AI Search: 5 Data-Backed Discoveries for SaaS E-Commerce Brands

84% of Citation Weight Doesn’t Belong to You

A client calls me. SaaS inventory management platform. « My content is solid, but ChatGPT doesn’t cite me. »

I check the metrics. 3 citations in ChatGPT over 6 months. All on competitor pages.

So what do I do? I flip the problem on its head.

Aleyda Solis’s study confirms it: for 15 SaaS brands analyzed, the weight of citations from outside the domain ranges between 84% and 93%. Plain language: your own pages make up only 7 to 16% of citation authority. Even in Finance & Accounting, where the internal percentage is highest (16.4%), the imbalance is massive.

Take the CRM & Sales category. 7.9% of ChatGPT citations come from the official site. 6.6% for Google AI Mode. The rest is third-party. Communities, peer publishers, comparison sites, forums.

A brutal fact. Even with a flawless blog, you’re invisible. The battle happens elsewhere.

I observe a natural reflex in my clients: produce more content on their own domain. Wrong move. Volume isn’t the bottleneck. Distributed proof is.

The signal AI models exploit relies on varied sources, not a single site. An ecosystem of proof convinces far better than your product page.

I built a map of the external ecosystem for this client: documentation, forums, detailed comparisons, YouTube videos, technical communities. In 4 months, citations doubled. He went from 3 to 7 in ChatGPT. And 12 in AI Mode.

Why? Because AIs find proof where it exists. Not just on your site.

The data from Aleyda Solis reveals a stark contrast in how ChatGPT and Google AI Mode build their proof ecosystems. This comparison shows which channels to prioritize for each platform.

ChatGPT and AI Mode Don’t Play by the Same Rules

This second discovery destabilizes SaaS e-commerce brands.

ChatGPT and Google AI Mode build their answers with radically different proof ecosystems.

Here are Aleyda Solis’s numbers:

• Peer software vendors (competing or complementary software publishers): 25.4% citation weight for ChatGPT, 16.1% for AI Mode.
• Communities & forums (Reddit, Quora, technical forums): 14.4% vs 11.4%.
• Video (YouTube): only 1% in ChatGPT, but 23% in AI Mode!
• Social (LinkedIn, Facebook, X): 2% vs 9%.

1% vs 23%. The ratio is 23 to 1. Video is nearly invisible in ChatGPT. Indispensable in AI Mode.

A client in e-commerce SaaS told me: « We made technical demos. But ChatGPT doesn’t mention them. » Of course not. ChatGPT doesn’t favor video. That content doesn’t fit its citation architecture. AI Mode, though, loves walkthroughs, tutorials, demos. It’s a citation magnet.

The lesson: don’t create uniform content. Segment. Structured text for ChatGPT. Video layer for AI Mode.

I’ll add an important point: in e-commerce SaaS, module integration videos, extension comparisons, and configuration demos are goldmines. AI Mode cites them massively.

Conversely, for ChatGPT, detailed text comparisons, technical documentation, and user feedback on Reddit carry weight. Don’t produce videos for ChatGPT. Create studies, integration pages, case analyses.

The game is different. You’re playing on two boards. With different rules.

The Two AIs Don’t Cite the Same Pages on Your Site

This is one of the most counter-intuitive discoveries in the study.

At a SaaS publisher, you’d think if ChatGPT cites a page, AI Mode will too. Wrong.

Aleyda Solis shows that each platform points to different zones of your domain.

Concrete example. For an e-commerce SaaS platform, ChatGPT often cites technical documentation pages, integration guides. Why? Because prompts tied to this SaaS type are oriented toward « setup, » « configuration. »

AI Mode prefers feature pages, pricing pages, comparison pages. It answers évaluation questions: « What’s the best SaaS for a booking site? »

A client in online payment SaaS experienced this firsthand. His « API integration » page made up 60% of ChatGPT citations. In AI Mode: almost nothing. His « recurring payment features » page was cited 4 times more by AI Mode.

The consequence: you can’t manage your AI presence with a single page inventory. You must audit citations separately for each engine, for each query type, and adjust content.

