Blogs That Rank in AI Search: The Claude Workflow

Ask ChatGPT, Perplexity, Bing or Grok for the best coffee concentrate in Australia and they name my brand. So does Google. That ranking came from blog posts, and the blog posts came from Claude, trained with a skill that does the research, writes the post, inserts the tables and product links, generates the images and publishes to Shopify. Not AI slop. Posts that get traffic and turn it into sales.

This is the setup and the workflow. The skill file is linked in the video description.

Why these posts rank where normal posts do not

AI assistants cite content that answers a specific question directly, backs it with structure (tables, lists, steps), links to real products, and sits on a site that is clearly about the topic. Most blog posts are an idea someone had, written in one sitting. These are built the way an SEO agency would build them: research first, then the post. Claude does the research part that nobody has time for, every time.

Setup part 1: Connectors

In the Claude desktop app, open Cowork and go to Connectors, Manage connectors. You need:

• Shopify, so Claude can read your products and publish the post.
• An image generator such as Higgsfield, if you want images created. Claude does not generate images well on its own.
• Semrush or Ahrefs if you have an account. Not required, but keyword data improves the research.

Make sure they are enabled in the project you are working in.

Setup part 2: Install the blog writer skill

A skill is a stored set of instructions, rules and connectors. Once installed, you type /blog writer instead of rewriting a long prompt every time. Click the plus, choose Skills, Manage skills, Create skill, Upload a skill, and upload the file. If "blog writer" does not appear in the skills list, restart Claude.

The workflow, start to finish

1. Run the skill. Type /blog writer. Claude reads the skill, logs into your Shopify store, learns your products and existing posts, then asks what you want: research and rank ideas, one finished post, a chosen topic, or a batch of posts.
2. Answer the brief. Which product should the post drive clicks to (I chose our latte kit). What to do when the post is done: files only for review, draft in Shopify, or publish live. Post style: how-to, comparison, Q&A, listicle, or let the research decide.
3. Let it research. It searches the web, reads competitor posts, checks Reddit discussions, pulls keyword data, and reads your store again. Then it returns ranked ideas with a confidence rating. Mine: "How to make a latte at home without a machine" rated high, plus four alternatives.
4. Pick one. Type the number. Claude writes the post (mine was 1,175 words with a how-to structure and a five-question FAQ), fact-checks claims against your product pages and corrects them, builds the HTML, and runs an SEO and AI-search checklist.
5. Review the draft. It gives you a markdown file to read and an HTML file for Shopify. Read it. If anything is off, say so: "change this, change this". It edits.
6. Add images. I asked for "one or two editorial images, using the real product image inside the scene so people see it in use". Higgsfield produced them and Claude inserted them. No image tool? Ask Claude for prompts and use another generator.
7. Publish. "This looks great, publish it to Shopify." Approve the permission prompts. The post goes live with tags, author, internal links to products and related recipes, the comparison table, steps, images and FAQ, styled by your theme. Check it on mobile.

Put it on autopilot

One post is good. One a week, forever, is how you rank. In Cowork, type:

Create a scheduled task that comes up with a different idea every week, writes the blog post and posts it to Shopify.

Set the day and time (Monday 9am for me). Choose draft rather than publish live until you trust the output, and tell it to label AI-generated images. Click schedule. It now appears under scheduled tasks and runs without you.

What makes the posts sell, not just rank

• Every post is written to drive clicks to a specific product, and links to it in context, not just at the end.
• Comparison tables and step lists give AI assistants something to quote.
• Claims are checked against your own product pages, so the post never contradicts your store.
• Internal links to other posts and products keep readers on the site.

For the principles behind this, read how SEO content reduces ad spend. For running the rest of the store the same way, see how I run 80% of my store with Claude. Members get the blog skill and our other marketing skills inside EcomOS.