E-commerce
June 28, 2026
The customer adds a serum to the basket. Their skin needs five to seven products in a morning/evening routine. A "you might also like" carousel suggests three SKUs without any order or logic. A multi-product routine via chat, however, sequences the steps, checks compatibilities, and fills the basket in a single action.
Alhena estimates a +38% AOV when an AI routine builder replaces isolated product recommendations on beauty traffic (Alhena, skincare routine 2026).
This guide #168 covers the routine recommendation chatbot for skincare, nutrition, and equipment. New use case: multi-SKU routine, not just a simple product recommendation. Distinct from contextual recommendations and cross-sell (#152).
Summary
Why does recommending a multi-product routine change conversion?
An average beauty basket often contains one to two items, whereas an effective protocol requires five to seven. The gap between purchase and real need is an untapped AOV lever.
What a routine brings
Confidence: search order explained
Consistency: compatible, non-contradictory active ingredients
Simplicity: one dialogue vs five product pages
Bundled basket: add all in one click
Routine vs isolated recommendation
Heeya points out that in 2026 beauty, the RAG bot can recommend cleanser, toner, serum, cream, SPF in the order of application based on skin diagnosis (Heeya, 2026 beauty chatbot). This is not an incidental upsell: it is a complete user journey.
How does it differ from cross-selling, upselling, and guided selling?
Four conversational levers, four logics.
Cross-sell (#152) and upsell (#151)
Cross-sell (#152) adds a one-off complement. Upsell (#151) pushes a premium version of the same SKU. The routine assembles multiple references in a sequenced protocol.
Guided selling (#150)
Guided selling (#150) guides towards 1 to 3 products via a tree. Here: multiple slots (morning/evening, week 1-4, nutrition stack).
Cosmetics advice (#146)
Cosmetics (#146) details the skin advice and INCI content. This guide #168 covers the transverse routine bot architecture for skincare, nutrition, equipment.
Which DTC verticals benefit from a routine bot?
The e-commerce routine bot shines when the use case requires multiple ordered products.
Skincare and treatments
Morning protocol (cleanser, Vitamin C, moisturizer, SPF) and evening protocol (double cleansing, night active ingredient, barrier). Revieve structures routine logic based on skin analysis + brand catalog (Revieve, routine advisor 2026).
Nutrition and supplements
Magnesium + omega + Vitamin D stack with intake timing and exclusions. See supplements (#147) for regulatory limits.
Sports and tech equipment
Home gym setup (mat + dumbbells + accessory) or ergonomic office setup (chair + footrest + monitor stand). See electronics (#148).
Opportunity signals
"Where do I start?" tickets, "not suitable on its own" feedback, and category AOV < 40% of the ideal routine basket.
How to structure a skincare routine via chatbot?
The skincare bot routine follows fixed slots, filled according to the skin profile collected in 5 to 7 questions.
Morning slots
Cleanser (texture according to skin type)
Targeted treatment (niacinamide, vitamin C)
Moisturizer
SPF if photosensitizing active ingredients are used
Evening slots
Makeup remover if wearing makeup, night active ingredient (retinol or AHA, not both for beginners), barrier cream. Alhena recommends checking ingredient compatibility before validating the protocol (Alhena, compatibility of active ingredients).
Chat presentation
Card per slot: photo, name, "why this step", numerical order. CTA "Add entire routine" + individual slot link.
Bot message example (copy-paste)
"Beginner combination skin routine, morning: 1) Gentle cleansing gel, 2) Niacinamide serum 5%, 3) Light cream, 4) SPF 50. Evening: cleansing oil if wearing makeup, barrier moisturizing cream. Application: apply from most liquid to thickest." Each line linked to the exact SKU in stock.
How to structure a nutrition routine via a chatbot?
The nutrition routine bot compiles a conservative stack without personalized medical advice.
Profile collection (4-6 questions)
Stated goal (energy, sleep, sport)
Diet (vegan, gluten-free)
Ongoing treatments → handoff
Pregnancy/breastfeeding → handoff
Products already consumed (avoid duplication)
Stack-type slots
Base (multivitamin or omega), goal target (magnesium sleep), sport option if active profile. Responses limited to authorized label claims. Systematic disclaimer: "dietary supplement, does not replace a varied diet".
Intake timing
Display morning/evening/meals according to the product leaflet, never dose adjustment by the bot.
Example sleep stack (3 slots)
Slot 1: magnesium bisglycinate (evening instructions). Slot 2: omega 3 (meal). Slot 3: melatonin if product is authorized on label, otherwise slot excluded and handoff if therapeutic request. Total basket displayed with reminder "do not combine with other in-house magnesium".
How to structure an equipment routine via chatbot?
The equipment routine bot assembles a coherent setup according to space, budget, and level.
Key questions
Usage (strength training, yoga, teleworking)
Available surface area (m², ceiling)
Total routine budget, not single product
Beginner vs. advanced level
Example home gym setup slots
Floor (mat), load (adjustable dumbbells), accessory (bands), recovery (roller). Verify compatibility of maximum floor weight + declared customer load.
