The Zero-Click Marketer
What happens when AI reads your emails, buys your products, and your customers never knew you existed
Zero-click in search, zero-click in email marketing, zero-click in the customer buying journey…AI continues to disrupt how customers (people!) interact with brands, products and services and this, in turn, disrupts the usual digital marketing / performance playbook marketers have been relying on for a while now…
It was inspired by what I read from Florian Delval, Jess Graham and Dela Quist on this topic that I wrote the article below. When this type of customer data and signals "disappears” we get to the Zero-Click Marketer.
I hope you enjoy (and, as always, just reply to this email and we will continue this convo!).
On January 8, 2026, Google announced Gmail was “entering the Gemini era.” The update gave 3 billion monthly active users AI-powered summaries of email threads, natural-language search across their inbox, and a new AI Inbox view prioritizing messages by inferred importance. All on by default. If you wanted it off, you had to go find the setting and disable smart features entirely, tabs included.
In September 2025, OpenAI launched Instant Checkout inside ChatGPT. In January 2026, Google announced the Universal Commerce Protocol at the National Retail Federation conference, backed by Shopify, Etsy, Wayfair, Target, Walmart, and Visa. Amazon’s “Buy for Me” feature has been live since April 2025, letting an AI agent purchase from third-party sites without the customer ever leaving Amazon. Some of those brands never agreed to participate.
Two disruptions hit within months of each other. In the inbox, AI moved from helping marketers write emails to standing between the marketer’s message and the human reader. In commerce, AI moved from recommending products to buying them. Both disruptions have the same structural effect: the customer is no longer the first audience for the marketer’s work. A machine is.
I have also read somewhere that, now with AI in the inbox, email marketing can change from an owned to an earned channel and this certainly captured my imagination and left me thinking how to adapt as a marketer.
The inbox becomes a query engine, not a browsing experience
Florian Delval, writing in his Foreign Key newsletter (which helped inspire this article), named the shift precisely: the zero-click mailbox. The concept borrows directly from what happened to Google Search. AI Overviews in Search give users answers without clicking a blue link. AI Overviews in Gmail give users answers without opening an email. Subject lines become the new blue links. And just like in search, the behavior changes before anyone notices.
Delval traces the precedent. Gmail’s spam filter. The 2013 Promotions tab. Apple’s Mail Privacy Protection in 2021. Each one shifted visibility. None of them changed the fundamental architecture of email. Gmail still delivered messages and let you decide what mattered. And yet, the January 2026 announcement crosses a different line. Gmail can now decide what the email is (semantic classification), whether you should care (AI relevance inference), and what you should do about it (to-dos, reminders, suggested actions).
That distinction matters. As Delval puts it, Gmail is moving from delivery and sorting to interpretation and decisioning. The inbox becomes something you query rather than something you browse. And if AI Overviews treat your email as a dataset and Gmail as a query engine, the rules for getting seen change from the ground up.
What we lose when curation replaces exposure
Dela Quist, who has spent thirty years in email marketing makes a different argument. For Quist, the real casualty is not opens or clicks. The real casualty is the inbox’s role as the last largely "uncurated” digital space (I had not yet thought of email marketing in this way!)
Search has been curated for years. Social feeds are ranked. Media is filtered and optimized. Email was the exception. If someone sent you something, it arrived. You decided what to do with it. Quist argues in his LinkedIn article “When Your Inbox Stops Delivering” that Gemini changes this. The inbox stops being a library and starts behaving like a secretary: helpful, efficient, and quietly judgmental.
The problem Quist identifies runs deeper than marketing metrics. He calls it the “illusion of completeness.” When an AI mediates your inbox, you feel fully informed because nothing feels missing. There is no obvious censorship, no warning light, no friction. But what you’re seeing is already the surface, curated and ranked and compressed. And once you believe the surface is complete, you stop looking underneath.
Email has always allowed for what Quist describes as “irrelevant-now, relevant-later” moments. Ideas landing sideways. Messages not mattering today but shaping how you think tomorrow. An AI inbox optimized for relevance removes those moments by design. Quist draws the parallel: maps made navigation easier and killed our sense of direction. Feeds made information abundant and collapsed attention. An inbox deciding what matters risks degrading contextual thinking. If you only see what the system believes aligns with your past behavior, you lose contrast. And contrast is where new thinking starts.
