AI Search Checklist: The URL-by-URL Worksheet

The September 2026 checklist: everyone has it. Nobody checks it.

I review 15 sites a week. All of them have downloaded Aleyda Solis’s checklist. Few have turned it into a URL-by-URL sheet.

The September 26, 2026 update is circulating. E-commerce teams read it. They save it in a Drive. Then nothing.

The block comes from execution, not from the grid. The weighted checklist remains a document. What’s missing is the operational worksheet: the page, the score, the action, the checked box.

Aleyda Solis built 12 steps, not 12 boxes to check mentally. The first two measure presence in AI engines. The next ten diagnose and fix readiness, content, extractability, entities, corroboration, commercial data, and localization. The last two validate impact. It’s a diagnostic protocol that runs over time. Not a one-off audit.

For an e-commerce team, the question becomes simple. Which URL moves the needle on the prompt that triggers a purchase? We can no longer think in terms of « site. » An 800-page site may well have only 40 URLs that matter for AI. Each of those 40 URLs needs a score, a priority, an owner.

The DOSE framework, as taught by Guillaume Attias at BMO Academy, helps explain why teams disengage. Dopamine comes from visible progress. Checked boxes. Endorphins come from the relief of complexity turned into steps. The URL-by-URL worksheet delivers both levers.

A checklist remains a library of good intentions until it becomes a roadmap. And a roadmap generates citations in ChatGPT Search.

I’ll show you the sequence. First the URL column. Then the score. Then the priority. Finally the checked box.

The worksheet starts with the URL column, not the score

The worksheet starts with a column, not a score.

You duplicate Aleyda Solis’s Google Sheet. You add a URL column on the left. Each row maps to a specific page. Not a topic.

For an e-commerce catalog, that changes everything. A product page plays a different role from a category page. A « buying guide » page plays a different role from a « comparison » page. Aleyda Solis’s checklist focuses on specific pages for specific prompts, not « the site. »

Aleyda Solis recommends 30 to 50 core prompts. That’s the foundation. The prompts cover priority products, audiences, journey stages, markets, and languages. Each prompt is tagged. Experimental prompts stay separate.

Then you cross those prompts with your URLs. A page can answer 1, 3, or 7 prompts. Only URLs that appear in the cross-reference earn a place in the worksheet. We check boxes for a reason: a page has a role in an AI-assisted purchase journey.

This simple row turns a checklist into a work plan. The team stops discussing the method. It discusses the page.

And that’s where complexity falls away. A URL column, a prompt column, a score column. Three columns are enough to start. The rest comes later.

Score each URL with three filters: presence, accuracy, click potential

Scoring a URL means measuring three things: presence, accuracy, click potential.

Aleyda Solis repeats it: a measurement protocol keeps everything solid. Each prompt runs 3 to 5 times over a 24 to 72 hour window. I document the platform, market, location, language, account state, and model. Otherwise scores shift from one day to the next and lose meaning.

We record six types of result: appearance, recommendation, link, comparative preference, precision, sources, competitors. That’s richer than « we’re there. »

Then we convert those signals into a score per URL. I use a simple scale: 0 for no mention, 1-3 for mention without recommendation, 4-6 for recommendation or link, 7-9 for link + won comparison, 10 for exact citation with attribution.

The trap is looking only at appearance. A page can be cited for a comparison while the competitor is recommended. The score must capture that gap. A URL scored 9 because it appears often, but is never recommended, misleads the team.

Each URL now has a score. The worksheet moves from inventory to diagnosis.

The third filter, click potential, depends on how the answer is phrased. The more the page provides a direct answer, the more likely it is to be cited as a source. That’s the extractability work, which Aleyda’s checklist places in step 4.

Prioritize: Severity × Importance / Effort, not the urgency of the last call

A raw score doesn’t tell you what to fix first. Priority is calculated.

Aleyda Solis uses a simple formula: Severity × Importance / Effort. Severity measures the gap between the current score and the expected score. Importance is the commercial value of the prompt for the page. Effort is time and resources.

