AI Search Checklist: The Weighted Grid of September 2026
Summarize this article with AI
Why September 2026 Changes Everything for AI Search
I’m going to tell you something agencies hate to hear: an AI Search checklist without weighting is no better than a grocery list. You tick boxes, you feel good, and citations stay flat. Why? Because no criterion carries the same weight. Aleyda Solis updated her checklist on September 20, 2026, with a weighted grid and a worksheet. Forget the best-practice list. We shift to a prioritization system.
The difference? Each criterion has a weight. Each step answers a key question. Each action falls under one of the three layers of Aleyda’s framework: presence, readiness, impact. Steps 1 and 2 measure presence. Steps 3 to 10 diagnose and correct readiness. Steps 11 and 12 quantify business impact and close the loop.
I’ve been observing a reflex in SEO teams for months: they rush to content. They rewrite pages, add Q&A sections, hope to get cited. Meanwhile, rarely anyone measures whether the site is already visible in ChatGPT, Perplexity, or AI Overviews. Rarely anyone checks whether pages are extractable. And rarely anyone separates proof pages from click pages.
The September 2026 update breaks that reflex. It places measurement before action. It forces you to define the journeys and prompts you want to influence. It imposes a repeatable measurement protocol. Only then do you touch content. That reversal of logic is the fundamental change.
Aleyda’s worksheet goes out to the 45,000 subscribers of her SEOFOMO newsletter, but it reads like a work plan. Twelve steps. Three layers. Scores. Page roles. A 2×2 matrix for citations and clicks. Nothing is left to chance.
We move past talking about « optimizing for AI » in a vacuum. We talk about weighting efforts where they move visibility. I want to detail here the weight of each criterion, not yet another list of 40 tick boxes.
La checklist d’Aleyda Solis n’est pas une liste à cocher : c’est un pipeline structuré en trois couches. Voici comment les douze étapes s’enchaînent, de la mesure de présence jusqu’à la validation finale.
Les 12 étapes en 3 couches
Le pipeline complet, de la mesure de présence à la boucle de validation
Presence, Readiness, Impact: The Three Layers That Structure the Grid
Aleyda Solis’s checklist does not start with content. It starts with a three-layer architecture. The first, presence, answers a simple question: do you appear? The second, readiness, asks a tougher one: are you ready to be cited, extracted, recommended? The third, impact, is the business question: what does it bring in?
Steps 1 and 2 cover presence. Define the commercially important prompts and journeys. Measure your current presence on AI platforms. That’s the foundation. With measurement comes diagnosis. With diagnosis comes credible optimization.
Steps 3 to 10 cover readiness. That’s the big chunk. Gap diagnosis. Page extractability. Decision-support content. Citation and click roles. Entity and positioning signals. Third-party source ecosystem. Machine-readable business data. Market-by-market localization. Eight out of twelve steps. Durable visibility is built right there.
Steps 11 and 12 cover impact. Reporting with full transparency, with confidence levels. Recurring validation loop. You verify that the expected signal moved. Otherwise, you adjust.
This sequencing imposes hierarchy. You cannot work on third-party sources until you have first diagnosed why the brand is absent. You cannot report business impact until you have defined the primary presence indicator before action.
The real weight shows in the structure: readiness steps weigh more than presence steps, but they depend on measurement. It’s a funnel. Measurement first, diagnosis next, action last. Those who reverse these steps optimize in the dark.
Presence is measured. Readiness is corrected. Impact is proven. Three layers, twelve steps, one logic.
Steps 1 and 2: Measure Your Presence Before Touching Content
Aleyda’s first step: define the prompts and journeys you truly want to influence. Not every possible prompt. Not just brand queries. The commercial decisions that matter. For an e-commerce site with 800 references, maybe « what is the best alternative to X for a €200 budget ». For a service provider, « how to choose between two approaches to reduce delivery lead time ».
The classic mistake: wanting to influence the whole spectrum. You end up measuring only awareness queries. Transactional journeys fall through the cracks. Aleyda’s grid asks you to source the real language of the audience. Not keywords typed in Google three years ago. The phrasings used in conversations with AIs.
