E-commerce
August 12, 2026
"How many points do I have?" "How do I use them at checkout?" "Can I combine them with -20%?" "What does the Gold level earn?" Four loyalty questions repeated in chat, even though the program is documented on a page that nobody reads.
LoyaltyLion estimates that 64% of consumers abandon a program if the rewards or the redemption journey remain confusing (LoyaltyLion, loyalty 2026). US Tech Automations notes that stores displaying balances and rules in a shopping context see 38% more redemption compared to opaque programs (US Tech Automations, loyalty friction 2026).
This guide #375 covers the AI chatbot for e-commerce loyalty programs: balance, rewards, rules, LOY-BOT intents, and Smile.io/Shopify integration. Distinct from pre-launch loyalty questions (anticipating go-live) and missing points (#374) (PTS-REC incident investigation): here, we focus on AI automation for current program balances and rules.
Summary
Why automate loyalty balance and rules via an AI chatbot?
A well-configured e-commerce loyalty chatbot answers queries about balances, earning rules, redemption, and tiers without agent tickets, while handing off missing points issues to PTS-REC (#374).
Five customer frustrations without a loyalty bot
Invisible balance: customers do not know how many points they have
Confusing redemption: how to apply points to the cart
Vague promo stacking: unclear if promo codes + points can be combined
Opaque VIP tiers: Gold threshold unknown
Silent expiration: points lost without warning
Antavo cites expiration before use (41%) and unattractive rewards (39%) among the major loyalty program irritants for 2026 (Antavo, 2026 Loyalty Report).
Angle #375
The loyalty questions guide lists pre-launch FAQs. The #374 PTS-REC handles credit issues. #375 deploys a loyalty AI bot with loyalty_balance, loyalty_redeem, and loyalty_rules intents for standard self-service.
DTC Example
Cosmetics brand using Smile.io, 124 loyalty_general tickets/month before LOY-BOT. After implementing balance bot + redemption guide: loyalty_bot_resolution 71%, total loyalty tickets -38%, checkout redemption rate +22%.
Loyalty bot vs generalist bot
A generic FAQ cannot read a customer's balance. LOY-BOT connects Shopify email + Smile API for a personalized response: "You have 412 points, which is a €4.12 discount."
Journey Touchpoint
PDP (points eligibility), cart (redemption), customer account (balance), post-purchase (expected earning). Contextual bot based on the page.
Issue Handoff
pts_missing intent with calculation mismatch → #374 PTS-REC, no self-invented credit by the bot.
Loyalty Bot ROI
Loyalty FAQ ticket = 7 min of agent time. Balance + redeem bot = 45 s self-service. Over 90 diverted tickets/month: 630 min saved.
Checkout Redemption
The redeem_how bot with a deep link to the logged-in cart increases dormant points conversion. Unused points = program perceived as useless.
How does it differ from #374 and the pre-launch questions?
Seven neighboring contents, seven roles on loyalty and AI automation.
Pre-launch loyalty questions
Loyalty questions: FAQ go-live mapping. #375 = bot deployed post-launch with live customer data.
Missing points (#374)
PTS-REC (#374): manual adjust incident investigation. #375 answers common rules; handoff to #374 if bug.
Shopify Customization (#20)
Bot customization (#20): general customer data. #375 applies to Smile/Yotpo loyalty layer.
Future referral (#376)
#376 will handle referral support codes/fraud. #375 covers general points program, not referral ops.
AI Governance (#142)
Governance (#142): saying vs doing. #375: bot explains redemption, does not credit points without agent.
VIP Escalation
VIP Escalation: Gold+ tier dispute. Bot cites tier benefits, handoff if policy exception.
#375 Promise
LOY-BOT framework, 12 intents, RAG rules corpus, balance/redemption/tiers flows, loyalty_bot_resolution KPI.
Which LOY-BOT intents should be prioritized for mapping?
Map the bot loyalty intents before corpus and API integrations.
Twelve LOY-BOT intents
loyalty_balance: points balance and € value
loyalty_earn_rules: how to earn points
loyalty_redeem_how: use points checkout
loyalty_rewards_catalog: available rewards
loyalty_tier_status: current level and next threshold
loyalty_tier_benefits: Gold/Silver benefits
loyalty_promo_stack: promo code accumulation
loyalty_expiration: points expiration date
loyalty_exclusions: non-eligible products
loyalty_history: last points transactions
loyalty_signup: join program
loyalty_pts_missing: handoff #374 PTS-REC
MVP Prioritization
Week 1: balance, earn_rules, redeem_how, promo_stack (80% of loyalty FAQ tickets). Week 2: tiers, expiration, rewards_catalog.
90-day ticket mining
Export loyalty tags. Quantify volume per intent. Validate deployment order.
Trigger page context
Cart: redeem_how, promo_stack
Account: balance, history, tier_status
PDP: earn_rules, exclusions
NLP Routing
"What are my Gold benefits?" → loyalty_tier_benefits. "I don't have my points from last week" → loyalty_pts_missing if time elapsed, otherwise loyalty_balance + credit date.
