E-commerce
July 19, 2026
"Will this case fit my model?" "What accessory am I missing to use the machine?" "You recommended the wrong filter to me." These messages arrive at the shopping cart, the chat, or after reception, when cross-selling ignores technical compatibility.
Heeya observes an add-to-cart rate 2 to 4 times higher on conversational cross-sell vs. static blocks, provided that the recommendation is relevant (Heeya, cross-sell reco 2026).
This guide #351 formalizes the compatible accessories AI chatbot: proposing the right add-on without generating returns. It completes support accessory (#350) (policy, ACC macros) and cross-sell (#152) with the focus on the AI use case for accessory compatibility and wrong accessory prevention.
Summary
Why automate compatible accessories by bot?
An incompatible accessory suggested by the bot = return, dispute and cross-sell refused next time. The bot must cross-reference compatible_models before any checkout link.
Three risks without a structured bot accessory
Blind cross-sell: best-seller, wrong model
Compatibility hallucination: LLM says "yes compatible" without data
Wrong accessory rate: return, lost accessory margin
Aculogi estimates that structured compatibility data reduces fitment returns by 20 to 40% (Aculogi, fitment data).
Angle #351
#350 documents ACC policy and agent workflow. #152 covers general cross-sell timing. #351 defines the dedicated compatible accessory bot: model lookup, compatibility gate, safe cross-sell.
Journey stage
Pre-purchase: "is it compatible?" and "what am I missing?" Post add-to-cart: logical add-on. Post-delivery: wrong accessory claim.
DTC Example
Tech accessory brand: 95 accessory tickets/month, cross-sell without lookup. After accessory_compat_check bot: 70% self-resolved, accessory_error_rate 4%, cross-sell accept rate +19 pts.
Safe cross-sell
Only suggest an accessory if model_id is in compatible_models. Otherwise ask for the model or escalate. Never generic "you might also like".
Bot accessory volume peak
Accessory_compat sessions surge at new parent launch, during accessory sales and tech holidays. Plan handoff backup and metafields audit on top 20 SKUs before these periods. A bot without updated data during a launch multiplies wrong accessories.
Cost of wrong accessory post-bot
Accessory return = round-trip shipping, restocking, agent time and negative review. Across 200 accessory sales/month at 12% error rate post cross-sell, the accessory margin melts away. A compatibility gate bot costs a metafield query, not 25 min of support dispute.
Safe cross-sell conversion impact
Compatible offer reassures and converts. Incompatible accessory post-purchase = return and customer hesitates to accept the next conversational cross-sell.
How does it differ from cross-selling and support accessories?
Seven neighboring contents, seven roles.
Support accessory (#350)
Guide #350: policy, 8 ACC macros, human workflow. The #351 implements the lookup AI and safe suggestion layer.
Cross-sell (#152)
Cross-sell (#152): timing, affinities, scripts. The #351 adds a mandatory compat gate before suggestion.
Spare parts bot (#344)
Spare bot (#344): post-life spare parts. The #351 = complementary accessories for main purchase.
Product compatibility
Product compatibility: pre-purchase matrix. The #351 deepens bot accessory + safe cross-sell.
Contextual recommendations
Contextual recommendations: passive widgets. The #351 responds in dialogue with model verification.
Upsell (#151)
Upsell (#151): better version of the same product. The #351 = distinct compatible accessory.
Promise #351
Intents accessory_*, compat gate tree, safe cross-sell, anti-hallucination, handoff wrong, KPI accessory_bot_resolution.
Which accessory intents should the bot classify?
Map the accessory bot intents before flows.
Ten accessory intents
accessory_compat_check: compatible with my model?
accessory_model_lookup: where to find model number
accessory_recommend: which accessory for my product
accessory_compare: diff between two accessory refs
accessory_essential: essential vs optional
accessory_cross_sell: post add-to-cart complement
accessory_not_compat: reasoned refusal + alternative
accessory_wrong_received: wrong accessory received
accessory_universal: universal with exclusions
accessory_third_party: third-party compatible or not
Mandatory session fields
parent_sku, model_id, accessory_sku_candidate, compat_confirmed (bool), cross_sell_trigger, photos_received. See taxonomy (#135).
Mining 90-day tickets
Export "accessory", "compatible", "case", "filter", "does not fit". Prioritize flows for top 20 accessory SKUs.
Affected verticals
Tech, household appliances, bike, auto, photo, smart home, baby, modular furniture. Each vertical has its own model_id fields: serial number, firmware revision, connector standard. The bot must map the customer vocabulary to the model_id metafield, not invent.
Distinction accessory vs spare intent
Accessory complements main product usage (case, filter, module). Spare part (#344) replaces a worn component. Classifying accessory_* vs spare_* avoids offering replacement parts as post-purchase cross-sell.
Verbatims to train
"Does this case fit my model?", "Which filter for my 2022 machine?", "You sold me the wrong accessory", "What am I missing to use the product?", "Universal compatible really?". Tag these phrases in the intent classifier.
How to build the gate accessory compat tree?
The arbre compat gate accessory blocks any proposal outside of compatible_models.
Six sequential gates
Gate intent: compat, recommend, cross_sell, wrong
Gate parent: main product or order identified
Gate model: confirmed model_id (photo if in doubt)
Gate accessory: requested or recommended accessory SKU
Gate compat: compatible_models metafield crossing
Gate output: PDP link, not_compat, handoff_agent
accessory_cross_sell branch
Post add-to-cart: read cart parent_sku → order model_id or ask → filter compatible_models accessory collection → propose 1 essential max.
accessory_recommend branch
"What am I missing?" → read essential_vs_optional metafield → list of compatible accessories sorted essential first.
Sequence #152 + #351
Timing cross-sell #152 triggers flow. Compat gate #351 filters before sending message. No cross-sell without gate OK.
Max turns rule
5-7 questions max lookup. Customer asks for human: handoff with partial model_id.
Which data sources does the accessory bot read?
The accessory bot responds from verifiable sources.
Five Shopify sources
Metafield accessory.compatible_models: JSON array
Metafield accessory.parent_product: product_reference
Metafield accessory.essential_vs_optional: priority cross-sell
Metafield accessory.compat_type: universal_strict / model_specific
Order line items: purchased parent, variant model_id
Documented Shopify metafields (Shopify, metafields 2025).
Accessories collection
Bot queries accessories collection filtered by compatible_models contains model_id. No full catalog LLM search.
Cross-sell affinities #152
cross_sell_primary metafield intersected with compatible_models. Affinity without compatibility = removed from bot recommendations.
Sheet sync #350
Same metafields as ACC ops support. See Shopify bot data.
Page /compatible-accessories
Help center hub: parent model lookup, list of filtered accessories. The bot reuses the same Storefront API query as the self-service widget. Consistency bot / page / agent = zero compatibility contradiction.
Order history as model_id source
Logged-in customer with parent order: bot reads variant model_id from order line. Avoids asking for model number again. Guest customer: ask for model_id or label photo before any cross-sell.
Post-launch metafields refresh
Each new parent or V2 revision = update compatible_models before activating bot cross-sell on this SKU. Merchandising notifies ops in Slack or Notion, bot only reads validated data.
How do I prevent the bot from suggesting an incompatible accessory?
The anti-hallucination accessory protects margin and cross-sell CSAT.
Five strict rules
Whitelist SKU: bot only mentions collection accessories + compat OK
Compat gate: no link if model_id is missing from metafield
No "universal" without listing incompatible_models
Cross-sell 1 essential: max 1 compatible proactive proposal
Fallback: handoff if model is not found
Accessory system prompt
"You never confirm compatibility without compatible_models. You never propose an accessory outside the matrix. You ask for model_id before cross-selling." See anti-hallucination.
Double customer confirmation
Before checkout link: "Confirm model [X], accessory [REF]." Reduces wrong accessory on the customer side.
Regression tests
25 scenarios: compat OK, not compat, universal exclusions, cross_sell post-cart, wrong received, model unknown. accessory_hallucination target 0%.
Which bot flows for pre- and post-purchase accessories?
The accessory bot intervenes on multiple touchpoints.
Accessory PDP Flow
Customer on accessory sheet: "compatible with my [model]?" → accessory_compat_check → yes/no + link or alternative.
Post add-to-cart Flow (#152)
10 s after parent add-to-cart: "Need an [essential accessory]?" → compat gate → 1 proposal max if OK.
"What am I missing" Flow
accessory_recommend: list essential then optional filtered by compatible_models, multiple PDP links OK.
Wrong item received Flow
Photos + compatible_models verification. If data error: handoff ACC-EXCHANGE (#350). If customer error: return policy.
Accessory handoff payload
parent_sku, model_id, accessory_sku, compat_result, cross_sell_trigger, photos[], order_id.
Example bot response
"Your iPhone 14 Pro is compatible with the Clear Case v3. Essential for daily protection. Add to cart: [link]. Budget alternative: [REF_B] also compatible."
Parent PDP chat widget Flow
Customer on main product sheet: "Which accessories for this model?" → bot reads page parent_sku → accessory_recommend filtered by compatible_models. No suggestions outside of the linked accessory collection.
Post-purchase email Flow
Email D+1 parent delivered: chat link "accessories compatible with your model". Bot retrieves order_id, confirms model_id, proposes 1 essential if not purchased. Timing #152 respected, compat gate #351 applied.
Accessory bot tone
Factual, not pushy. Confirm compatibility before linking. If not_compat: explain why and offer compatible alternative if one exists. Never downplay the risk of a wrong accessory.
How do I connect the accessory bot on Shopify?
The Shopify accessory bot setup aligns metafields #350 and cross-sell #152.
Technical Checklist
Populate accessory.compatible_models top 50 SKUs
Tag accessories collection + parent_product links
Configure accessory_* intents tree section 4
Connect post add-to-cart cross_sell safe trigger
Write anti-hallucination prompt section 6
Test 25 regression scenarios
Weekly accessory_error_rate dashboard
Parent PDP Widget
"Compatible accessories" filtered by metafield. Bot widget uses same data as self-service block.
