Content Triage AI 2026: When to Create Content, When to Stop (Framework + Aleyda Worksheet)
Summarize this article with AI
The Problem: Your Content Exists, But It No Longer Gets Clicks
I review 15 sites per week. They all have the same problem.
Articles collecting dust. « Complete guide » pages that bring in nobody anymore. Entire blogs built to capture keywords that today generate barely 10 visits per month.
The culprit isn’t SEO. It’s AI.
Search engines answer directly in the interface. No need to click. The user gets their answer and never goes further. Your content lives in a traffic-less bubble.
Aleyda Solis put precise numbers on it. In her analysis of 40 leading sites across 4 industries, clicks from AI answers represent a very small share of total organic traffic. And crucially, the pages cited in AI answers aren’t the same ones receiving visits.
In other words, a page can shine in AI (citations) without generating a single click. Another can bring valuable visits without ever being cited. Two different jobs. Two different mechanics.
Yet most e-commerce brands keep creating content the old way. They spend thousands on « quick explanation » articles that AI absorbs without ever sending traffic.
What’s missing? Real triage. A framework to decide which content to create, which to stop, and which to consolidate.
What Aleyda’s Analysis Changes in How We Think About Content
Aleyda doesn’t just observe the phenomenon. She proposes a decision framework I tested immediately on my own clients. Her premise: AI completely satisfies the user when they’re seeking a quick explanation, a definition, a date, a raw fact. In contrast, AI doesn’t replace a website when the user needs proof, up-to-date data, tangible comparisons, contextual guides, or a place to take the next action.
The idea is to map your content against two axes: click value (does the user still need to visit your page after seeing the AI answer?) and citation value (will your content be picked up as a source by AI, other sites, publishers?). The third dimension is business value: does the page directly support conversion, retention, or support?
This triage changes everything. Brands often confuse brand awareness with traffic. They imagine a page generating citations will mechanically send traffic. That’s wrong.
Aleyda says it clearly in her analysis:
« The pages receiving AI referrals and those cited in AI answers are not always the same and serve different jobs. »A proprietary study can get dozens of citations without sending a single visit. A pricing or checkout page can receive qualified visits without ever being cited.
From this finding, she built 6 criteria to evaluate each piece of content. I’ve applied them to 14 projects since April. Result: 70% of « informational » content fails the test.
Avant de décider si créer ou supprimer du contenu, testez-le sur les 6 axes du framework Aleyda. Le radar ci-dessous vous permet d’évaluer rapidement un page existante ou en projet : cliquez sur chaque critère pour en lire la définition complète.
Les 6 critères de viabilité du contenu
Évaluez votre contenu sur les dimensions clés de la triage AI
The 6 Criteria to Decide If Creating Content Is Still Worth It
Aleyda's framework rests on 6 dimensions I'll detail here as I've learned and applied them. Each existing or planned piece of content gets scored across these axes. You don't need a scientific score; a rough estimate is fine.
These 6 criteria aren't an algorithm. They're your compass. In practice, you adapt them to your audience, priority keywords, real performance, and business model.
Voici comment le framework s'est concrétisé chez un client e-commerce spécialiste du matériel de rando. Sur 184 pages analysées, nous avons supprimé les articles sans résilience aux réponses IA et consolidé ceux à fort potentiel. Le résultat parle de lui-même.
Du triage à la croissance: cas e-commerce réel
Suppression de 48 articles = +31% de sessions en 3 mois
Client Case Study: 48 Articles Deleted, +31% Organic Sessions
One of my e-commerce clients, a hiking gear specialist, handed me 184 pages of informational content. Guides, comparisons, technical specs, definitions.
I pulled out Aleyda's framework. I scored each URL across the 6 criteria.
97 pages had near-zero click resilience. AI answered the question without any click needed. Example: "what is an IPX4 protection rating." 2,700 words to explain a definition. Result: 4 organic clicks in 3 months. The content was fine but not unique. No brand mention. No business value. No proprietary advantage.
Conversely, 11 pages of proprietary weather data collected by their field sensors scored well on citation potential, brand mention, and proprietary advantage. These pages sent only 320 visits a month, but were consistently cited in Google AI answers and outdoor press articles. Their job wasn't clicks—it was brand mention.
I applied the framework's logic. 48 articles deleted. 23 articles consolidated into 7 pillar pages focused on decision support and proof. 19 pages kept unchanged because they filled a clear job. All in under 6 weeks.
