Weekly Adobe Commerce SEO copy audit posted to Slack

Every Monday at 7am, run an SEO copy audit against my Adobe Commerce catalog and post AI-drafted improvements to Slack for merchandiser review. The workflow is an assistant to the merchandiser: it drafts and shares, it never writes back to Magento.

Step 1. Pull the catalog. Use List Products (adobe-commerce) with a searchCriteria filter to fetch only enabled products (status = 1) that are visible on the storefront (visibility != 1). Page through until the accumulated item count reaches total_count. Keep the fields needed for the audit: sku, name, meta_title, meta_description, meta_keyword, short_description, category_ids, and the key custom_attributes (brand, size, color, material) if present.

Step 2. Score every product for weak SEO signals. Flag a product if any of the following are true: missing meta_description, meta_description shorter than 60 characters, missing meta_keyword, generic or duplicated meta_title (matches the product name verbatim, is empty, or appears more than once across the catalog), or short_description shorter than 40 characters. Record which rules each product tripped.

Step 3. Rank by recent revenue impact. Use List Orders (adobe-commerce) with a searchCriteria filter of created_at from the last 30 days and status in complete, processing, shipped. Aggregate units sold per SKU from the order line items. Sort the flagged products so the best-selling offenders come first. Take the top 20; if fewer than 20 products are flagged, take all of them.

Step 4. Draft new copy for each of the shortlisted products, pulling from the existing product name, attributes (brand, material, size, color), and category names. For each product produce: a meta_title targeting 55 to 65 characters that leads with the product identity and includes one differentiator; a meta_description targeting 140 to 160 characters that summarizes the benefit, mentions a key attribute, and closes with a soft call-to-action; a tightened short_description under 300 characters that keeps the merchandiser voice, cuts filler, and front-loads the promise. Never fabricate specs the product data does not support.

Step 5. Post the recommendations to Slack with Send a Message (slack) to the merchandising channel. Format the body as a compact Slack mrkdwn table, one section per product, with: SKU and product name as the header, then the current meta_title / meta_description / short_description, then the proposed replacements clearly labeled, then a one-line note on which SEO rules the product tripped and its 30-day units-sold rank. Keep each entry scannable so the merchandiser can eyeball a winner in seconds and paste it into the Magento Admin. If the payload is too long for one message, split across multiple posts in the same thread.

Guardrails. Do not call any Adobe Commerce write operation. This workflow is read-only against Magento by design so the merchandiser stays in control of the storefront copy. If there are no flagged products in a given week, post a short Slack note saying the catalog passed the audit.