Here’s how I work: I extract URLs cited by ChatGPT across the 20 most important queries for the client. Same for AI Mode. I compare. Where there’s a gap, I build.

In this case, I created an intermediate page that addressed both integration and évaluation intent. Result: citations up on both platforms.

Pages That Gain Citations Aren’t the Ones That Capture Traffic

A striking figure: roughly 70% of pages cited by AI receive virtually no AI traffic.

Wait. How is that possible?

Aleyda Solis’s study distinguishes two things:

1. The page that serves as proof for the AI to formulate its answer.
2. The page where the user (or AI agent) is redirected after the answer.

They’re not the same.

Here’s how it works. ChatGPT cites a comparison on an external site. This comparison legitimizes the answer. But the user clicks the link leading to the app, configuration tool, or management interface. That page is what captures traffic.

I observed this phenomenon with a client in SaaS e-commerce. His « Shopify vs Our Solution » comparison page was the 3rd external citation source. But AI traffic landed 80% on his free trial signup page. These two pages had no direct connection.

So what do you do?

First, audit the post-citation funnel. For each citation, identify the actual landing page. Then align your strategy: strengthen the cited page to maintain the citation, but also optimize the landing page to convert AI traffic.

In this client’s case, I added a clear call-to-action on the comparison page toward the free trial, with specific tracking. Within a month, 15% of signups came from this path—invisible before.

AI traffic isn’t monolithic. It’s a two-step flow. Too many companies work on the first and forget the second.

No Universal Playbook: Your SaaS Has a Specific Job

Aleyda Solis’s fifth discovery is interesting.

There’s no single way to optimize for AI search. It all depends on the main job your SaaS category performs for the user.

Let me pull three examples from the study:

• CRM & Sales: the user wants to evaluate, compare, then log into their tool. The AI journey prioritizes comparators, reviews, and the login page.
• Project & Work Collaboration: the user seeks to configure an integration, understand workflows. Citations point to technical documentation, forums, and demo videos.
• Accounting & Finance: the user must solve a specific accounting problem. Support pages, tax articles, and usage guides are most cited. Visitors land directly on the feature.

For a SaaS e-commerce platform, the job is hybrid. Depending on whether it’s a sales platform, catalog management tool, booking system, or payment software, the AI journey will vary.

I see this with my e-commerce clients. A dropshipping platform needs to be cited in comparisons and integration guides. A payment SaaS should appear in articles on conversion rates, security, and recurring billing. Each category has its own proof sources.

Simple action: identify the 20 questions your prospects ask ChatGPT. Spot the sources cited for each question. Classify by proof type. Then invest where the gap with competitors is largest.

One of my clients, a booking SaaS tool, multiplied citations by 6 in 2 months by focusing only on comparisons with a detailed feature table. Why? Because that’s the content type ChatGPT valued in their category. Nothing else.

How to Apply These Discoveries to Your SaaS E-Commerce

Here’s how I work with each SaaS e-commerce client.

1. Audit current citations. I list the top 15 queries where the brand should appear in ChatGPT and AI Mode. I note the cited URLs (yours, competitors’, third-party).

2. Map the proof ecosystem. For each query, I identify the most-cited external sites. I build a list of 20 to 30 priority domains.

3. Platform-differentiated content strategy. For ChatGPT, I create or strengthen integration pages, detailed comparisons, documentation pages. For AI Mode, I develop a video layer of tutorials, demos, use-case stories.

4. Align citations with traffic. For each cited URL, I verify where AI traffic actually lands. I build a clear conversion path between the two. No gaps.

5. Adapt to your category’s role. I map the decision journey. What does the user expect after the answer? A demo? Technical docs? Free trial signup? I build the complete AI funnel.

This process isn’t theoretical. I apply it every week.

One SaaS e-commerce client, specialized in marketplace management, went from 5 AI citations to 18 in 3 months. AI traffic jumped +340%. Signups followed.

All that without writing a single blog post. Just by playing on the ecosystem, formats, and citation-to-traffic alignment.

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn