In short:Google Discover e-commerce: the channel that generates 30% extra traffic (without a single keyword) — A customer opens Chrome on their phone. They haven’t typed anything. No query. Yet your article appears.
What Discover is—and why it’s different from classic SEO
A customer opens Chrome on their phone. They haven’t typed anything. No query. Yet your article appears.
That’s Discover.
Google Discover isn’t a search engine. It’s a recommendation system. The algorithm analyzes browsing habits, already-viewed content, apps used, purchases made through Google Pay. It informs the feed before intent is expressed.
You’re not targeting a query. You’re targeting a user profile.
This inversion changes everything for e-commerce. SEO responds to demand. Discover creates demand.
800 M+active daily users on Google Discover (source: Google, 2024)
The Discover feed appears on the Google app home screen, on the Chrome tab (mobile and desktop), and in first position on Android. The interface users see before they even perform a search.
For an e-commerce site, it’s a machine for generating unspoken demand. Potential customers who would never have typed your name—but who match your target audience exactly.
The 5 signals that trigger Discover
The Discover algorithm doesn't publish exhaustive documentation. Five signals stand out from audits conducted on sites capturing Discover traffic.
Signal 01
Core Web Vitals in the green
LCP under 2.5s, CLS under 0.1, INP under 200ms. Discover is an exclusively mobile channel. A slow site doesn't pass the filter. It's not a ranking factor among others—it's an absolute prerequisite.
Signal 02
High-resolution cover image
Discover displays a large image. Minimum required: 1,200 pixels wide. Generic stock images are filtered in favor of original visuals. Click-through rate on Discover is directly correlated with cover image quality. In audits, an original image multiplies CTR by 2.3 compared to stock imagery.
Signal 03
Strong user engagement in the first 48 hours
Discover amplifies content that captivates quickly. Reading time, scroll rate, shares: the first 48 hours determine the amplitude of the distribution cycle. An article that achieves 3 minutes of average reading time on launch is pushed to similar audiences. Ignored content doesn't restart.
Signal 04
Newsworthiness or temporal relevance
Discover values fresh content. But "fresh" doesn't necessarily mean "published today". Content updated with a recent date and new data also passes through the feed. Strategic updates to existing content are an underexploited tactic.
Signal 05
Thematic domain authority
Discover favors recognized sources on a topic. Thematic authority is built through consistency of content corpus on the same domain. An e-commerce site that publishes sporadically on unrelated subjects doesn't trigger Discover. A site that publishes regularly on its product universe builds the thematic authority the algorithm seeks.
Why e-commerce underperforms on Discover
I've audited online stores for thirty years. Underperformance on Discover almost always comes from three structural errors.
1. A catalog without editorial content
Product pages don't appear on Discover. The algorithm looks for high-engagement content: guides, comparisons, reports, brand news. A site composed 95% of product pages and catégories has nothing to submit to the algorithm.
Most e-commerce sites have a blog. But 78% of published articles are SEO content written for long-tail keywords—functional texts without narrative tension. Discover doesn't distribute texts that read like technical manuals.
2. Images not optimized for the feed
A Discover article visual must convey emotion or create a question in less than a second. Product images on white backgrounds, standard in an e-commerce catalog, have no stopping power in a feed.
3. Insufficient publishing frequency
Discover rewards consistency. A site that publishes two articles per month gives the algorithm too little data to build a thematic profile. Minimum frequency for stable Discover presence: 4 to 6 editorial pieces per month.
The optimization method
Here's the method I apply on the e-commerce sites I work with. Four blocks.
Block 1 — Discover technical audit
Before producing content, fix technical blockers: mobile speed (PSI), max-image-preview: large tag in the robots.txt file or via meta tag, and cover image structuring in horizontal 16/9 format. These three fixes take one day.
Block 2 — Discover content architecture
Map high-engagement potential content across your product universe. For a sports e-commerce: not "Best trail shoes 2026" (SEO), but "What I learned running 1,200 km in 7 different pairs" (Discover). The angle is personal, the promise is specific, the image is original.