On an e-commerce site, a high-margin product page with a score of 2 on a transactional prompt is a P1. A blog post with a score of 6 on an informational prompt is a P3. The score isn’t everything. Commercial potential often carries more weight.

The worksheet must include a Priority column and an Owner column. Assign an owner to every fix. With my clients, owned fixes leave the « To do » column quickly; unowned fixes linger for weeks.

In practice, we only handle P1s the first week. P2s the second week. P3s once the rest is stable. This rule avoids scattering.

Relief comes from the end of complexity. A team with 17 P1s knows exactly what to do Monday morning. It has a clear list instead of 214 floating recommendations.

Prioritizing with Aleyda’s formula removes the noise. We fix what has the most impact for the least effort, not what’s most visible. That’s the difference between a worksheet and a wish list.

Track progress URL by URL: the table that creates dopamine

URL-by-URL tracking makes progress visible. That’s dopamine. The DOSE framework taught by Guillaume Attias at BMO Academy leverages exactly this mechanism.

Each worksheet row has a Status column: To do, In progress, Verified, Impact measured. A status change triggers a micro-satisfaction. The cell changes color, and that’s enough.

Relief comes from simplification. The original checklist has 12 steps, 10 characteristics, 3 layers. The worksheet brings it down to one question: which URL, which action, which status.

At one e-commerce client, we processed 64 URLs in 17 days. The shared table kept the pace. Every morning, the team saw 3 or 4 checked boxes. The question « where are we » disappeared.

Format matters. Checked boxes must jump out, with color. A gray table without formatting doesn’t do the job. Dopamine comes from contrast.

This mechanic is functional. It turns AI Search optimization work, often anxiety-inducing, into a series of finished steps. That’s endorphins: complexity reduced, and relief that follows.

A simple Google Sheet with 6 columns is enough: URL, Prompt, Score, Priority, Action, Status.

Checked boxes become the dashboard. Each morning, you see where you stand, without a meeting. Each URL moved from « To do » to « Done » proves the system is running. AI Search tracking is clarity, not noise.

From table to results: measure impact URL by URL, not in aggregate

A checked box isn’t enough. You need to measure impact URL by URL.

Aleyda Solis dedicates steps 11 and 12 to reporting. We replay the core prompts after correction. We compare the initial score with the one measured at 30 days. We note the gaps by platform and by market.

The classic trap: aggregating results. A global score rising from 4.2 to 5.1 says nothing about pages that stagnate or regress. The URL-by-URL worksheet shows where the effort paid off.

Example of tracking after 30 days:

Stagnant rows are signals. Either the page has an extractability gap. Or the prompt is lower commercial priority. Or the competitor moved faster.

Measurement must respect the initial conditions: 3 to 5 runs per prompt, over 24 to 72 hours, same account, same model. Keep conditions identical so variations stay meaningful.

When a URL’s score reaches 8 or 9 and stays stable, we stop optimizing. We move to the next one. That ends the perfectionism that blocks progress.

We leave vague aggregates behind. Each URL is tracked separately. We know what worked, what didn’t move, and why. That’s the level of detail an e-commerce team expects to validate an AI Search budget.

La checklist ne s’exécute pas toute seule. Voici le pipeline minimal, sept mouvements nommés, qui transforme la grille en worksheet opérationnel page par page.

What I’d tell a team stuck on the checklist

The deliverable is the worksheet. The checklist is only one step.

An e-commerce team rarely gets stuck from lack of knowledge. It gets stuck when a URL-by-URL sequence is missing. Aleyda Solis’s grid gives the logic. The worksheet gives the execution order.

The minimal workflow fits in a few moves: duplicate the Google Sheet, add the URL column, cross-reference 30 to 50 core prompts, score each page, prioritize with Severity × Importance / Effort, assign an owner, check off each week.

The format matters little. Visible progress keeps the team moving.

The next time you open an AI Search checklist, only one question matters.

How many of your priority URLs have an AI score this week?

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