries for your niche and test them on Perplexity, ChatGPT, and Gemini. Log in a spreadsheet how often your brand appears relative to your direct competitors.

Branded search in Google Search Console.

Monitor impressions and clicks for queries that explicitly contain your brand name. An increase in these searches often signals that users discovered you inside a generic AI response and then moved to Google to look you up directly.

Anomalous Direct traffic in GA4.

If you notice a spike in "Direct" sessions that can't be explained by offline campaigns or email marketing, there's a good chance you're picking up users who encountered your brand in an LLM response. Cross-referenced with Search Console data, this lets you track the indirect effectiveness of your passive visibility.

Three traps nobody accounts for

Trap 1 — Handing your narrative over to the LLMs

The gravest risk of having no proprietary editorial strategy is the complete surrender of control over your own company story. If you don't write, if you don't document your processes, if you don't stake your claim on third-party sources with official data, LLMs will fill the informational vacuum by drawing on whatever textual fragments they can find online. The AI doesn't choose silence. It chooses the available source — even if it's outdated, partial, or plainly hostile.

The Parmalat case is the most telling I know of in Italy. After the financial collapse of 2003, the corporate structure was completely reorganised, and the brand returned to market with new industrial assets. Yet for years, virtual assistants kept surfacing the bankruptcy and the bond scandal as the primary context associated with the name. The sheer volume of text, the informational density, and the authority of the journalistic sources that had covered the collapse in 2003 far outweighed the copy on the relaunched company website. It took a massive programme of editorial production, digital PR, and semantic structuring to teach the algorithms that today's Parmalat was a fundamentally different entity from the one of the past.

Neglecting your editorial presence isn't a neutral choice. It's handing the keys to your reputation to whichever algorithm happens to pass by.

Trap 2 — Visibility without conversion (The Ghost Brand)

Being cited by AI assistants within category lists risks becoming a hollow vanity metric if it doesn't translate into real purchasing preference. Everyone knows you exist, you appear in every list Perplexity generates, but nobody chooses you when it comes to signing the contract.

This short circuit happens when digital presence is purely descriptive, without a clear Unique Positioning Statement repeated obsessively across all written output. If the AI summarises your business as "there's a communications agency in Naples, among them X, Y and Z", you're part of the scenery — not the answer to the problem.

The solution is to anchor your copy around a differentiating angle so sharp it forces the algorithm to include it in the synthesis. Not "communications agency" but "an editorial and creative studio specialising in cross-border positioning for SMEs operating between Italy and the United Kingdom, working directly between Caserta and Shoreditch". When the AI constructs its response, it won't be able to drop you into a generic list: it'll have to specify what makes you different.

Trap 3 — The structural volatility of AI models

A brand that has perfectly optimised its information architecture to be extracted as a primary source by GPT-4 might wake up one morning to find it's been downgraded in GPT-5's responses. Models constantly modify algorithmic weights, grounding criteria, and commercial partnerships with web data providers.

A zero-click strategy is not a one-off configuration task. It requires ongoing editorial and technical maintenance. The only defence against this volatility is redundancy: never tie the fate of your visibility to a single AI platform. Distribute your semantic authority broadly enough that it holds firm regardless of which assistant the user happens to be querying.

The right question isn't how many clicks I got. It's how many times I was the answer.

We're facing a paradigm shift that demands an immediate cultural reset. A brand's value is no longer measured by the number of sessions logged on a proprietary server, but by its ability to occupy digital conversations at the exact moment a buyer's opinion is forming — even when those conversations happen in digital rooms we don't own the keys to.

The new metric emerging in the most advanced brand tracking reports is AI Mention Share: how often your brand appears in LLM-generated responses for a set of queries relating to your category, measured against the frequency of your direct competitors. That's where the real competitive game is being played.

The good news — and I say this with no motivational agenda, purely as an observation of what I actually see working — is that winning this challenge doesn't require a multinational's budget. It requires editorial discipline, consistency in producing content dense with real data, precision in the semantic structuring of code, and a clear territorial positioning. All of it perfectly within reach of a City of London tech company just as much as a precision engineering workshop on the Caserta industrial estate, or a graphic design studio in the heart of Hackney.

Conclusion

London, 12:15pm. Outside on City Road, the rain has stopped. A shaft of cold light breaks through the low Shoreditch clouds and catches the screen of my laptop. I close the Perplexity tab. I haven't clicked on any of the three names that came up in this morning's analysis. I've generated no traffic. I've triggered no pixels. I haven't inflated the session statistics on any of their sites.

And yet the decision is made. One of those three names — that packaging studio that had the nous to seed the web with structured definitions, precise data on materials and production processes, verified mentions in trade publications — has lodged itself in my head as the only logical answer to my problem.

In a fortnight I'll open the address bar and type their name directly. Their systems will register a "Direct" traffic instance. Their marketing manager will look at the monthly visits graph and smile, convinced that user arrived by pure chance or a lucky moment of inspiration.

They'll never know the game started fourteen days earlier, in silence, on an AI interface, without a single click from anyone.

In the marketing landscape of 2026, if you don't appear in the answers, you don't exist. It doesn't matter how good the product is sitting in your warehouse.