One of his sharpest observations: ghosting by algorithm is indistinguishable from human intent. If a message from a trusted contact never surfaces because the system scored it as low value, you don’t experience censorship. You experience absence. And you assume that absence was the sender’s choice.
The buyer journey collapses at the same time
The inbox disruption would be significant on its own. But it is hitting at the exact moment the purchase journey is being compressed by the same forces.
Jess Graham, founder of Banquet and former Global Product Marketing lead at Instagram, has been tracking this collapse in her Substack newsletter The Feast. She has some provocative analysis based on some sobering data. Salesforce and Adobe independently found approximately 39% of consumers are already using AI to shop. Not curious about it, not open to trying it. Already doing it. SparkToro’s December 2025 research found less than a 1% chance the same brands appear across two AI responses to the same query. According to Graham, your brand is not being rejected but flickering in and out of existence with zero transparency.
Graham also introduces two frameworks worth holding onto. The first is the distinction between Algorithmic Legibility and Agentic Invisibility. Algorithmic Legibility is about being readable to the machine: structured product data, optimized content, plentiful reviews. Agentic Invisibility is what happens to the human experience when the machine does the choosing. These are two sides of the same coin. A brand readable to the machine becomes invisible to the person. The agent evaluates, the agent buys, and the customer never saw the packaging, never compared, never felt anything.
Graham’s term for this is “evaluation compression.” The cognitive labor of comparing, weighing, and deciding shifts off-screen entirely. The customer never sees the analysis. An agent purchasing you is not a customer choosing you. The distinction matters because choice builds attachment. Procurement does not. As Graham writes: nobody has ever fallen in love with sourcing.
Her second framework is the Discovery Tax: the compounding cost brands pay every quarter they fail to address this shift. The components are specific. Reduced premium tolerance as customers lose attachment to brands they never chose. Substitution volatility as agents swap you out on marginal differences in rating or price. Rising retention costs to replace the relationship-building previously done through natural exploration. And growing dependence on agent-facing optimization you do not control.
The industry will sell optimization. The fundamentals will not change.
Both disruptions are already producing the predictable industry response: optimize for the machine.
Quist is blunt about this in his Linkedin article “Writing for the Inbox Robot.” He argues the email industry is about to rebrand inbox placement as “summary placement,” selling the same product twice. The pitch will shift from “we’ll get you out of the spam folder” to “we’ll get you to the top of the summary.” Score the language. Tune the structure. Hack the signals. Sell confidence as a metric. The incentives come back in new clothing.
Quist’s counterargument is structural: permission does not change, the competitive set does not change, email is still opt-in distribution. The brands in someone’s inbox are there because the person gave permission at some point. That permission is the asset. AI re-orders the pile. It does not replace the pile. The underlying competition stays the same: the same subscriptions, the same brands, the same permission base.
Graham makes a parallel argument about commerce. Brands are racing to win “Top of Algorithm,” structuring product data for agents instead of humans. And the effort is not wrong, exactly. Brands cited in Google’s AI Overviews do see higher organic and paid clicks. AI-referred traffic spends more time on site. Algorithmic Legibility works. But the board meeting question (”how fast do we win this new game?”) obscures a harder question: what are brands winning when they win this way?
The deeper problem, in Quist’s framing, is that AI is not a suppression layer. It is an amplification layer. Good performance gets rewarded faster. Bad performance gets punished faster. Weak tactics lose their camouflage because the system is watching outcomes, not intentions. If the robot surfaces your best offers with the best timing to the right people and performance still does not move, the question gets sharper: do people want what you are selling badly enough to keep hearing from you?
That is a value question. Not a deliverability question. And Quist argues it is the question the email industry has always tried to keep off the table.
The robot responds to the same levers humans do
Here is the part most marketers will miss, and it is Quist’s most counterintuitive point.
AI inboxes are not built to think independently. They are built to approximate what humans tend to want, using behavioral feedback at scale. This means they naturally over-weight the same behavioral levers already proven to work on people, because those levers are how humans reveal preference under uncertainty. Any system trained to predict preference will learn the same patterns.
