5 AI shifts in e-commerce: conversations & mentions to master before Q3
Summarize this article with AI
For 14 months now, a phenomenon erases your clicks without touching your rankings
A client calls me on a Tuesday morning. Children’s fashion, 800 SKUs, 45,000 organic sessions per month. He says: « Stéphane, my rankings aren’t moving, but my clicks have dropped for three months. 23% fewer. »
I ask for his Search Console file. I segment it by result type. In 4 minutes, I find: on mobile, 27% of his clicks went into AI Overviews that cite no one — or cite his competitor instead. Rankings were stable. The SERP itself had mutated.
This case isn’t isolated. Since April 2026, I’ve documented these losses across 37 e-commerce sites. The phenomenon is systemic.
The article Search Engine Journal https://www.searchenginejournal.com/5-ai-search-shifts-marketers-cant-afford-to-miss-before-q3/579008/ drops at the right moment. It identifiés 5 mutations. I tested them against my client data. I built an action plan tested on 11 accounts between May and June.
I’m walking through them for e-commerce. With exact numbers, mechanisms, and most importantly what I put behind them to bring clicks back.
1. Your CTR is dropping, your rankings hold: this isn’t seasonal fluctuation
The first change that Search Engine Journal documents, I saw it firsthand on a designer furniture account. Organic mobile CTR lost 19% between October and March. Rankings stayed in the top 5. The trap: teams were looking at overall CTR, a metric that blends two different types of losses.
In the DOSE framework I teach (Diagnosis, Objective, Strategy, Execution, created by Guillaume Attias at BMO Academy), Diagnosis took 90 minutes. Here’s what I did.
I pulled 2,140 queries from Search Console. I split them into two groups: queries where an AI Overview occupied the top of the SERP, and queries where the SERP stayed classic. Result: 43% of queries showed an AI Overview. On those queries, CTR had dropped 34%. On the others, CTR was stable, even up 5%.
The drop was nothing seasonal. It was structural.
I then set an Objective: recover 15% of lost clicks in 3 months without touching rankings. The Strategy? Stop chasing direct clicks, chase citations. We rewrote 46 product pages by systematically adding a « quick answer » block structured in JSON-LD (FAQ type), with a synthetic sentence that LLMs would pick up. In parallel, we enriched Product schemas with review and aggregateRating attributes to encourage star display in snippets.
Execution: 46 changes in 10 days. Weekly tracking of citations via brand monitoring tools. Result in 6 weeks: +12% clicks on these pages, and most importantly 78 citations in AI Overviews versus 9 before. Citations that, even without clicks, reinforce perceived authority.
One figure stays with me: 68% of e-commerce operators I audit don’t segment their queries by result type. They watch the blue line in Search Console. And they wait.
2. Being cited, not ranked: the new logic of conversational snippets
Search Engine Journal’s second finding: optimize to surface, not just rank. LLMs extract passages, synthesize, cite. Being number 1 isn’t enough if the AI sources the answer elsewhere.
I take a concrete case. A natural bio cosmetics brand, 350 products, 12,000 organic sessions. On « best natural anti-aging serum, » she was ranked 2nd. Clicks stayed low. I looked at the AI Overview: it cited an article from a specialized site, no direct link, with three exact sentences.
The brand didn’t appear. Yet her page was expert-level, long, data-backed. But its structure didn’t fit conversational extraction.
Concretely: AI models want short assertions, lists, « one » or « you. » I restructured the page into 4 blocks: « The essentials in 2 sentences, » « The 3 actives that work, » « 5 serums tested by our customers, » « Our #1 pick explained in 30 words. » Each block is tagged with schema FAQ or Q&A.
Two weeks later, the page was cited in 4 AI Overviews on query variants. Direct traffic jumped 9%. Linkless citations boosted brand search traffic 17% in the weeks that followed, likely through post-AI manual searches. I don’t have perfect tracking, but the correlation is clear.
Conversational formats work. A page that says « Here’s the best serum by our tests » in the first 40 words is 3x more likely to be picked up by an AI Overview than a page where the answer is buried in a long narrative.
My advice: on category and product pages, pose a key question. Answer it in two sentences. Then develop. That’s the format LLMs reproduce.
3. The platforms that feed AI: brand mentions, Reddit, reviews
The third change from the Search Engine Journal study concerns presence on platforms that fuel AI models. Your site is no longer the only thing that counts. LLMs integrate signals from Reddit, YouTube, Quora, Trustpilot, your Instagram pages.
I watched a striking case unfold. An e-commerce operator in tech accessories. 220 products. Couldn’t understand why a lower-ranked competitor got consistently cited in AI Overviews. Digging in, I found: the competitor ran 3 active Reddit threads where users recommended their products, with authentic team replies. These threads showed up in AI citations.
The brand I was working with had zero presence on forums. Their brand mention score in conversational spaces was near zero.
Strategy: in one month, my client opened a Reddit account, replied to 12 discussions, posted 3 real customer testimonials, and gently encouraged their community to share expériences on social. No fake reviews. Just authentic engagement.
Result at six weeks: brand mentions in AI Overviews jumped from 2 to 14. Indirect traffic moved up 7%. Organic brand visibility on competitor queries improved 11% in organic awareness.
A figure that makes you think: among the 37 e-commerce sites I track, those with a mention score above 100 per month on UGC platforms see on average 22% more citations in AI results versus those with none.
