Zero-Click Search: What Happens to Your Content ROI When Most Searches Never Leave Google

For fifteen years the informational blog post was the reliable top of an eCommerce content funnel. You answered "what is a good return rate for apparel," you ranked, you earned the click, and some fraction of those readers eventually bought. That machine is quietly breaking, and most operators are still funding the broken part.

This is the content-mix decision under zero-click: which content to stop making, which to protect, and which to fund harder. It is a companion to — not a repeat of — our piece on whether generative engine optimization is worth the investment. That article answers when to fund GEO as a program. This one answers a different question: given that a large share of searches now end without a click, how should you reallocate the content budget you already spend across content types? Keep the two apart. One is a channel-timing decision; this is a portfolio decision.

How much search is really zero-click now?

Start with the number that reframes everything. SparkToro, analyzing Similarweb clickstream data for U.S. Google searches from January through April 2026, found that 68.01% of searches ended without a single click to the open web — up from 60.45% in 2024. The share of searches sending at least one click to an external site fell 9.51 percentage points in two years, a 22.9% decline in the thing your content funnel actually runs on.

The mechanism behind the acceleration is the AI answer. Pew Research Center tracked 68,879 real Google searches from 900 U.S. adults and found that when an AI summary appeared, users clicked a traditional result in just 8% of visits — against 15% when no summary appeared. That is a 47% relative drop in click-through the moment Google decides to answer the question itself. Clicks on the sources cited inside the summary happened in only 1% of visits.

And AI answers are appearing on more queries over time, even as the exact rate wobbles. Semrush's tracking put AI Overview prevalence at 6.49% of queries in January 2025, peaking near 24.6% in July 2025, then settling around 15.7% by November after Google recalibrated. The direction is a ratchet, not a fad. When an AI Overview does appear, Similarweb data cited across the 2026 reporting puts the zero-click rate around 83%; on Google's newer AI Mode, closer to 93%.

Why does zero-click break the content P&L, not just traffic?

Here is the part the "zero-click survival" advice skips. The damage is not evenly spread across your content. It concentrates almost entirely in one bucket: content whose answer fits inside a paragraph.

Think about what an AI summary can and cannot replace. "What is a healthy gross margin for eCommerce?" is a definition — it fits in three sentences, so Google now serves it and the click evaporates. "Should I sell wholesale or DTC given my margins?" is a judgment that depends on the reader's specific numbers — it does not compress into a summary, so the click survives. Same content program, two completely different fates under zero-click.

That means your content P&L has silently split in two. One half is producing pageviews that no longer convert to clicks, let alone customers. The other half still earns its click and still converts. If you keep budgeting by old habit — a steady drip of informational posts because "content marketing works" — you are funding the dead half at the expense of the live one.

The correct unit of measurement changes accordingly. Cost per pageview and even cost per click are now misleading, because the highest-volume informational content generates impressions that never become visits. The only honest unit is cost per acquired customer, per content type. Measured that way, a large slice of most operators' content libraries has quietly moved to an infinite cost-per-customer — all spend, no downstream click. This is the same trap we describe in the SEO mirage, where traffic that does not convert flatters the dashboard while the business stalls; zero-click just removes even the vanity traffic.

Which content stops paying off?

Stop, or stop making more of, the content whose entire value was delivering a definition or a generic procedure that an AI now delivers in place.

• Definitional and "what is / how does" posts. "What is a chargeback," "what does DTC mean," "how does dropshipping work." These are the first casualties. The answer is a paragraph; Google serves the paragraph; no one needs your page. Because the click-through has been structurally removed, more of these posts is pure cost.
• Generic how-to and listicle content with no proprietary judgment. "10 tips to reduce cart abandonment," "how to write product descriptions." Interchangeable with a thousand other pages and trivially summarizable. AI systems synthesize these into a single answer and cite none of you in a way that sends traffic.
• Trend-chasing explainers. "What is [new platform feature]." By the time you rank, the feature is documented by the platform and summarized by the AI. You are subsidizing someone else's answer engine.

The discipline here is subtraction, and it is uncomfortable because these posts often still show impressions in your analytics. Impressions are the last thing to die. Do not confuse a page that still appears in search with a page that still does work. Ask the harder question: in the last 90 days, did this piece produce a click that produced a customer? For most informational TOFU under zero-click, the honest answer is no.

Which content still earns its click?

Three kinds of content survive zero-click, and they survive for structural reasons you can rely on — not because Google is feeling generous.

Comparison and decision content. "BigCommerce vs Shopify at scale," "3PL vs in-house fulfillment for a 500-order-a-day store." These resist summarization because the answer is conditional on the reader's situation, and the reader knows it. They arrive already knowing there is no one-line answer, so they click to do the reasoning. This is the spine of a decision-intelligence catalog, and it is exactly where an AI summary sends people deeper rather than ending the session.

Transactional and product-adjacent content. Category pages, buying guides tied to specific SKUs, product pages themselves. A buyer comparing two espresso machines with a $600 basket is not satisfied by an AI paragraph — they need to see, configure, and buy. This content sits closest to revenue, which is the whole point: its click has a purchase attached, so its cost per acquired customer is the lowest in your library.