Quist walks through specific examples. Anchoring: a high number next to a lower one creates “value” for humans, and a summarization system will surface salient deltas for the same structural reason. Loss aversion: “your points expire” triggers a prevention response in humans, and a robot trained to be a helpful assistant will treat loss language as an actionable task. Social proof: “10,000 reviews” and “bestseller” serve as strong metadata signals because they are concrete entities and quantities. Scarcity and countdowns: “ends tonight” creates urgency for humans, and a literal system will treat time boundaries as actionable constraints.
Humans respond emotionally. Robots respond to patterned salience. But the direction of the response is the same. This matters because it means the fundamentals of effective marketing do not change in a machine-mediated world. They become more measurable, more quickly. The robot does not make creative less important. The robot makes creative more accountable.
What gets lost has a price tag
Graham’s historical argument gives the cognitive loss a financial shape.
In “You Call It Friction, I Call It Human Experience,” she traces a pattern across two decades of commercial decisions. Comparison sites turned credit card brand-building into rows of APR numbers. DTC turned subscriptions into inertia (a subscriber who receives something month after month without thinking is not loyal, they are passive). Agents turn purchase into procurement. At each stage, the human experience of browsing, considering, and wanting got optimized away.
Her case studies are specific. Fab.com built a daily design marketplace around curation. Within six months, a million people signed up. Then Andreessen Horowitz arrived with $40 million and growth targets. Merchandising teams optimized for volume. The curated flash sale became a catalogue. Valued at $1 billion in 2013, Fab sold in 2015 for somewhere between $15 million and $50 million. The founder’s internal memo: “I guided us to go too fast. I enabled us to lose our core focus.” Stitch Fix launched on a genuine idea (human stylists who learned your taste) and IPO’d at $1.6 billion. Under investor pressure, it replaced the human styling layer with an algorithmic model. The stock is down more than 90% from its IPO price. Glossier recovered by returning to what Emily Weiss had originally built: community as the product, goods as the proof. By 2024, they were profitable again.
The pattern is consistent. Extraction does not compound.
Graham directly challenges Shopify CEO Tobi Lütke’s claim at the UCP launch event that agentic commerce enables “serendipity.” She cites Christian Busch’s research: luck happens to you, serendipity requires your participation. What Lütke described is targeting. One builds attachment and anticipation. The other is a brown box showing up on the doorstep.
The zero-click marketer as a new professional identity
This is the part I have not seen anyone else name yet. The disruption demands a new type of marketing practitioner, one who has internalized a dual audience (machine first, human second) and refuses to optimize for one at the expense of the other.
The zero-click marketer is not a specialist in AI tooling or a rebrand of the email deliverability consultant. The zero-click marketer is a practitioner whose work operates on two layers simultaneously.
The machine layer determines whether your message gets surfaced at all. This means structured, semantically clear content. Plain text may outperform designed HTML in an AI inbox because the summarization engine reads text, not layouts. Entity-rich language gives AI something to parse, categorize, and surface accurately. Delval predicts the emergence of an “Email Context Protocol”: a standard framework with clear CTAs, explicit dates, and structured information elements the AI system is built to process. The behavioral economics levers Quist identifies (anchoring, loss aversion, scarcity, social proof) work on the machine for structural reasons, not emotional ones, but they work. This layer is about signal clarity, not manipulation.
The human layer determines whether the message matters once it reaches a person. This means relationship value strong enough to survive algorithmic filtering. Ryan Phelan, writing on MarTech.org, argues that sales-only messaging will get buried deeper than the spam folder. An AI agent will bury brands whose only message is “buy this.” Marketing based on relationships, where the subscriber has genuine reasons to stay engaged, will fare differently. Graham’s concept of “Strategic Friction” belongs here: deliberately creating moments of agency, surprise, and meaning in the buyer journey rather than optimizing everything smooth. (I also addressed Strategic Friction as part of marketing processes). Graham mentions the example of Gentle Monster, the South Korean eyewear brand, employs sixty people to design its stores and six people to design its glasses. The stores are immersive art installations. You go for the experience and then buy glasses. They have built something the algorithm does not know how to evaluate: a reason to walk in.
