The Traffic Pool Is Shrinking, and Which Channel Fills It Matters Less Than We Think

For years, the goal of digital marketing has been to figure out which channel sends the most visitors. We obsess over the ratio of organic search versus social or referral traffic, treating the "visit" as the primary metric of success. But this focus assumes that the total amount of traffic available is a constant, or at least growing. This connects with Is Google Fixing B2B Marketing? when the same signal needs a clearer operating decision.

The reality is shifting. We are entering an era where the total pool of available clicks is shrinking. When the goal is no longer just getting someone to land on a page, but ensuring a transaction happens regardless of where the user is, our entire approach to visibility has to change.

The distraction of the traffic ratio

Recently, the industry has been locked in a debate over how much traffic AI assistants actually send compared to traditional search. Several high profile studies have highlighted a stark gap. For instance, Marie Haynes analyzed GA4 data across 100 sites and found that ChatGPT sent a tiny fraction of the sessions that Google organic did, barely reaching 0.73% even when including all LLMs.

Other research supports this. The SEO Works found organic search delivered 39 times more traffic than AI assistants. Similarly, Kevin Indig tested whether AI visibility eventually translated into branded search or direct traffic and found no reliable link.

The common conclusion from these numbers is that AI sends almost nothing, so we should ignore "Generative Engine Optimization" and double down on Google. While it is true that Google remains the dominant source of visitors today, this conclusion misses a critical detail: the total volume of clicks is dropping across the board.

The danger here is focusing on the relative performance of channels while ignoring the absolute decline of the pool. If Google sends 99% of the traffic and AI sends 1%, but the total number of people clicking through to websites is plummeting, winning the "ratio" battle is a hollow victory. The same pattern also shows up in LLM Traffic Converts Differently, where the practical question is how the signal becomes visible.

Expert Interpretation: The tradeoff here is between short term stability and long term viability. If you only optimize for the current ratio, you are optimizing for a disappearing world. The decision you need to inspect is whether your current KPIs are based on "visits" or "outcomes." If your primary metric is still session count, you are measuring a shrinking asset.

Why the click is disappearing

People aren't necessarily switching from one search engine to another; they are simply stopping the act of clicking. We are seeing a move toward interfaces that provide the answer in place, removing the need to visit a website entirely.

Data from Bocconi University researchers suggests that ChatGPT produces an outbound click in only about 5.2% of conversation sessions, whereas Google queries result in clicks 31.1% of the time. More concerningly, the availability of AI search has already begun to reduce traditional search usage.

Even Google's own ecosystem is contributing to this. Recent data indicates that for every 1,000 searches in the U.S., only about 232 clicks actually reach the open web. This creates a strange paradox: Google can report record high query growth while simultaneously seeing slower revenue growth and fewer clicks leaving its ecosystem.

The "shrinking pool" is the result of the search experience becoming a destination rather than a doorway. When the answer is delivered on the results page or within a chat interface, the visit to your site becomes an unnecessary step for the user.

Expert Interpretation: This is a fundamental shift in user behavior. The tradeoff is convenience for the user versus visibility for the brand. You cannot "fight" this trend with better keywords; you have to accept that the "visit" is no longer the guaranteed first step of the customer journey.

Shifting from visits to completed actions

If the visit disappears, the transaction still needs to happen. This is where the concept of the "agentic web" becomes vital. We are moving toward a world where AI agents do not just provide information, but execute tasks on behalf of the user.

Google is already moving in this direction. Leadership has indicated that agents will act as a unifying layer to book reservations, fill out forms, and make purchases. In this model, the agent doesn't send a visitor to your site to book a room; the agent books the room directly.

This changes the fundamental unit of success. We are moving from the "visit" to the "completed action." The goal is no longer to get a human to land on a landing page, but to be the option the agent selects to finalize the request.

Interestingly, agents don't choose the same way humans do. A McGill study comparing AI agents to human data from Expedia found that agents search more deeply and never decline to book. While position still matters, it matters differently; for example, agents were found to inspect the middle of a list the least, which contradicts traditional human heatmaps. A useful companion note is Cloudflare’s PACT Is Not Live Yet, because it looks at a nearby part of the same system.

Expert Interpretation: The critical decision here is to stop treating your website as the only point of conversion. If an agent can finalize a sale without a human ever seeing your homepage, your "conversion rate" as measured by GA4 becomes a vanity metric. You must start optimizing for "agent selectability."

Building an agent ready business

To survive a shrinking traffic pool, you have to make your business legible to the machines that are now mediating the transaction. This requires moving away from prose and toward structured data.

Audit your actual pool

Stop looking at the ratio of AI vs. Google. Instead, track your total inbound sessions from all sources monthly. Compare this against the total impressions for your category queries. By segmenting this by page type, you can see exactly where the leak is happening and decide at what point a decline in traffic requires a pivot in strategy.

Develop an undominated offer

Agents make decisions based on attributes they can read, such as price, ratings, delivery speed, and terms. To be chosen, you need an "undominated offer." This means looking at the 3 to 5 key attributes in your category and ensuring no competitor beats you on all of them simultaneously. If you can't be the cheapest and the fastest, you must clearly state your specific trade off so the agent can categorize you correctly.

Standardize your reputation data

An agent cannot "feel" a brand's vibe; it reads review scores and counts. Ensure your reputation data is current, structured, and consistent across all platforms. If an agent sees conflicting ratings across different directories, it may perceive the business as a risk and skip it entirely.

Convert policies into data

Stop hiding your shipping, returns, and warranty policies in long paragraphs of text. Agents struggle with prose but excel with data. By placing these policies in structured fields, you allow an agent to compare your terms as easily as it compares your prices.

Maintain complete, machine readable feeds

For retail, this means filling every single attribute in the Merchant Center for price and availability. For service providers, it means publishing packages, service areas, and availability in a machine readable format. The goal is to feed the "Universal Cart" and other agent layers directly.

Ensure factual consistency

Your name, pricing, hours, and specifications must be identical across your website, your social profiles, and third party directories. When an agent cross checks these sources and finds a conflict, it reads that inconsistency as a risk factor, which can lead to your business being dropped from the short list.

Expert Interpretation: The tradeoff here is between "creative marketing" and "technical precision." While humans like storytelling, agents require data. The decision you must make is to prioritize the technical infrastructure of your offer over the aesthetic presentation of your website. In an agent driven economy, being "easy to parse" is a competitive advantage.