The second step requires a repeatable measurement protocol. Five presence indicators: is the brand shown, recommended, linked, winning in comparisons, described accurately? Each indicator is measured separately. Then you choose a primary indicator based on the business model. A content site will not track the same indicator as an e-commerce site.
I’ve seen teams skip this step. They optimize pages at random, then find the brand was already well positioned but poorly described. They lose six weeks. Measuring first saves time. It tells you where to act.
Concretely, Aleyda’s worksheet has you record prompts, platforms, typical responses, description gaps. You get a presence map. Not a fuzzy global score. A snapshot per journey. That’s the foundation of weighting.
If you don’t measure today, start small. Seven representative prompts. Three platforms: ChatGPT, Perplexity, Google AI Overviews. Track responses once a week. You’ll see patterns. That’s enough to launch the checklist.
Sur les douze étapes, huit tombent dans la couche readiness. C’est précisément là que se joue la pondération — et donc l’essentiel du travail d’optimisation.
Où se concentre le poids de la checklist
Répartition des 12 étapes sur les trois couches d’Aleyda Solis
Steps 3 to 6: Where the Real Weighting Happens
Steps 3 to 6 carry the most weight in readiness. Gap diagnosis, extractability, decision-support content, citation and click roles. That’s the heart of Aleyda Solis’s weighted grid.
Step 3: diagnose before correcting. Aleyda lists ten readiness characteristics. Page accessibility. Content extractability. Entity clarity. Naming and category consistency. Presence of third-party evidence. Structured business data. Localization. For each, you map observed patterns into hypotheses. You prioritize with a score. The key question: which characteristic best explains the observed pattern?
A concrete example. A financial services site never appears in Perplexity responses. The content is good. Internal linking is fine. But the main content is rendered in JavaScript after loading. Answer engines don’t isolate it. The diagnosis points to extractability, not editorial quality. That’s Aleyda’s step 4: are your pages accessible and extractable?
Step 5: build decision-support content, and prioritize what still has a role after the answer. Aleyda has you score six directional dimensions: need coverage, granularity, comparability, evidence, format, positioning. You don’t copy the page mix from another category. You deprioritize, consolidate, or improve certain content.
Step 6: separate citations from clicks. A page can earn citations without generating clicks. Another can generate clicks without being heavily cited. Aleyda asks you to act on the 2×2 matrix, not on a single « AI visibility » score. Proof pages feed citation. Destination pages capture the click. Both roles are measured separately.
The weighting of these steps is massive. A correct diagnosis saves months of useless optimization. An extractable page beats every rewrite in the world. Separating roles avoids sacrificing destination pages to proof logic. This is where Aleyda’s checklist becomes a system.
Steps 7 to 10: The Signals Everyone Underestimates
After content, Aleyda places four steps many teams forget: entities, third-party sources, business data, localization. They carry heavy weight in AI citations.
Step 7: align entity, naming, and positioning signals. AIs rely on explicit entities. Your company profile, product names, catégories must be consistent across all pages. Facts must be explicit. No internal jargon. No ambiguous descriptions. Aleyda insists on the checks to run.
Step 8: work the third-party source ecosystem. Aleyda shows that composition counts more than ratio. Not a question of « backlinks » in the classic sense. It’s the diversity and credibility of the sources describing your brand, products, services. Classify the role of each source: proof, description, comparison. Score citation opportunities on six dimensions. Maximize formats that produce durable evidence. Avoid manipulative activities.
The September 2026 update adds a nuance: the platform mix changes the budget. You no longer weigh only the quantity of mentions. You evaluate contextual quality. A specialized press article can outweigh ten directories.
Step 9: make commercial and transactional information readable by machines. By business model, Aleyda asks you to clarify what must be explicit: price, availability, lead times, guarantees, terms. Product feeds, merchant programs, implementation checks. Structured data alone falls short. AIs read visible content and structured data together.
Step 10: localize by market, not just by language. Aleyda provides proof: word-for-word translation falls short. You need to adapt evidence, local sources, local entities, content formats. Test per market.
These four steps are not extras. They condition the credibility of your content. Excellent content poorly surrounded remains invisible. An inconsistent entity dilutes trust. Unreadable business data blocks transactional responses.
