E-commerce
June 28, 2026
You launch a collection, change a delivery time, remove a SKU. Your Shopify store is updated in 10 minutes. Your help center, your Gorgias macros, and your AI bot are still quoting the old version. This is the catalog / knowledge base drift, a silent cause of unnecessary tickets and bot hallucinations.
RAG tests show that the error rate can jump from 10 to 66% when the retrieved context is obsolete (Tianpan, obsolete RAG docs 2026). Gorgias is launching Opportunities to detect knowledge conflicts and gaps automatically (Gorgias, Opportunities 2026).
This guide #140 covers KB maintenance aligned with the Shopify catalog: trigger events, product launch workflow, helpdesk sync. It complements the KB structure and cleaning the corpus (#103) with a focus on continuous freshness, not the initial architecture.
Summary
Why the catalog / KB misalignment is costly for support?
When your knowledge base diverges from the live Shopify catalog, every channel amplifies the error: bot, agents, help center, email marketing.
Measurable Consequences
Preventable tickets: "the bot says out of stock, product available"
Useless returns: obsolete size guide on an old cut
Low CSAT: false response perceived as a lie
SEO help: indexed page with incorrect specs
Bot trust: 1 visible error = widget abandonment
DTC Example
Supplements brand, reformulation of star product (new dosage). PDP and Shopify stock updated Day 0. Help center and bot corpus: old dosage still cited Day +21. Peak tickets "can I take 2 capsules?": 47/week. Sync KB + product help metafield: 6/week within 10 days.
Convert over 2,000 customers on average per month with Qstomy.
The world’s 1st Shopify AI dedicated to customer conversion
Empowering 200+ e-commerce merchants
How does #140 differ from existing KB guides?
Three neighboring articles, three different questions.
KB Structure
Knowledge base structure: taxonomy, article templates, linking. #140 = post-build maintenance.
Corpus Cleanup (#103)
Clean up bot corpus: one-shot audit before AI launch. #140 = recurring rituals with every catalog change.
Shopify Training
Train Shopify chatbot: sources and initial sync. #140 = drift governance product, collection, promo.
Which catalog events trigger a KB update?
Map the KB maintenance triggers on the Shopify admin side.
Product events
New SKU / collection: help sheet, PDP questions inset, bot chunk
Specs change: dimensions, ingredients, compatibility
Discontinuation: remove bot chunks, redirect help, agent macro
Permanent OOS variant: update availability, suggested alternative
Price / bundle modified: promo items, stacking conditions
Cross-functional events
Changed return/shipping policy, new international market, pre-order launch, final sale clearance, carrier/delivery time change. Shopify Catalog propagates price and stock to AI channels in ~60 s; your narrative KB does not follow automatically (Shopify, Catalog 2026).
Mental webhook
Every product or policy save in Shopify admin should trigger a Slack #kb-review notification if the SKU is in the top 50% of support tickets. Automate via Zapier or internal flow if launch volume > 2/month.
How to define a single source of truth?
Without a source of truth, each team updates their own corner and drift returns.
Recommended DTC Hierarchy
Shopify /policies/*: returns, delivery, privacy (canonical legal)
Product Metafields: specs, care, compatibility (canonical product)
Gorgias Help Center: derived articles from policies + product hero SKUs
Qstomy Bot Corpus: RAG chunks synced from 1-2, no manual copying of PDP
Agent Macros: dynamic Shopify variables when possible
Golden Rule
Never rewrite a product spec in 4 different places. Help center summarizes and links to PDP; bot reads metafield or API; agent macro =
What workflow should be followed for each product launch?
Integrate the KB into the product go-live, not as an afterthought after launch.
Launch sequence (D-7 to D+3)
D-7: help metafields filled in (usage, size, compatibility)
D-5: help article "How to use [product]" drafted
D-3: 5 support questions anticipated from beta testers
D-1: bot chunk ingest + shadow test of 10 questions
D0: publish help + agent macro PROD-XXX
D+3: review unmatched tickets for the new collection
Complete launch support plan and misunderstood products.
Owner
Product owner writes metafields. Support owner publishes help + bot. Marketing notifies if PDP copy changes post-launch.
Clean discontinuation
SKU removed: deactivate bot chunk, switch help article to "product replaced by [X]", agent redirect macro, 301 redirect help URL to successor. Forgetting this step = ghost tickets on products that cannot be found at checkout.
How to manage seasonal collections and promotions?
BFCM and sales create the most violent KB drift peak of the year.
Before campaign
Promo help article: dates, exclusions, coupon stacking, final sale by collection. Shopify tag `final_sale` → bot rule rejects standard return. Macro PROMO-2026 with expiration dates. See promos and support, reduce promo tickets.
During campaign
15-minute daily review: top 3 unmatched bot issues related to promos. Fix corpus or macro same day.