Desk setup
Ergonomic chair, footrest, laptop stand, lighting. Link to furniture (#149) for dimensions and delivery.
What compatibility rules and safeguards should be implemented?
A routine compatibility engine prevents the bot from recommending dangerous or inconsistent combinations.
Skincare
Retinol + AHA same evening = warning or alternating days
Exclusion of fragrance INCI if allergy declared
Pregnancy: blocklist retinol, strong acids
Mandatory SPF if retinol or vitamin C is used during the day
Nutrition
No accumulation of two products with the same active ingredient beyond the daily label dose. Immediate handoff for drug interactions. See regulated (#119).
Equipment
Total routine budget respected. Stock confirmed on each slot before presentation. No recommendations outside the brand catalog.
RAG without hallucination
Each routine SKU must exist in the live catalog. See hallucinations (#123).
Document the matrix
Notion table: active A + active B → OK / alternate / forbidden. Validated by product manager. Versioned: any formulation change triggers a matrix review before bot redeployment.
How do you present the routine in chat and fill the cart?
The routine UX chatbot compresses the journey from five product pages into a single conversation.
Message format
Intro: "Here is your [goal] routine in X steps." Block per numbered slot. Total price summary + bundle savings if applicable. Main CTA: "Add the X products."
Agentic checkout
Alhena describes a redirect to a pre-filled checkout after routine validation, avoiding multi-PDP navigation (Alhena, agentic checkout).
Post-routine personalization
Customer removes a slot: recalculation without breaking the remaining order. Request for an alternative for slot 3: swap SKU with the same skin tag, not a completely new routine.
Profile saving
Store validated routine for repurchase and cross-sell refill. See replenishment subscription.
Mobile first
On smartphones, display a maximum of 3 visible slots with a "see more" button rather than a block of seven cards. The add-all CTA remains sticky at the bottom of the widget.
How to measure AOV and the quality of recommended routines?
Chatbot routine KPIs validate business value and advice relevance.
Conversion metrics
Routine session AOV vs PDP alone
Routine add-all rate / presented routines
Average items per purchased routine
Completion flow conversion diagnostic → routine
Quality metrics
Post-routine inadequacy feedback
"Routine does not fit" tickets
Nutrition/skincare medical escalations
Post-routine session CSAT
Benchmark
Skincare pilot target: routine AOV +25 to 38% vs control, 4+ products per purchased routine, flow abandonment rate < 35%.
Which routine bot errors cost returns and trust?
Five AI routine anti-patterns to avoid in production.
Frequent errors
Overloaded routine: 8 strong active ingredients for a beginner
Inconsistent order: SPF before cleanser
SKU out of stock in the presented routine
Health promise on nutrition stack
Ignoring existing routine customer already using retinol
Over-personalization
Asking for too much sensitive data without added value is scary. 5 to 7 questions are enough for a first protocol.
Trust link
See undecided assistant (#36) and agent vs chatbot (#162) for advising posture vs aggressive sales.
How does Qstomy build multi-product routines?
Qstomy builds routines using configurable slots, linked to Shopify tags and compatibility rules.
Routine Features
Diagnostic flow: 5-7 conditional questions
Morning/evening slots: tag u2192 SKU mapping
Compatibility matrix: active ingredients, allergies, pregnancy
Add all to cart: pre-filled Shopify checkout
Swap slot: alternative options without restarting the flow
Quantified DTC Scenario
Skincare brand with 45 SKUs, u20ac38 AOV (1.2 products/cart). Deployment of a beginner 4-slot routine bot: routine sessions AOV u20ac38 u2192 u20ac67, add-all rate 31%, routine items/cart 3.8, mismatch returns -14% vs single-product buyers, "where to start" support tickets -39%.
Operational Lesson
Limiting beginner routines to 4 slots converted better than intimidating 7-product protocols. Slot 5 upsell offered 14 days post-purchase. Sessions where the customer modified a slot (28% add-all conversion) outperformed fixed routines without swaps (19%), indicating that flexibility strengthens protocol commitment.
Explore AI sales agent, customer support, Shopify, request a demo.
Which playbooks to launch a routine bot in 3 weeks?
Playbook 1: audit slots (1 day)
Define the ideal protocol per persona (dry skin beginner, home gym athlete). List SKUs per slot. Catalog gap.
Playbook 2: Shopify tags (2 days)
Metafields slot_morning, slot_evening, concern, level. Apply to top 30 SKUs. Notion compatibility matrix validated by PR.
Playbook 3: bot flow (3 days)
Diagnostic questions, scoring rules, routine card template, 20 gold set cases per vertical.
Playbook 4: category pilot (1 week)
Hero collection only, 50% traffic to routine widget. Measure AOV vs. control.
Playbook 5: REG guardrails (1 day)
Nutrition/skincare disclaimers, blocklist claims, pregnancy/reaction handoff. 30 min agent training.
Playbook 6: extension and refill
Routine saved to customer profile, Day+30 refill stack email, quarterly quarterly upsell slot.
Useful links
Recommending a routine via chatbot is not about stacking products: it is about offering a coherent protocol that the customer understands, buys with confidence, and renews over time.
Enzo
June 28, 2026