Out-of-scope bot intents
Refund claim, chargeback dispute, GDPR request: immediate handoff without attempting loyalty response.
How to structure the LOY-BOT framework into six blocks?
The LOY-BOT framework structures loyalty bot deployment into six reproducible technical blocks.
Six LOY-BOT blocks
LB-1 Customer Auth: Shopify logged-in email or guest order email verification
LB-2 Loyalty API: Smile/Yotpo customer points + tier + activity
LB-3 RAG Corpus: program rules, FAQ, indexed rewards catalog
LB-4 Intent router: NLP → 12 intents section 3
LB-5 Response template: dynamic balance + static RAG rule
LB-6 Handoff rules: pts_missing → #374, VIP exception → agent
Balance response example
"Hello [First Name], you have 412 points (value €4.12). Next expected credit: +89 pts on August 14 (shipped order #7842)." Account redemption link.
Redeem response example
"At checkout, log in, go to the 'Loyalty Points' section, and check 'Use my points'. 100 points = €1 discount. Minimum 200 points for redemption."
Shopify enrichment data
Customer VIP_GOLD tags, order_count, last_order_date. Bot personalizes tier_benefits. Link personalization (#20).
Fallback mode
Smile API down: bot quotes static RAG FAQ without live balance + "balance temporarily unavailable, try again or contact us".
Dynamic PDP earning calculation
PDP price €79 × 2 pts/€ = bot responds "This purchase earns you 158 points" without an earn_rules ticket.
Multichannel widget
Same LOY-BOT on site, mobile app wrapper, and post-purchase email link "See my points".
Which RAG corpus should be populated for the program rules?
The loyalty RAG corpus feeds stable rule responses that the API does not cover alone.
Eight LOY-RAG corpus blocks
Earning rate: X pts/€, rounding, shipping included or not
Credit delay: D+3 post-shipping
Redemption: min points, conversion value, checkout steps
Promo accumulation: yes/no, sale exceptions
Tiers: Silver/Gold thresholds, benefits, annual reset
Expiration: 12 months inactivity, D-30 email alert
Exclusions: gift cards, wholesale, marketplace
Rewards: non-monetary rewards catalog (samples, free shipping)
Corpus sync
Each marketing policy change → RAG update + test 5 bot questions same day. Link bot FAQ cleanup.
FR/EN Version
Corpus per locale if multilingual store. Intent detect customer language.
Anti-hallucination
Bot cites source RAG block. If question out of corpus: "I do not have this information, I will connect you" no invention of point rates.
Which balance and redemption conversational flows?
Loyalty conversational flows guide customers from balance checking to redemption without needing a ticket.
Balance → redeem flow
Customer: "My points?" → balance bot + "Would you like to use them now?" → redeem_how + cart deep link if logged-in.
PDP earn flow
PDP Widget "Earn 158 points with this purchase" → click opens contextual earn_rules bot showing SKU price.
Tier upgrade flow
"How much for Gold?" → tier_status: "You need €120 more in eligible orders by Dec 31. Gold benefits: free shipping, early access."
Cart promo_stack flow
"15% off code + points?" → promo_stack: "The promo code excludes earning points on this cart. You can use 300 existing points (-€3). Both benefits cannot be combined."
Signup flow
Non-member visitor: loyalty_signup → benefits + account creation link + 100 pts welcome bonus.
Quick reply buttons
Balance | Use points | My benefits | Missing points. Speeds up mobile resolution.
Expiration alert flow
Customer "Are my points expiring?" → loyalty_expiration: "120 points expire on Sept 30. Use them at checkout or exchange them for a catalog reward."
Proactive cart message
Cart > €50 + logged-in customer + points > 200: bubble "You have 412 pts (-€4.12), how would you like to apply them?"
How to integrate Smile.io, Shopify, and the chat widget?
The LOY-BOT technical integration connects Shopify, Smile.io, and the chat widget without double customer entry.
Smile.io API
GET customer by email: points_balance, vip_tier, points_activity. 5-minute cache to limit rate limit.
Yotpo Loyalty / LoyaltyLion
Equivalent endpoints for customer rewards. Adapt LB-2 depending on the installed app.
Shopify Customer Account
OAuth logged-in session: bot skips email prompt. Guest: order number + email verification before balance (privacy).
Checkout extensibility
If Shopify Plus redemption: bot deep links to checkout with loyalty redeem hint parameter.
Webhooks sync corpus
Smile tier change → log activity bot can reference. Order fulfilled → bot loyalty_history updated.
Omnichannel POS
Web + store balance if sync is active. Otherwise, bot references the omnichannel policy.
API Rate limit
Smile limits requests/min: 5-minute balance cache per customer_id. Bot announces "balance at time T" if cache < 5 min.
Staging tests
Smile sandbox + Shopify test order before production: validate balance, tier, and redeem_how across 3 profiles (Silver, Gold, guest).
What kind of governance: declaring the balance without crediting?
