The Death of the Link Click: How Answer Engines Change Everything

For twenty-five years, digital marketing ran on a single atomic unit: the click. Every model. CPC, CPM, session-based analytics, last-click attribution, conversion-rate optimisation, took the click as the moment of truth. Everything upstream was an input; everything downstream was an outcome. The click was the bridge.

That bridge is eroding. AI Overviews on Google, Perplexity's answer pages, ChatGPT Search, and Gemini's grounded responses all share one feature: they construct the answer in place. The user reads the synthesised response. They often absorb the recommendation. They may, or may not, click through. Google's own published data and every third-party study we have read since 2024 show the same pattern: on queries that trigger an AI-assembled answer, click-through rates fall 20% to 60% compared to the pre-AI baseline.

This is not a marginal change. It is a phase transition in the economics of discovery. And every marketing organisation still instrumented for the click-centric era is flying partially blind.

Here is the nuance that most CMOs are getting wrong in 2026: the commercial influence of search did not fall in line with click-through rates. It moved. It is now embedded in the answer the user reads before they decide whether to click.

A user who reads an AI Overview saying "The three most reputable Singapore-regulated forex brokers are X, Y, and Z, with X known for institutional-grade execution and the tightest spreads on major pairs" has already been influenced. If they click through to X, great, your analytics sees it. If they do not click and instead open X's app next week because the name stuck, your analytics sees nothing, but the influence is the same.

In our measurement work across B2B and premium B2C brands in 2025, we found that roughly 30–50% of the commercial influence of organic search had moved into the answer layer, meaning attribution models were silently under-crediting organic and search-adjacent investment by roughly that amount. The budget reallocations that followed, when teams finally saw this, were significant.

If your growth org runs on the following four measurement systems, you are already measuring an incomplete picture:

1. Organic traffic as the primary SEO KPI. Organic traffic is now a lagging, partial signal. A brand can lose 30% of organic traffic while gaining commercial mindshare if the lost traffic is on queries now answered upstream. Judging SEO health by traffic alone in 2026 is like judging brand health by ad-click volume.
2. Last-click attribution. Last-click has always been a crude model, but it was at least internally consistent. Once the AI answer is injecting influence upstream of the click, last-click systematically over-credits the final touchpoint (often a direct-navigation visit) and under-credits everything that shaped the intent inside the answer.
3. Session-based engagement metrics. Time on site, pages per session, scroll depth, all assume the user arrives on your site to learn. When the learning now happens inside the answer, users arrive on your site already decided, transact quickly, and bounce. Your engagement metrics look worse while your conversion efficiency looks better.
4. Content ROI measured by traffic. A cornerstone article that is cited in 40 AI answers per week and generates 300 direct-navigation visits a week from users who remembered the brand is wildly more valuable than a SEO-tuned article generating 10,000 bounce-heavy organic visits. Traffic-based ROI cannot see this.

Each of these systems needs to be supplemented, not replaced, with visibility-layer instrumentation that measures what happens inside the answer itself.

We now instrument every senior growth org with four metrics that sit alongside, not instead of, traditional KPIs:

• Share of Voice (SoV): of all commercially relevant prompts in your category, what percentage of AI answers mention your brand? Measured monthly across ChatGPT, Gemini, Perplexity, and Google AI Overviews via platforms like Workduo.ai.
• Citation Rate: when mentioned, how often is your brand named first, named with supporting detail, or cited with a direct link? This distinguishes genuine authority citations from throwaway mentions.
• Sentiment Index: the average sentiment (positive / neutral / qualified / negative) of the sentences LLMs construct about your brand. This is the AI-era equivalent of a Net Promoter Score, measured in the language the machine uses, not the customer survey.
• Brand-anchored assisted revenue: revenue from users who arrived via direct navigation, branded search, or a non-search channel within 14 days of a measurable uptick in AI visibility events. A mixed-methods metric, but directionally more honest than last-click in the AI era.

These four metrics do not replace sessions, conversions, or CAC. They sit alongside them and explain the 30–50% of commercial influence that click-centric measurement can no longer see.

The traditional funnel, awareness, consideration, decision, purchase, assumed that "awareness" happened across broad media and "consideration" happened on branded properties after a click. In the AI era, these stages collapse. A user asks Perplexity a high-intent question and, in a single answer, progresses from awareness to consideration to near-decision, before ever visiting your site.

We now design funnels around three new stages sitting on top of the traditional four:

1. Stage 00. Entity visibility. Is the brand even known to the LLM? Does it appear when prompted? Is it described correctly? (Addressed via entity signal engineering, Wikidata, schema, sameAs links.)
2. Stage 0A. Answer inclusion. When relevant commercial prompts are asked, does the brand appear in the answer, and in what framing? (Addressed via GEO + cornerstone content + citation graph work.)
3. Stage 0B. Answer preference. Does the answer frame the brand as the leading choice, one of several, or a hedged option? (Addressed via signal engineering, digital PR, review management, sentiment work.)

Only after these three stages does the traditional click-based funnel begin, and the traditional funnel now converts at two to three times the historical rate, because the users arriving have already been pre-qualified by the AI answer.

I run PPC for a living. I am telling you: the unit economics of paid search are changing in 2026 in ways that will unsettle most CFOs unless they understand why.

Three things are happening:

• Informational-query CTRs are collapsing on ad inventory that now sits below an AI Overview. Defending top-of-funnel paid placements on informational keywords is rapidly becoming bad math.
• Commercial-intent CPCs are rising because the users who do click are higher intent, and more valuable, than the pre-AI cohort. Paid converts better than it did, but the cost is climbing to reflect that quality.
• Branded search volume is fragmenting. Some users who would have typed your brand name now ask an LLM instead. Branded paid defence spend is less universally necessary, but more important on the specific branded queries that AI still directs clicks on.

The smart 2026 PPC playbook, the one we are running for our clients, is to reduce informational-query defensive spend, concentrate budget on the 5–15% of commercial-intent queries that still click at high rates, and move the saved budget into the organic/GEO layer that is now doing the top-of-funnel work. This is not a reduction in growth investment. It is a re-weighting toward where influence now happens.

The deepest implication of all of this is cultural, not tactical. For two and a half decades, marketing teams optimised for a unit, the click, that was cheap, trackable, and binary. The new unit is the sentence: the string of words an LLM constructs about your brand, inside the answer a prospective customer reads before any behaviour is logged. That sentence is harder to measure, harder to directly control, and vastly more influential than any single click ever was.

The shift asks a lot of us. It demands that we measure things we used to ignore. It demands that we influence a layer we used to treat as outside our control. It demands that we stop celebrating the click and start caring about the citation.

The brands that will win the next ten years of discovery are the brands that make this shift early, completely, and unsentimentally. The click is not coming back in the shape it used to exist. The answer is the new atomic unit. Manage the sentence. The click will follow.