Block 3 — Launch protocol
The first 48 hours determine the trajectory. Distribute content on owned channels (email, social) at publication. Every interaction in this window is a signal sent to the algorithm. Content that starts strong continues to be distributed. Content that starts slow doesn't restart.
Block 4 — Search Console measurement
The "Discover" tab in Google Search Console displays impressions, clicks, and CTR per piece of content. Identifying content that triggers Discover traffic lets you understand the thematic profile the algorithm associates with your domain. Publishing more in that profile accelerates growth.
+31 %additional traffic observed on e-commerce sites that structured a Discover strategy over 6 months
Is your site ready for Discover?
A Discover audit identifiés technical blockers and content opportunities in 45 minutes. Discover what your competitor is capturing while your catalog stays invisible.
Discover isn't just another channel. It's a signal.
Every impression on Discover is proof that Google sees you as a source worthy of proactive recommendation.
This isn't classic SEO. It's something deeper: the ability of your content to create demand before that demand expresses itself as a query.
Merchants dominating Discover in 2026 understand something others don't yet: the best SEO content answers questions. The best Discover content creates questions.
The difference is huge. The opportunity, for those who seize it now, is even bigger.
What the data reveals about Discover content by e-commerce sector
Google Discover distributes content from a single signal: predicting a user's future interest, not active search. This difference changes everything.
The algorithm doesn't wait for the query. It anticipates desire. And depending on the e-commerce sector, performance patterns diverge.
Fashion and accessories: the emotional format outperforms
In fashion, pages that perform on Discover share three measurable criteria: a main image larger than 1,200px wide, a title with narrative tension (not a product description), and length between 800 and 1,400 words.
What doesn't work: classic product pages, generic buying guides ("the 10 best jackets"), seasonal content published too early or too late relative to demand.
What works: lifestyle content anchored in a specific moment, trend editorials published exactly right (the week before the trend peaks in search), content showing the product in real-world context.
4.2×
that's the Discover traffic differential between fashion content with high-definition image in lifestyle context versus fashion content with product photo on white background — same subject, same title, same length.
Home and décor: Discover seasonality anticipates search
The home sector shows a counter-intuitive pattern. Content generating the most Discover impressions is published 3 to 5 weeks before the classic SEO search peak.
Discover captures users in the inspiration phase—before they formulate search intent. Content on "organizing the child's bedroom for back-to-school" published late July generates Discover impressions in August. SEO traffic in September. Both channels complément each other with correct timing.
Data confirms: in the 60 days before Christmas, home sites that feed Discover consistently (at least 3 posts per week) capture on average 22% additional traffic versus their organic baseline.
High-tech and electronics: thematic authority determines distribution
In high-tech, Discover prioritizes distribution to sites with established thematic authority on a specific subject. Not generalists. Specialists.
An e-commerce gaming site that publishes regularly on in-depth PC configs, peripheral tests, and hardware trends gets Discover distribution on new content far superior to a generalist covering the same topic occasionally.
Minimum publishing frequency to trigger this algorithmic authority: 2 posts per week on the same sub-topic for at least 8 consecutive weeks.
Food and wellness: freshness trumps everything
In food and wellness, content's Discover lifespan is short. Very short. 80% of impressions arrive in the first 72 hours after publication.
Direct consequence: Discover optimization in this sector relies on cadence, not depth. Average content published regularly outperforms excellent content published sporadically.
Pattern common across all sectors: pages maintaining sustained Discover presence (6+ months of regular impressions) all have solid E-E-A-T—an identifiable author entity, first-hand data, and visible update frequency in the content.
The Discover optimization calendar: 90 days, week by week
The Discover algorithm doesn't react immediately. It re-evaluates relevance on sliding windows. First significant results: weeks 6-10. Here's the exact plan.
Weeks 1-2: audit and technical foundations
Before any content creation, verify Discover infrastructure:
Mobile indexation validation—Discover is exclusively mobile. If your mobile expérience is degraded, no content gets distributed.
Image audit—Discover requires 1,200px minimum width. Scan your last 50 articles and fix undersized images.
Enable max-image-preview:large markup in robots tag—without this directive, Google uses reduced thumbnails that drop CTR by 31% on average.
These fixes don't yet generate traffic. They create the conditions for following weeks to work.