Storefront API query
products(collection:accessories) filter compatible_models. No generic best-seller recommendations.
Phased Launch
Phase 1: accessory_compat_check top SKUs. Phase 2: cross_sell safe post-cart. Phase 3: full accessory_recommend.
Which KPIs should be measured on the accessory bot?
Without KPI bot accessory, it is impossible to prove cross-sell safe ROI.
Seven Key Metrics
accessory_bot_resolution: resolved without agent / accessory bot tickets
accessory_error_rate_bot: wrong returns after bot proposal
cross_sell_safe_accept_rate: acceptance post compat gate
accessory_compat_bot_rate: confirmed compat / lookups
accessory_handoff_rate: escalatons / accessory sessions
compat_to_conversion_bot: orders post-bot link
CSAT intent accessory_bot: satisfaction post-response
DTC Benchmark
Goal accessory_bot_resolution > 65%, accessory_error_rate_bot < 5%, cross_sell_safe_accept_rate > 20%, CSAT accessory bot > 4.5/5.
Correlation #350
Compare overall accessory_error_rate vs bot. Bot must lower wrong accessory without lowering cross_sell volume.
Monthly Review
Top handoff reasons: model missing, SKU without metafield. Enrich compatible_models based on cause.
Segment by Intent
accessory_compat_check vs accessory_cross_sell vs accessory_wrong_received have distinct KPIs. A low cross_sell safe accept rate can stem from timing #152, not a compat gate issue. Analyze separately.
accessory_error_rate_bot Alert
Alert threshold > 6% over sliding 7 days: pause proactive cross_sell, audit metafields of top 10 proposed SKUs, review hallucination logs. Resume bot after data correction validated by merchandising.
What edge cases and escalations should be anticipated?
Five accessory bot edge cases require escalation.
Model not found
Label photo requested. If missing: handoff, do not propose cross-sell.
Universal with exclusion
Bot cites incompatible_models. Customer excluded model: accessory_not_compat + alternative if it exists.
Parent revision V1 vs V2
Bot lists differences. Customer chooses. In case of doubt: handoff.
Cross-sell refused twice
Do not persist. Bot offers other help or closure. No incompatible accessory spam.
Wrong accessory post-bot
Audit compat log. Data error → exchange. Customer error → policy #350.
Third-party accessory
accessory_third_party: clear policy. Do not guarantee fit for untested competitor product.
Multi-accessories same model
Customer asks if two accessories are compatible together: bot checks each SKU in compatible_models. If conflict documented in compat_notes (e.g. two exclusive modules): handoff.
Customer changes mind on model mid-flow
Reset session model_id, pass compat gate again from model gate. Do not reuse previous compat_result on new model_id.
How does Qstomy offer compatible accessories?
Qstomy processes accessory_* intents from compatible_models, order history and cross-sell affinities.
Capabilities
accessory_compat_check: model_id → yes/no + link
accessory_cross_sell_safe: 1 essential post-cart if compat
accessory_recommend: filtered essential + optional list
accessory_compare: diff of two compatible refs
Handoff wrong: photos + compat audit log
Encrypted DTC Scenario
Tech brand, 95 accessory tickets/month, cross-sell without gate.
After Qstomy accessory bot: 72% auto-resolved, accessory_error_rate_bot 3%, cross_sell_safe_accept 24%, CSAT accessory bot 4.7/5.
Explore AI support, Shopify, request a demo.
What is the checklist for launching the accessory bot?
Accessory bot checklist (10 steps)
Audit accessory tickets 90 d (#350)
Populate compatible_models top 50 accessories
Document compat gate tree section 4
Align cross-sell affinities #152 with compat gate
Configure accessory_* intents
Draft anti-hallucination prompt section 6
Test 25 regression scenarios
Enable safe cross_sell post add-to-cart
Weekly accessory_error_rate_bot dashboard
Monthly review of wrong accessory and metafields
In brief
#351 = bot accessory compat, not cross-sell alone (#152)
Compat gate: mandatory before offering
10 accessory_* intents: compat, recommend, cross_sell
Anti-hallucination: SKU whitelist + compatible_models
KPI accessory_error_rate_bot: target < 5 %
FAQ
Difference with support #350?
#350 = ACC ops policy. #351 = AI bot lookup and safe cross-sell.
Difference with cross-sell #152?
#152 = timing and scripts. #351 = compat gate before any offering.
Can the bot offer 3 accessories?
Proactive cross-sell: 1 essential max. Recommend on demand: filtered list OK.
Spare link #344?
#344 = spare parts. #351 = complementary accessories for main purchase.
Without compatible_models metafield?
Agent escalation. No bot offering.
Going further
Test safe cross_sell staging: parent add-to-cart, verify that the bot only offers compatible_models OK accessories.
Share this #351 guide with merchandising: a cross-sell without compat gate costs more in returns than it gains in margin.
Enzo
July 19, 2026