After campaign
Deactivate expired promo articles (Hide in Gorgias, do not delete for SEO). Re-sync help URLs. Audit conflicts between 30-day return policy vs 14-day sale policy.
How to maintain Gorgias Knowledge and re-sync sources?
In Gorgias Knowledge, keeping content fresh requires targeted manual actions.
Audit Last updated
Filter Knowledge by the Last updated column. Prioritize articles older than 90 days on top-volume intents (returns, shipping, star product). Gorgias recommends drafting → publishing with a changelog (Gorgias, managing AI Agent content).
Re-sync URLs and site
If your KB ingests the Shopify site or help center URLs, trigger a re-sync after a policy change or PDP redesign. Auto-sync does not guarantee narrative consistency if multiple pages say different things.
Duplicate consolidation
Merge 3 "return" articles into 1
How to use Gorgias Opportunities to detect drift?
Gorgias Opportunities analyzes bot conversations to surface knowledge gaps and conflicts without manual ticket auditing.
Two types of opportunities
Resolve conflict: guidance says 14 days, help article says 30 days
Fill knowledge gap: customers ask about a new product with no chunk
30-min weekly workflow
AI Agent → Opportunities. Tackle the top 3 by affected tickets. Edit, disable, or delete conflicting content. Resolve. Opportunities do not update on their own if you have already fixed them elsewhere: dismiss if obsolete.
Limitation
Opportunities detects conversational symptoms, not always the catalog root cause. Cross-reference with the 7-day modified products export in Shopify admin.
Typical conflict example
AI guidance "free return" vs. help article "€4.90 return fee except defect". The bot alternates answers depending on the chunk retrieved. Resolve conflict: merge into 1 rule with conditions (defect = free, change of mind = fee).
How to use product metafields for the bot?
Shopify metafields anchor the product KB to the live catalog.
Useful support metafields
custom.care_instructions: care, washing
custom.compatibility: compatible models
custom.size_notes: fit, between sizes
custom.faq_json: 3-5 atomic Q&As per SKU
custom.support_last_review: support review date
Bot sync
Tier 2 bot reads product API + metafields on demand, not a stale static chunk. Hero SKUs (top 20% tickets): dedicated chunks + metafield. Long tail: live API is sufficient. Craftshift recommends Shopify Knowledge Base for FAQs and tone to supplement Catalog (Craftshift, Catalog and KB 2026).
What rituals and RACI for continuous freshness?
The KB catalog governance consists of short rituals and clear roles.
Simplified RACI
Product: metafields, specs, discontinuation
Support lead: help center, macros, Opportunities
Marketing: notifies promo/copy changes 24 hours in advance
Ops: shipping policy, carrier, lead times
Calendar
Weekly 20 min: Top 3 Opportunities + unmatched product bot. Monthly: audit of Last updated top 20 articles. Quarterly: 4×4 policy matrix (help, site, bot, macros). Post-launch: Day+3 and Day+14 review of each collection.
Freshness Indicator
Display "Last support review" on top-volume help articles. Goal: 100% of top 20 intent articles reviewed within 90 days. An article without a review date is treated as stale by the bot team.
How does Qstomy synchronize KB and Shopify catalog?
Qstomy reduces KB catalog drift by reading Shopify live instead of a frozen export.
Maintenance Features
API product sync: real-time stock, price, and variants
Metafields support: dynamic chunks per SKU
Drift alert: unmatched spike post-catalog modification
Export gaps: questions without chunk → Notion brief
Policy sync: /policies/* as canonical source
Quantified DTC Scenario
Electronics brand, 180 SKUs, 12 launches/year. Before Qstomy: manual corpus re-ingestion 2×/year, 19% unmatched post-launch. After metafields sync + policy + drift alert: unmatched post-launch 4%, KB maintenance time -6 h/month, "false product info" tickets -38%.
Explore AI support, Shopify, request a demo.
Which operational playbooks should be launched this week?
Playbook 1: 2-hour drift inventory
Export Shopify products modified in the last 30 days. Cross-reference with the help center 'Last updated'. List modified SKUs without a KB review. Prioritize the top 10 by ticket volume.
Playbook 2: source of truth matrix
Notion: Policy / Product / Promo × Shopify / Help / Bot / Macro. Red cell = contradiction. 1 fix per week until green.
Playbook 3: product launch template
Duplicate the sequence from day-7 to day+3 in section 5. Assign owners. Integrate into the marketing go-live Notion page.
Playbook 4: Friday Opportunities ritual
30 min: top 3 conflicts + top 3 gaps. Resolve or dismiss. Log action in Slack #kb-updates.
Playbook 5: 20-question bot freshness test
Pull 20 recent product tickets. Ask the bot. Score accuracy vs live PDP. < 90% = fix metafield or chunk before scaling ads.
Useful links
A living KB follows the catalog like a product: small, frequent updates are better than a massive annual audit.
Enzo
June 28, 2026