The loyalty bot governance defines what the AI can say and what remains agent-only.
Bot can say
Balance, tier, policy rules, redemption steps
Expected points earned on current order
Expiration date if API provides it
Bot cannot do
Manual points adjustment (→ #374 agent)
Override exclusion policy
Promise bonus undocumented in RAG
Modify tier manually
Align governance (#142) and limits (#124).
GDPR balance
Display balance only after email/order identity verification. No third-party balance.
Audit logs
Log intent, displayed balance, handoff yes/no. Monthly review of hallucination samples.
Which loyalty_bot KPIs should be measured?
Loyalty bot KPIs measure ticket deflection and redemption impact.
Eight key metrics
loyalty_bot_resolution: resolved without agent / loyalty intent sessions
loyalty_ticket_deflection: decrease in loyalty_general tickets post-bot
loyalty_bot_csat: post-conversation loyalty survey
redeem_after_bot: checkout redemption 24 hours post redeem_how bot
intent_mix_loyalty: distribution of 12 intents
loyalty_handoff_rate: pts_missing + VIP → agent
rag_miss_rate: unanswered questions in corpus
signup_from_bot: loyalty_signup → account created
DTC Benchmark
Target loyalty_bot_resolution > 65%, loyalty_ticket_deflection -30%, redeem_after_bot > 15% of redeem_how sessions.
Weekly Dashboard
Intent volume, resolution, pts_missing handoff, top 5 rag_miss. Share with #product + #support.
PDP widget A/B test
PDP with vs. without "Earn X pts, ask the bot". Measure earn awareness and earn_rules tickets.
Cost of rag_miss
Each rag_miss = 7 min agent handoff. Top 5 rag_miss → add LOY-RAG block following week.
What edge cases and handoffs should be anticipated?
Six loyalty bot edge cases requires a handoff or specific rules.
Guest without account
loyalty_signup is the priority. No balance without account creation.
Duplicate email account
Two Smile profiles for the same person: handoff to merge agent #374 PR-6.
Points expired yesterday
Bot shows expiration activity log. No bot re-credit, rare agent policy exception.
B2B wholesale
B2C program only. Macro redirect to pro account.
Referral question
"My referral?" → future handoff #376 or RAG referral FAQ if simple.
API timeout
Fallback to static FAQ + async balance ticket. Do not invent a number.
Anti-patterns to avoid
Invented balance without API, instant credit promise vs D+3 policy, ignoring promo_stack, loop without handoff #374, outdated RAG corpus. Monthly regression test: twenty gold standard questions vs Notion policy.
How does Qstomy deploy LOY-BOT on Shopify?
Qstomy deploys LOY-BOT: 12 intents, Smile API sync, RAG rules corpus, and PTS-REC handoff.
Loyalty bot capabilities
loyalty_balance_live: balance + € value Smile API
loyalty_rag_rules: earn, redeem, promo_stack indexed
loyalty_tier_lookup: status + points to next tier
loyalty_redeem_guide: checkout steps + cart deep link
loyalty_pts_missing_handoff: → #374 PTS-REC
loyalty_context_pdp: SKU points earning current price
Quantified DTC Scenario
Fashion brand Yotpo Loyalty, 98 loyalty tickets/month, loyalty_bot_resolution 0% before Qstomy.
After LOY-BOT MVP 4 intents: loyalty_bot_resolution 68%, loyalty_ticket_deflection -35%, redeem_after_bot 18%, loyalty_bot_csat 4.5/5.
Explore AI support, Shopify, request a demo.
What checklist is needed to launch LOY-BOT?
LOY-BOT Checklist (10 steps)
Audit loyalty tickets for past 90 days and prioritize MVP intents
Draft 8 LOY-RAG corpus blocks
Connect Smile/Yotpo customer points API
Deploy balance, earn, redeem, promo_stack intents
Dynamic balance response templates + RAG rules
Handoff pts_missing → #374 configured
PDP Widget "Earn X pts" + bot trigger
Anti-hallucination tests with 20 edge questions
Weekly loyalty_bot_resolution dashboard
Quarterly marketing sync corpus review
In short
#375 = standard loyalty AI bot, not incident #374
LOY-BOT: 12 intents, API + RAG
MVP: balance, earn, redeem, promo_stack
Bot states balance, does not adjust points
KPI loyalty_bot_resolution: target > 65%
FAQ
Difference with missing points #374?
#375 = balance FAQ and rules. #374 = incident credit investigation.
Can the bot credit points?
No. Agent handoff PTS-REC if variance is confirmed.
Guest checkout?
loyalty_signup or order email verification before showing balance.
Smile vs Yotpo?
LOY-BOT adapts LB-2 API based on installed app.
Referral program relationship #376?
#375 general points program. #376 = referral codes ops/fraud.
Going further
Test twenty mystery shop loyalty questions: verify API balance = checkout balance and pts_missing handoff works correctly.
Share this #375 guide with product and support: a connected loyalty bot transforms an opaque program into a self-service that boosts redemption.
Enzo
August 12, 2026
