A content-marketing workflow for Shopify blogs that actually scales

Most Shopify store owners treat the blog as a thing they “should” do, half-heartedly, between everything else that pays the bills. That’s a fair instinct — Shopify’s article editor is buried in the admin, the SEO surface is shallow, and publishing one article a month into a 200-article archive that nobody maintains feels like throwing pennies in a well.

But the stores that take the blog seriously — the ones whose category pages rank because their cluster of supporting articles ranks — almost all run a loop that looks roughly like this:

keyword research → outline → draft → review → publish → measure → refresh

The part where Shopify’s native tooling falls down is “publish” and “refresh,” because the article editor is one article at a time and the rest of your stack lives outside of it. Specter is the piece that closes that loop. Here’s the whole workflow with Specter slotted in where it belongs.

1. Keyword research and topic shortlist

Use whatever you already use — Ahrefs, Semrush, Search Console, Google Keyword Planner, or a spreadsheet of competitor articles. Output is a shortlist of topics, primary keywords, intent, and target URL slug.

This step is yours. Specter doesn’t do research. The reason to mention it: keep the shortlist as a markdown file in your articles folder (or a sibling folder) so the AI you’ll use later can see what’s queued, what’s published, and what’s been refreshed. Context wins.

2. Outline in markdown, locally

Open a new .md file in your editor of choice. Write the outline by hand or have an AI generate one based on the keyword and the competing articles. The outline is just headings (##, ###) and one-line summaries underneath each.

Doing this locally rather than in the Shopify admin matters because (a) the file already lives in the folder Specter syncs, so it’ll publish itself when ready, and (b) your AI tool of choice has the full context of every other article on the blog when it helps you outline.

3. Draft — your hand plus AI assist

This is where most “AI content” workflows go wrong. They generate a whole article with one prompt, the writer skims it, it ships. Six months later you have an archive of articles in nobody’s voice that all sound vaguely the same.

A better pattern, and the one the hands-on Claude guide walks through, is to use AI surgically inside a draft you’re shaping:

• You write the intro because the intro is the article’s voice.
• AI fills in supporting sections from your outline.
• You rewrite the AI’s sections in your voice.
• AI tightens the prose, suggests internal links to other articles in the folder, and drafts the meta description.

The folder context is what makes this work. The AI can see every other article on the blog, so its internal-link suggestions are accurate and its tone stays consistent across the archive. That’s something a one-article-at-a-time generator inside the Shopify admin can’t do.

4. Review and approve

Read the draft top to bottom in your editor. Read the diff if the AI made the last pass. Fix anything that sounds robotic. Check the frontmatter: title, handle, blog assignment, tags, SEO title, SEO description, feature image URL.

This is the step that protects you from publishing slop. Don’t skip it. The whole reason Specter is a manual two-way sync rather than an auto-publisher is so this step exists.

5. Publish via Specter

Open Specter and run the dry-run preview. You’ll see exactly which articles in the folder will be created or updated on the next push. For a single new article it’s obvious; for a batch of five or ten ready-to-ship drafts, the preview is the safety net.

Push. Specter streams the changes back through the Shopify Admin API in the background. Long batches don’t time out the way ad-hoc API scripts do. Articles land on your store blog with their full frontmatter intact — handles, blog assignment, scheduled date, SEO fields, the lot.

If you’ve set scheduled dates in the frontmatter, Shopify takes it from there.

6. Measure

Search Console, GA4, whatever you use. Watch the article over 4–12 weeks. Note which articles climb, which stall, which get clicks but no engagement.

Specter doesn’t do measurement. Tag the articles in their frontmatter with whatever taxonomy is useful to you — “published-2026-q1,” “category-launch,” “review-piece” — so later passes can filter the folder by tag without you having to remember which is which.

7. Refresh the back catalog

This is the step that almost nobody does and that has the biggest SEO upside.

Every six months, sweep the archive. For each article older than 12–18 months:

• Are the facts still right?
• Do the internal links still resolve?
• Does the meta description still match what’s in the body?
• Is there a newer keyword pattern that should be in the title?
• Has the product or category page it points to changed?

The Shopify admin makes this miserable — you’d open every article one by one. With Specter, the back catalog is a folder. Hand it to Claude with a refresh prompt, read the diffs in your editor, run the dry-run preview, push. A sweep that would have been a week’s work becomes an afternoon.

The same machinery handles the bulk SEO sweeps that don’t need a full refresh: bulk-rewrite SEO titles and meta descriptions across every article when you change your title pattern, or regenerate missing descriptions across the archive in one pass.

If you also want a hands-off content backup as a side effect of all this — every article on your disk, every version in Git if you commit the folder — that’s covered in the backup guide. It’s the same machinery; you’re just keeping the files.

What changes when the loop closes

A few things start to compound when “publish” and “refresh” stop being the bottleneck:

• Cadence stops being constrained by tooling. If your bottleneck was the Shopify admin, you can publish three articles in the time it took to publish one.
• The archive stops rotting. Refreshing 30 articles in an afternoon makes it a thing you’d actually do twice a year, instead of never.
• The blog starts to feel like one body of work. Internal links are consistent, tone is consistent, meta descriptions are consistent — because your AI tool saw the whole folder while it edited.
• You own the content. Plain markdown on your disk. If you ever leave Shopify, your articles come with you in a format every other CMS can ingest.

That last one matters more than people think on day one. Run this loop for two years and the folder of .md files is one of the more valuable assets the store has.

The mental model

Shopify is your store and your publishing destination. Your Mac holds the working drafts and the canonical archive as markdown. Your AI does the heavy text work. Specter is the part that closes the loop between “I have a folder of drafts” and “they’re live on the store blog” — and, six months later, between “the archive is stale” and “the archive is fresh.”