Proprietary-data and original-analysis content. Your own benchmarks, cost breakdowns, teardown numbers, and hard-won operating data. This is the one category that gets more valuable under zero-click, not less — for a counterintuitive reason. When an AI summarizes a definition, it needs no source. When it cites a specific number — "returns cost apparel sellers X% of revenue" — it has to attribute it, and it attributes it to whoever produced it. Proprietary data is the content most likely to be quoted and linked, and the visitor who follows that citation arrives late in their journey and close to a decision. Multiple 2025–2026 analyses (compiled by Omnibound from Semrush and others) put AI-referred visitors' conversion and value well above ordinary organic traffic precisely because of that late-journey intent.

Where should the content budget go instead?

Here is the reallocation framework — the spine of this piece. Score every content type on two axes: zero-click exposure (how completely can the query be answered without a click?) and conversion proximity (how close is the reader to a purchase?). Then apply the move.

The reallocation itself is a transfer, not a cut. Every hour you pull out of the "Stop" rows moves into three destinations, in this order of priority:

1. Down-funnel first. Move hours to comparison, buying-guide, and category content — the rows with the lowest cost per acquired customer. Because the click here has a purchase attached, this is the highest-return content hour you have, zero-click or not. If those pages do not yet convert the clicks they get, fix that before making anything new; a converting page is worth more than a new one, and conversion rate is only a vanity metric when you measure it without cohorts and margin.
2. Into proprietary data. Fund the original benchmarks and cost teardowns that make you the source. This is the content that earns citations in AI answers and the high-intent clicks that follow them. It compounds: once you are the number an AI quotes, you are quoted repeatedly. This is also the content that makes a dedicated GEO program pay later — the sequencing is laid out in the GEO-versus-SEO investment decision.
3. Toward paid, deliberately, where the math clears. If a chunk of your discovery genuinely evaporated and you cannot rebuild it organically fast enough, some of that budget belongs in paid — but only after you have run the real math of paid ads versus organic. Reallocating to paid without that math just swaps a free dead channel for an expensive one.

How do you run the reallocation without cutting too deep?

The failure mode is overcorrection — gutting the informational layer so aggressively that you lose the topical depth that makes your decision pages rank in the first place. Three guardrails keep the reallocation surgical.

Cut by cohort, not by category. Do not delete "all informational content." Pull the last 90 days of data and identify the specific pages with zero converting clicks and no internal-link value. Those are the ones to retire or consolidate. Keep the definitional pages that still feed your money pages.

Consolidate upward. A thin "what is a 3PL" post and a rich "3PL vs in-house fulfillment" decision page should become one authoritative page, with the definition folded into the top. You keep the topical coverage, lose the standalone dead page, and hand the AI a fuller thing to cite. This also positions you for the shift toward agentic commerce, where machine buyers read structured, authoritative content rather than a scatter of thin posts.

Instrument for the new reality before you cut. You cannot reallocate what you cannot see. Segment AI-referral traffic where analytics can identify it, watch unattributed-direct lift against content you have structured for citation, and — critically — track cost per acquired customer per content type, not per pageview. If you make product feeds and data machine-readable as part of this, you also feed the product side of AI discovery, covered in how AI product discovery actually works.

The through-line: zero-click did not kill content. It killed one kind of content and made two others more valuable. The operators who lose are the ones who keep funding the dead kind out of habit. The ones who win treat their content library like a P&L, retire the line items that no longer produce a customer, and move the money toward the purchase and toward the data only they own.

FAQ

Does zero-click search mean content marketing is dead for eCommerce?

No. It means one layer of content marketing — summarizable, top-of-funnel informational posts — has lost its click-through, while comparison, transactional, and proprietary-data content still work. The move is reallocation, not retreat. Judge each content type on whether its query can be answered without a click and how close the reader is to buying, then fund the types that still produce customers.

Should I delete all my old informational blog posts?

Not on instinct. Some definitional posts send zero clicks but still support money pages through internal links and topical depth. Cut by cohort: identify the specific pages with no converting clicks and no link value over the last 90 days, and consolidate the useful ones into deeper decision pages rather than deleting them. Reallocate effort; do not vandalize your topical footprint.

If AI answers reduce clicks, why does proprietary-data content get more valuable?

Because a summary of a definition needs no source, but a specific statistic must be attributed. When an AI cites a number, it credits whoever produced it — and the visitor who follows that citation arrives late in their journey, close to a decision. Original benchmarks and cost breakdowns are the content most likely to be both quoted and linked, which is why they are the one category to fund harder under zero-click.

How do I measure content ROI now that pageviews and clicks are misleading?

Switch the unit to cost per acquired customer, measured per content type. High-volume informational content often shows healthy impressions while producing no clicks and no customers — an effectively infinite cost per customer that pageview metrics hide. Segment AI-referral traffic where you can, watch unattributed-direct lift, and reallocate toward the content types with the lowest cost per acquired customer.

Is this the same as deciding whether to invest in GEO?

No, and conflating them wastes money. GEO is a channel-timing decision — when to fund a program that optimizes for AI answer engines, covered in the GEO-versus-SEO investment decision. This is a portfolio decision — how to redistribute the content budget you already spend across content types now that most searches end without a click. Get the portfolio right first; the GEO program is a later overlay on top of it.