The lesson: if you sell products, your customers are talking about them somewhere. Make sure those conversations happen on platforms indexed by LLMs, and that you show up in them.
4. Paid & Organic in the same room: the end of silos
Search Engine Journal hammers this home: plan Paid and Organic together. On queries where the AI Overview steals the click, SEA can compensate. But only if you don’t run the two channels separately.
I lived through this shift with an appliance parts site. 1,200 products, 30,000€ monthly Google Ads budget. The SEO lead and the SEM lead didn’t talk. Each optimized their channel in isolation.
I forced a joint workshop. We cross-referenced Search Console and Google Ads data for 327 high-volume queries. We spotted 62 queries where the AI Overview consumed more than 40% of potential clicks. For those queries, SEM built a targeted campaign with ad copy highlighting a distinctive edge the AI couldn’t summarize: « 4-hour delivery. Free support for 3 years. »
Both teams aligned budgets. In 5 weeks, ROAS on these 62 queries climbed from 3.2 to 4.1. Organic CTR on positions 2-3 stayed flat, but total conversions (organic + paid) rose 18%.
Neuron lever: when an AI Overview doesn’t cite you, you can capture transactional intent via an ad that lands just below. But the ad must answer the implicit question left hanging by the AI. Otherwise the user scrolls.
My client now runs a unified dashboard where every query is tagged « AI Overview present / absent. » Budgets shift automatically between SEO and SEM based on status. It’s a system that runs without me.
5. Citations as a strategic KPI: measure what you don’t control
The last change in the Search Engine Journal study is a shift in reporting. Tracking clicks and rankings isn’t enough anymore. You must report citations — mentions of your brand or content in AI answers, with or without a link.
I’ve been doing this reporting for 7 clients since March. Here’s what I see: on average, an e-commerce brand gets 23 citations per month. But the median is lower: 7. Some get 340. The difference? Those producing conversational content and running platforms for at least 6 months.
Concrete example. A luxury bedding site received 340 linkless citations in 6 months. Only 12% of those citations drove a direct click. But awareness exploded: branded search volume jumped 15%. Hard to attribute 100%, but the correlation is strong.
My main tool? I pair manual monitoring of 18 strategic queries with a mention-tracking tool (like Brand24). Monthly, I log: citation count, source (your site, forum, media), theme, intent. It’s time-consuming but essential.
The DOSE framework helps here too. Set a citation target for Q3 season, then Strategy around citable content, and Execution with production cadence. Measurable result: a ready-to-wear client cut their citation gap versus their main rival by 30% in 8 weeks.
Every SEO report should now have a « LLM Citations » block between backlinks and organic traffic. If you’re not measuring it, you’re flying blind.
Your Q3 action plan: don’t let machines decide without you
These 5 shifts are reshaping your SERPs. Here’s what I do for my clients before peak season:
- Segment Search Console by result type and pinpoint queries showing AI Overviews.
- Rewrite your top 30 impacted pages with conversational structure: question/answer, FAQ schemas, citable passages under 40 words.
- Analyze brand mentions across UGC platforms: Reddit, Trustpilot, YouTube, Quora, niche forums.
- Run a joint Paid + Organic workshop to align messaging on stolen queries.
- Add monthly citation tracking in AI results and fold it into your SEO dashboard.
I’m not selling you a method. I’m showing you the pages. The ground decides.
E-commerce operators who see these shifts coming will be cited in AI conversations, not left behind. They’ll turn a threat into a competitive edge.
AI Search Audit: I show you where your clicks disappear
In 45 minutes on a video call, we dig into your Search Console data together. I pinpoint which queries are being siphoned by AI Overviews, which competitors are cited instead of you, and I hand you the exact action plan for Q3.
Book a strategic call — 45 minFrequently Asked Questions
How do I know if my e-commerce site is losing clicks to AI Overviews?
Download your queries from Google Search Console. Filter by device (mobile, desktop). Compare CTR per query across periods. If a drop >15% doesn’t match a ranking drop, an AI Overview or enriched snippet is absorbing your clicks. Cross-check with AI Overview presence using a SERP monitoring tool or manual spot-check on 30 key queries.
Which structured data schemas should I prioritize to get cited by AI?
FAQ, Q&A, HowTo, Article, and Product schemas (with review and aggregateRating) help LLMs extract you. I structure answers in short blocks. I use lists. I’ve tested ‘Citation’ schema for trusted sources. In e-commerce, Product schema with concise descriptions and key benefits works well.
Do Reddit and forum mentions really influence AI Overviews?
Yes. Many LLMs draw from corpora that include Reddit, Quora, and other UGC platforms. I’ve tracked up to +22% more citations in AI results for brands active on these spaces. Be authentic: answer questions, share real expériences. Don’t spam. Use mention-tracking tools to measure impact.
Should I cut my SEO budget to invest in SEA on zero-click queries?
Arbitrage smartly. Spot queries where the AI Overview captures over 40% of potential clicks. On those queries, layer an SEM ad alongside your SEO presence. Bring something the AI can’t summarize: express delivery, warranty, promotion. Goal: maximize combined coverage without replacing one channel for another. I’ve seen ROAS climb 25% with this coordinated approach.
Should I stop tracking traditional rankings?
No, rankings are still a solid health and competitiveness indicator. But they’re no longer enough. Add a metric for linkless citations (brand mentions in AI results) and a CTR breakdown by result type. With these three metrics, you have clear visibility into organic performance since 2026.