The zero-click marketer also needs to reckon with measurement. Open rates were already unreliable after Apple’s MPP in 2021. In a zero-click inbox, they become meaningless. But here is the feedback loop problem Delval identifies: Gmail’s own algorithm currently infers importance from engagement signals like opens and clicks. If users adopt AI Overviews and AI Inbox as their primary consumption mode, Gmail itself starts losing those signals. Fewer opens. Fewer clicks. Less direct engagement to measure. The system interpreting your email starts losing the data it uses to interpret your email. New metrics will need to emerge. The marketers who instrument and test those new signals first will have a structural advantage.
Few predictions of my own: maybe the new email metrics will include metrics that are more in-line with branding than performance marketing: time on message, number of messages saved (versus swiftly deleted), and even heat maps can become more relevant.
Frequency as the compound lever
Quist offers one more argument worth sitting with. Once you have earned the right to be in someone’s inbox (the subscriber has not unsubscribed, complained, or blocked), the final email marketing lever returns where it has always been: frequency.
This is counterintuitive to the instinct many marketers will have, which is to send less in a machine-mediated inbox. Quist’s logic: if someone does not opt out, more impressions still win. Not because creative does not matter, but because frequency is how you compound whatever value you have earned. If AI Overview treats the inbox as a dataset, frequency means more data points the system has to work with. The question is whether the value per message holds up at higher frequency. If it does, the robot rewards it. If it does not, the robot punishes it faster than a human ever would.
This is where Quist and Graham’s arguments converge. Quist says: making the offer good enough that people stay. Graham says: giving people a reason to form a relationship. Those are the same argument stated from different starting points, one from the email practitioner’s chair and one from the brand strategist’s.
The experimental marketer’s move
The playbooks for a zero-click world have not been written yet. The Gmail AI Inbox is still rolling out to testers. The Universal Commerce Protocol is weeks old. eBay has already banned AI shopping agents. Amazon is expanding them. The rules are being written in real time.
In the framework I'm exploring in my upcoming book Strategic Marketing Skills: The Experimental Marketer (Kogan Page, October 2026), I call these moments where the playbook hasn't been written the intersection of Explore and Experiment. Explore is the work of studying how new systems operate before best practices calcify. Experiment is the work of testing before you have proof, running structured trials with new content formats, new signals, new measurement approaches while the cost of learning is still low. The marketers who did both during the early days of first-party data strategy (for example) built advantages their competitors spent years trying to close. The zero-click inbox is the same kind of window.
What is clear is the shape of the problem. The zero-click marketer operates in a world where the customer’s inbox has a bouncer and the customer’s purchase path has a proxy. Optimizing for the machine is necessary work. It is not sufficient work. The brands and practitioners who will lead are the ones building genuine relationship equity strong enough to survive algorithmic filtering, creating value the system rewards, and developing the human resonance the system does not know how to evaluate.
Sources:
Florian Delval, “The Zero-Click Mailbox,” Foreign Key (Substack), February 26, 2026
Dela Quist, “When Your Inbox Stops Delivering, It Starts Deciding,” LinkedIn, 2026
Dela Quist, “Writing for the Inbox Robot,” LinkedIn, 2026
Jess Graham, “You Call It Friction, I Call It Human Experience,” The Feast (@comefeast, Substack), 2026
Jess Graham, “Meanwhile, as You Were Optimizing,” The Feast (@comefeast, Substack), 2026
Ryan Phelan, “Your next marketing challenge may be winning over the AI in charge of the customer’s inbox,” MarTech, April 24, 2025
Google, “Gmail is entering the Gemini era,” Google Blog, January 8, 2026
Data sources cited by Jess Graham: SparkToro brand consistency research (December 2025), Seer Interactive AIO/CTR study (September 2025), Salesforce AI shopping adoption (2025), Adobe AI-driven traffic study (2025), BCG and Moloco “AI is Collapsing the Marketing Funnel” (2026), BoF-McKinsey State of Fashion 2026, Christian Busch The Serendipity Mindset (Riverhead Books, 2020)