Opaque Funnel: The Moment Before
How AI moved into the marketer’s first point of influence
For decades, marketers built around a simple assumption: if a customer was entering the market, the brand had a chance to reach them before the decision began to harden.
That assumption shaped the modern marketing system.
Brand campaigns created memory. Search captured intent. Social generated awareness. Influencers created aspiration. Retail media intercepted shoppers near purchase. Loyalty programs encouraged return. Performance marketing optimised the path from impression to click to conversion.
The funnel was never perfect. But it was visible enough to manage.
A marketer could see where traffic came from. They could observe search queries. They could measure click-through rates. They could compare campaigns, optimise landing pages, retarget visitors, and attribute outcomes with at least some confidence.
The buyer journey had many gaps, but the enterprise had instruments pointed at it.
AI-led discovery changes that.
A growing number of customers are no longer beginning with a search result page, a product listing, a brand website, or a comparison article. They are beginning with a question.
“What is a good watch to gift my husband under SGD 250?”
“Which skincare brand should I trust for sensitive skin?”
“What is the best festive gift for clients?”
“Should I buy this phone now or wait for a better deal?”
“Which CRM is better for a growing sales team?”
These questions do not behave like search keywords. They are not merely requests for information. They are requests for judgement.
The customer is not asking for a page of options. They are asking an AI system to reduce the market on their behalf.
That is the moment before.
Before the search. Before the click. Before the landing page. Before the retargeting pixel. Before the brand has any direct opportunity to intervene.
The customer asks. The AI answers. A shortlist begins to form.
The marketer was not in the room.
The funnel did not disappear. It became opaque.
It is tempting to describe this as another channel shift.
Search changed. Social changed. Marketplaces changed. Mobile changed. Influencers changed. Marketers adapted each time.
But AI-led discovery is different because it does not simply add another surface where brands need to appear. It inserts an inference layer between marketing activity and customer consideration.
That layer interprets the customer’s need, evaluates available evidence, compares possible options, and recommends a path forward. The customer may still click. They may still search. They may still visit a brand website. But by then, the first act of consideration may already have happened elsewhere.
This is the opaque funnel.
Marketers are still funding visibility, influence, acquisition, and retention. But a growing share of customer consideration is now being shaped inside AI systems the business cannot observe, measure, directly access, control, correct, or update with the speed and certainty modern marketing requires.
A brand may be visible in campaigns but absent from the AI-generated shortlist.
It may be well positioned on its own website but misrepresented by third-party sources.
It may have loyal customers whose relationship history is invisible to the assistant advising their next purchase.
It may spend heavily to create awareness, drive traffic, improve recall, or win consideration. Only to lose the decisive moment inside an answer the marketing team never saw.
The damage does not appear as a neat line item called “AI invisibility.” It appears as demand leakage: rising CAC, weakening ROAS, distorted attribution, lower organic yield, underperforming campaigns, and retention investments that no longer fully protect customer preference.
The funnel has not disappeared. It has become opaque at the very moment where consideration is formed and preference begins to harden.
The festive season is the stress test
Most structural changes in marketing arrive quietly. They first appear as anomalies: a channel underperforms, traffic softens, paid efficiency declines, organic yield changes, attribution becomes harder to explain, even when campaign execution looks competent.
The festive season makes these anomalies harder to ignore.
Festive demand compresses the entire marketing system into a short window. Purchase intent rises. Promotions intensify. Gift decisions multiply. Price comparison becomes more active. Consumers become more open to switching brands. Retailers and D2C brands increase spend to defend attention.
In that environment, AI-led discovery becomes commercially legible.
A customer who once typed “best watches under SGD 250” into Google may now ask an AI assistant for a recommendation based on recipient, budget, taste, occasion, and social context. A shopper comparing offers may ask whether a festive deal is actually worth it. A repeat customer may ask for alternatives without realising that the AI assistant has no memory of their loyalty history with the brand.
The shift is no longer theoretical. Bain reported in April 2026 that 44% of US online buyers it surveyed either mostly start their shopping journey in a large language model or split their search between AI tools and traditional search engines. Adobe reported that traffic from AI sources to U.S. retail sites rose 693% year over year during the 2025 holiday season, followed by 393% year-over-year growth in Q1 2026.
Those growth rates need context. AI referral traffic is still emerging, and Adobe has previously described it as modest compared with major channels such as paid search and email. But the signal is becoming harder to dismiss because the quality of those visits appears to be improving. Adobe reported that AI-referred retail visitors converted 42% better than non-AI traffic in March 2026, reversing the pattern from a year earlier when AI visitors converted at nearly half the rate of other visitors.
At the same time, the visible search journey is becoming less reliable as a source of clicks. SparkToro and Similarweb reported that 68.01% of Google searches in the first four months of 2026 ended without a click. This figure excludes searches within Google’s mobile app, where zero-click rates are likely higher still, based on Similarweb’s desktop and mobile clickstream panel. Ahrefs found that Google AI Overviews reduced clicks to top-ranking content by 34.5% in 2025, and by 58% in its 2026 update.