Steps 11 and 12: Reporting Without Overpromising and the Validation Loop
Most tools promise a single AI visibility score. Aleyda refuses that shortcut. Step 11: report without overpromising. Keep scale in view. Three layers: presence, readiness, business impact. Each layer has its own confidence level.
Presence is measured with first-party data accessible on AI platforms. Readiness comes from internal scores. Business impact relies on conversions, revenue, leads, with a confidence level: observed, inferred, probable. A sale attributed to an AI citation lacks the certainty of a direct click.
Aleyda asks you to configure the observable layer carefully. Which first-party data is available on ChatGPT, Perplexity, Google AI Overviews? Which events can you track without overinterpreting? Which signals serve as leading indicators rather than outcomes?
Step 12: recurring validation loop. After each type of change, you validate a specific signal. The worksheet lists what to validate: extractability modification, addition of support content, reinforcement of a third-party source. Cadence is defined. The minimal workflow is clear: measurement, hypothesis, action, validation.
This loop is what makes the checklist durable. You don’t optimize once and for all. You build a cycle. AI platforms evolve fast. Citation criteria change. Only a validation loop keeps you aligned.
Reporting without overpromising avoids internal disappointment. You show pages that become cited, journeys that gain clarity, clicks that arrive. A score increase alone tells nothing. There you have concrete results. More honest. More useful.
Applying the Weighted Grid to Your Site: The Action Plan
Aleyda Solis’s checklist, in my work, is a work plan. I apply it with my clients, adapting the worksheet.
Week one: define seven representative decision journeys. Record presence on three platforms. Choose the primary presence indicator based on the business model.
Week two: diagnose gaps across the ten readiness characteristics. Note each pattern. Formulate one hypothesis per gap. Prioritize with Aleyda’s score. Correct only once diagnosis is complete.
Week three: check extractability of priority pages. Make the main content accessible and isolable. Separate proof pages from destination pages. Deprioritize or consolidate content with no role after the answer.
Following weeks: align entities, reinforce third-party sources, clarify business data, localize by market. Then set up three-layer reporting and the monthly validation loop.
An e-commerce client with 800 references followed this sequence. The first month, no content overhaul. Just diagnosis and extractability correction. Two product pages ended up cited in comparative responses. Without a single rewrite. Order of magnitude I observe with my clients: gains come first from structure, not text.
Aleyda Solis’s weighting saves you from rewriting 300 pages into the void. It tells you where to act first. The 12 steps are not equal. Measurement and extractability weigh more than answer length. Third-party sources count more than synonyms. Honest reporting does more than a single score.
You can keep ticking boxes without hierarchy. Or weight your efforts where they move citations. Which of your ten most important pages is invisible in AI responses today?
AI Search Audit: I Weight Your Pages Live
I show you the pages that matter, the real weight of each criterion on your content, and the three actions that will shift your citations. The first call is a live audit, not a sales presentation.
Book a strategic call — 45 minFrequently Asked Questions
What’s new in Aleyda Solis’s AI Search checklist in September 2026?
It moves from a best-practice list to a weighted grid: 12 steps, 3 layers. Presence, readiness, impact. The worksheet asks you to measure before optimizing, and to separate proof pages from click pages. Prioritization changes. It’s far more than a simple update.
How do you weight the checklist criteria instead of treating everything equally?
I score each gap on the 10 readiness characteristics with Aleyda’s diagnostic score. One hypothesis per pattern. Then I prioritize extractability and page roles ahead of content. The initial presence measurement gives me the main indicator to track.
Where do I start if my site never appears in AI responses?
List seven typical decision journeys. Check whether ChatGPT, Perplexity, and Google AI Overviews cite you. Scrutinize the ten readiness points, starting with page extractability. Correct structure before touching content.
Does the checklist also apply to e-commerce sites and local content?
Yes. Step 9 lays out the business data machines can read, based on your business model. Step 10 goes beyond language: it addresses market-by-market localization. Product feeds, prices, local evidence are among the signals covered.
How long does it take to see AI Search results with this method?
The first extractability changes produce citations within a few weeks. Aleyda’s validation loop works in short cycles: measurement, hypothesis, action, validation. For business, I look at confidence levels. A single short-term score says nothing.
