Skip to main content

Playbook 8: Clean Up Catalog Debt in One Session

How to eliminate the critical catalog warnings and poor scores that are silently reducing AI agent confidence across your entire catalog — in a single focused session.

Most catalogs accumulate debt over time — products added without alt text, categories set at one level instead of a full taxonomy path, metafields left blank. Each individual gap is small. Together they add up to a catalog that AI agents approach with low confidence. This playbook shows you how to clear the most damaging debt in a single focused session.

Goal: Zero critical warnings remaining and at least 80% of active products scoring "Mediocre" or above — with your worst products cleared out of "Bad" and "Poor".

Time to complete: 2–4 hours depending on catalog size.

Features used: My Store → Catalog Guardrails · My Store → Catalog Health Tracker · My Store → Important Warnings · My Store → Product Listing Editor


Step 1: Configure Guardrails Before You Start

Before you start fixing, make sure your score is measuring the right things. Guardrails that are misconfigured will show false warnings — gaps that aren't actually gaps for your catalog. Fixing your score first means every fix you make during this session is a real improvement.

  1. Go to My Store and click the Catalog Guardrails button.

  2. Review each active scoring criterion. Adjust any that don't match your standards:

    • If you don't use tags for product discovery, leave tag validation off (it's off by default). If you do, set a realistic Usage Limit Threshold first.

    • Adjust the description word count thresholds (default: 300 minimum, 2,000 maximum) to match your actual standards — technical products may need more words than simple accessories.

    • Check the Accepted Category Types list only includes categories you actively sell in — otherwise flagging becomes noisy and loses meaning.

    • Use Apply Rules To to scope enforcement — e.g., Published products only, so draft work-in-progress doesn't pollute your warnings.

  3. Click Save settings and close the panel.


Step 2: Review the Damage

Catalog Health Tracker — showing the distribution of products across health tiers

  1. Click Refresh scores on the Catalog Health Tracker to get a fresh read.

  2. Look at the distribution bar. Note how many products sit in each tier — Excellent, Mediocre, Poor and Bad. This gives you a realistic target for the session. If 60% of your products are Mediocre, moving them all to Excellent in one session isn't realistic. Moving the Poor and Bad products to Mediocre — and a handful of Mediocre to Excellent — is.

  3. Set a concrete target before you start: "I'll clear all Bad products and resolve the three most common warnings."


Step 3: Work Through Warnings in Priority Order

Important Warnings panel — showing issue types with affected product counts and fix actions

The Important Warnings panel shows you which issues affect the most products. Work through them in this order — not alphabetically, not by product, but by impact.

  1. Expand the Important Warnings Detected section in My Store.

  2. Start with missing image alt text. This is the highest-impact issue for AI agents and the fastest to fix. Open each affected product's listing, go to the AI Suggestions panel, and click Apply on the alt text suggestion. The suggestion already has the product attributes in it — review it briefly first (use Edit to adjust if needed).

  3. Move to poor categorization / low category depth. For each affected product, open the Category field, click Suggest a better category and apply the AI recommendation if it looks right. Aim for at least four taxonomy levels — the minimum recommendation for agentic discovery.

  4. Address incomplete metafields. For each product with missing metafields, open the Metafields section and fill in the flagged gaps.

  5. Save thin descriptions for last — they take the most time but have less immediate impact than the structural fixes above.


Step 4: Fix Products One Tier at a Time

Product Optimization Opportunities table — filterable by health tier and showing Fix Now actions per product

Once you've cleared the top warnings, shift to the product table and work by tier.

  1. In the Product Optimization Opportunities table, use the Health filter to show only Bad products. These are the most critical — resolve them first.

  2. Click Fix Now on each Bad product. Review the AI Suggestions and apply the highest-impact ones. You don't need to fix everything — just enough to move the product out of the Bad tier.

  3. Switch the filter to Poor and repeat.

  4. If time allows, work into Mediocre. Focus on products with the most warnings (highest alerts count) — those have the most room to improve with the least effort per fix.

📌 Note: You don't need to get every product to Excellent. Moving Bad products to Poor, and Poor to Mediocre, still meaningfully increases agent confidence across your catalog. Progress matters more than perfection in a single session.


Step 5: Verify Your Progress

  1. When you're done fixing, click Refresh scores on the Catalog Health Tracker.

  2. Check the new distribution. Compare the Excellent/Mediocre/Poor/Bad counts to where you started.

  3. Check that the Important Warnings you targeted are now resolved (zero products affected).

  4. Note the products that still need work and add them to a follow-up list. One session won't clear everything — but it should clear the critical debt.


Related Articles

Did this answer your question?