The exact numbers will vary by market, category, and platform. But the direction is clear enough for marketers to take seriously: AI-led discovery is growing from a small base, while the click value of traditional search visibility is being compressed.
Even that may understate the strategic role of AI discovery. Referral data captures the visible click from an AI source. It does not capture the invisible influence of an AI answer that shapes the shortlist before the buyer later arrives through Google, a marketplace, a direct visit, or a paid ad. For marketers, that distinction matters. The opaque funnel is not only about where traffic comes from. It is about where consideration was formed.
The festive season is not the cause of the opaque funnel. It is the moment when the cost of it becomes visible.
The moment of intent has moved upstream
The classic marketing model assumed that first consideration could be shaped through visible channels.
A customer might see an ad. Search for a category. Read a review. Visit a marketplace. Watch an influencer video. Compare options. Visit the brand website. Add to cart. Convert.
Each step produced signals. Imperfect signals, but signals nonetheless.
AI-led discovery changes the order.
The customer can now compress several steps into a single prompt:
“I need a premium but not flashy Deepavali gift for clients under SGD 100. It should feel useful, not generic. What should I consider?”
That one prompt contains:
budget
occasion
recipient type
social intent
aesthetic preference
fear of seeming generic
category openness
purchase urgency
implied trust criteria
In traditional search, the marketer might have seen fragments of that intent across multiple queries and clicks. In AI-led discovery, the system sees the full context at once.
Then it judges.
It may recommend categories. It may name brands. It may exclude options. It may warn against certain products. It may compare retailers. It may create a shortlist before the customer ever reaches a visible marketing surface.
This is why the moment before matters.
The marketer has not merely lost a click. The marketer has lost access to the point where intent becomes structured.
Visibility is no longer the same as inclusion
Marketers are used to thinking in terms of visibility.
Did the campaign reach enough people? Did the brand rank? Did the ad appear? Did the influencer post perform? Did the marketplace listing get impressions? Did the email get opened?
AI-led discovery introduces a harder question:
Was the brand included in the answer?
A brand can have visibility and still lose inclusion. That distinction becomes more important as AI summaries and assistants separate being present in the information ecosystem from being selected into the answer.
It can be present on Google but absent in ChatGPT. It can be active on Instagram but ignored in Gemini. It can have strong marketplace presence but weak recommendation presence. It can rank for keywords and still fail to appear when the customer asks a richer, more personal question.
That distinction is critical.
Visibility is exposure. Inclusion is selection.
The opaque funnel is not merely about whether the customer saw the brand. It is about whether an AI system considered the brand worthy of being surfaced at all.
This is the new marketing anxiety.
Not “Are we on page one?” Not “Did our ad get served?” Not “Did the customer click?”
But:
“Did the AI include us in the shortlist?”
The business still pays for the visible journey
The uncomfortable reality is that most marketing budgets are still organised around the visible journey.
SEO teams optimise for rankings. Paid search teams bid on intent. Brand teams buy reach. Content teams publish explainers. Influencer teams create social proof. Retention teams build loyalty flows. Ecommerce teams optimise product detail pages.
None of this becomes irrelevant.
But all of it becomes insufficient if a growing share of consideration is happening upstream in AI-generated answers.
If the AI assistant has already shaped the customer’s shortlist, paid search may be capturing intent after preference has begun to harden. SEO may be optimising for rankings whose click value is declining. Influencer content may create awareness without becoming usable evidence. Loyalty may fail to protect repeat consideration if the assistant does not know the relationship exists.
This is how CAC rises without an obvious culprit.
This is how ROAS weakens while campaign execution looks competent.
This is how organic traffic underperforms even when rankings appear stable.
This is how attribution models become less useful: they are measuring the visible part of a journey whose decisive moment happened elsewhere.
The marketer is still funding the funnel. But part of the funnel has moved into a system the marketer cannot see.
The new question for marketing leaders
The first response to this shift has been tactical.
Optimise for AI search. Create clearer content. Improve structured data. Seed third-party mentions. Track AI citations. Monitor brand visibility across platforms.
These actions matter. They will become part of modern marketing hygiene.
But the larger question is not tactical. It is operational.
Who inside the marketing organisation is responsible for the moment before?
Not the campaign. Not the click. Not the landing page. Not the conversion event.
The moment before.
Who knows whether the brand appears in AI-generated answers? Who knows which competitors are being recommended instead? Who checks whether the brand is being described accurately? Who can connect those invisible recommendation moments to CAC, ROAS, retention, and pipeline?
Most organisations do not have a clear answer.
SEO does not fully own it. Brand does not fully own it. PR influences parts of it. Product marketing shapes parts of it. Ecommerce controls parts of it. Analytics struggles to measure it. Legal arrives only when the misrepresentation is severe enough to matter.
The opaque funnel cuts across all of them.
That is not an AI search problem. It is a marketing operations problem that no one has yet been asked to solve.