3 months later, the site's organic traffic was up +31%. The deleted pages had zero negative impact. They were generating nothing already. Consolidation focused authority on remaining pages, boosting visibility on transactional and decision-making queries.
The gain didn't stop there. 2 proprietary studies launched from existing data. Cost? 14,000€. Citations earned since generated 7 specialist media backlinks and 1,200 brand mentions in the following year. No ad budget. Just content architecture aligned with AI usage reality.
Tous les types de contenu ne souffrent pas également des réponses IA. Voici comment se répartissent les scores de résilience et de citation selon le profil de page. Les types en bleu résistent bien aux IA; en orange, c'est risqué.
Résilience à l'IA par type de contenu
Lequel vaut vraiment la peine d'être créé en 2026?
Which Content Types to Prioritize? Those That Resist AI
Cross-referencing the 6 criteria, Aleyda maps the content worth prioritizing. Not all play the same role, and that's fine. Some shine in citations, others in clicks. What matters is they fill at least one job clearly connected to business.
Here are the families I now use with my clients, with directional scores I adapt per site.
1. Brand and Entity Pages
Role: establish canonical facts about the organization. What it is, what it offers, who owns it, where it operates, why it's trustworthy. These pages aren't for pure traffic but for entity clarification and authority verification. Medium click resilience score, high citation potential, very high brand mention.
2. Proprietary Data and Original Studies
Role: provide unique, current information no one else has. Market analysis, sensor data, exclusive surveys. Click value is often low, but citation value and brand mention are huge. This is content that gets picked up by press and AI. It demands effort but blocks copycats.
3. Decision Support Tools and Personalized Content
Role: let the user configure, compare, simulate. AI can't deliver a personalized answer with the same constraints and proof. A product comparator pulling real-time inventory, for instance. Very high click resilience. Direct business value.
4. Pages That Enable Action Completion
Role: checkout, customer account, quote, booking. AI can't make you pay. The user must come to your site. Maximum click resilience. Citation potential is zero, but that doesn't matter.
These 4 families now cover 84% of profitable organic traffic for my e-commerce clients. "Complete guide without proof" content falls off the radar.
The Worksheet: How I Use It With Clients
Aleyda offers a downloadable worksheet you'll find on her article. Here's how I apply it in direct audits.
Step 1. Export all indexed URLs
I start with a Search Console export filtered on the last 12 months. I isolate pages that generated less than 2 clicks per month, and those that lost over 70% of their traffic after AI Overviews appeared on their main query.
Step 2. Score each page across the 6 criteria
No need for complicated spreadsheets. 3 columns do it: URL, content type, qualitative score per criterion (low / medium / high). The goal isn't perfection—it's fast triage.
Step 3. Bucket into 3 Groups
• Delete or deindex: pages with no clicks, no citations, no brand mention, no business value. They consume crawl budget and dilute authority.
• Consolidate: pages covering the same topic superficially. Merge them into one pillar page rich in proof, data, interactive elements.
• Invest: pages with strong citation potential, brand mention, or direct business value. Add proprietary data, field expertise, demos.
Step 4. Validate With Real Numbers
Compare the triage against performance: engagement rate, assisted conversions, backlinks generated, citations spotted via brand mentions. The framework gives direction; real results validate or adjust.
In 2 hours of audit, I review up to 500 URLs. My clients see immediately what's worthless and where to invest. The decision is never about belief. It's about triage.
I'll show you which pages deserve your budget
We take 2 hours. Your Search Console export, your catalog. We apply Aleyda's 6 criteria live. You walk out with the list of URLs to cut, consolidate, and invest in. No presentation. Just an action plan.
Book a strategic call — 45 minFrequently Asked Questions
Can deleting pages cause traffic loss?
If the page gets no clicks and has no citation or business value, deleting it costs nothing in traffic. It frees crawl budget and authority for the rest of the site.
Should I follow Aleyda's worksheet exactly?
No. The scores give you direction and need adjusting based on your audience, priorities, and business model. It's a compass, not a verdict.
How long to audit an e-commerce site with this framework?
With Search Console export and quick scoring, I review 500 URLs in 2 hours and deliver the 3 groups: delete, consolidate, invest.
Which content types generate the most AI citations without clicks?
In-house studies, exclusive and verifiable data, original research. Their click resilience is low, but brand value stays very high.
Should I still write informational blog articles?
Only if they provide proof, current data, or decision support AI can't deliver alone. Pure definitions no longer pay off.