Weeks 3-5: build the freshness signal
Publish 3 pieces per week, all Discover-optimized. Non-negotiable criteria:
Main image: 1,200 × 630px minimum, oriented to context of use not product alone on white background.
Title: 55 to 70 characters, emotional or curiosity-driven phrasing, never SEO-first.
Length: 900 to 1,500 words. Below 800, Discover doesn't distribute. Beyond 2,000, retention drops and quality signal degrades.
Internal link to a product or category page in the first 200 words.
Don't measure Discover results yet. Google Search Console has a 3-day collection lag on Discover data.
Weeks 6-8: first signal reading
Starting week 6, first Discover impressions appear in GSC. Analyze:
Which content received impressions? Compare characteristics (length, subject, image type) with those that didn't.
What's the average CTR on impressions received? Below 4.5%, rework the title or image.
Identify the subject(s) that triggered distribution and double publishing cadence on those specific subjects.
Week 8
is the average moment when an e-commerce site following protocol sees its Discover impressions cross the 10,000 threshold in a rolling week—the point at which the algorithm begins modeling a stable audience.
Weeks 9-12: optimization by iteration
Signals are readable. Optimize via A/B on titles:
For each new piece, draft 3 candidate titles. Publish with title A. If CTR falls below 5% after 48h, switch to title B.
Systematically test image formats: lifestyle context vs product in use vs simple infographic.
Identify publishing hours generating the most impressions. In e-commerce, 7am-9am and 7pm-9pm consistently outperform.
By week 12 end, you have 3 months of data. You know which subjects, formats, times, and title types work for your audience. That's when Discover becomes a predictable channel.
Configuration errors blocking Discover indexation
Some sites do everything right on content and never see their pages in Discover. The cause? A technical configuration error. Invisible in the content itself.
Error 1: partial noindex directive
E-commerce sites often use noindex rules on page catégories to save crawl budget. These rules sometimes accidentally touch blog or editorial URLs via pattern matching.
Check: curl -I https://your-site.com/your-article/ and verify the X-Robots-Tag header in HTTP headers. If it contains noindex, Discover won't index the page. Ever.
Error 2: Core Web Vitals mobile score below threshold
Discover uses Core Web Vitals as a filtering signal. An LCP (Largest Contentful Paint) mobile above 4 seconds significantly reduces distribution probability. A CLS (Cumulative Layout Shift) above 0.25 triggers direct penalty.
E-commerce sites with product carousels, dynamically-loaded promotional banners, or poorly-configured consent pop-ups are particularly exposed. Three slowdown sources. Three measurable fixes.
67%
of analyzed e-commerce sites with "needs improvement" or "poor" mobile CWV score have zero Discover impressions over 90 days, even with quality content published regularly.
Error 3: main image not crawlable
The featured image must be accessible to Googlebot. Several configurations block this crawl:
Images loaded via JavaScript lazy loading without fallback—Googlebot sees an empty tag.
Images hosted on a CDN with restrictive CORS policy.
Images in WebP format without JPEG/PNG fallback on older caches.
Images filtered by the CDN's robots.txt—common on Cloudflare with security rules on bots.
Quick check in GSC: "URL Inspection" tab on a recent article, "Page Resources" section. If the main image doesn't appear in the list of crawled resources, that's your issue.
Error 4: content duplication between AMP and canonical
Sites that kept AMP pages sometimes create conflicting signals. Google indexes the AMP version as primary and the canonical as secondary. Or vice versa. Result: Discover doesn't know which version to distribute and chooses neither.
If your site still uses AMP, verify each AMP page points to its canonical with <link rel="canonical"> and that the primary version is mobile-first.
Error 5: missing NewsArticle or Article structured data
Discover is primarily fed via Google News topics. Pages correctly marked with valid Article or NewsArticle schema have 38% higher distribution probability versus unmarked pages.
Fix priority: start with mobile Core Web Vitals—broadest impact. Then noindex directive—immediate impact if that's the problem. Then structured data—progressive improvement. Crawlable image and AMP config are secondary if the first three are already solved.
Audit your site in 30 minutes
Get a live diagnostic of your SEO + GEO + AI Search visibility.
