Pricing Platform Bundles Docs
Flagships SK Chat SK Voice SK CRM
Core Platform Modules
SK Billing SK Inventory SK Calendar SK Tickets SK Works SK Catalog SK Queue SK Upsell SK Notifications
Verticals Hospitality
Sign in Start free
SK Upsell · smart product recommendations

Margin-aware suggestions.
Boost ticket sizes, respect customer bounds.

Maximize average order value dynamically at checkout, booking, or chat flows. SK Upsell evaluates product recommendations deterministically using tenant-configured Zod AST rules and margin-tilted expected values, protected by decline suppression thresholds. Smart recommendations, dynamic guardrails.

Start freeExplore features →
0 LLM hot-path latency · Expected value math · Decline-sensitive suppression · Core Suite
< 5ms matching0 LLM hot-path calls
Margin-aware mathmaximizes expected value
Decline suppressioncapped to 3 session declines
No per-seat taxflat suite pricing
Smart Attachments

Maximize profit, not annoyance.
Rules engine meets margin science.

Standard recommendation widgets push random items or spam the checkout with unrelated inventory. SK Upsell evaluates rule parameters deterministically at checkout, matching item margins and client history to trigger the highest-yielding suggestions.

Expected value (attach × margin) sorting

Pitches are sorted by their expected contribution value. By multiplying the historical conversion rate of a recommendation by its exact unit margin (`Expected Value = Attach Rate × (Price - Cost)`), the engine prioritizes suggestions that bring real profit rather than high-volume low-margin noise.

  • Auto-calculates gross margins dynamically from inventory
  • Weights rankings using historical attach rates
  • Applies stable tie-breakers to ensure consistent sorting order

Decline-sensitive session back-off

Aggressive upselling kills conversion rates. SK Upsell tracks guest interaction state: if a customer dismisses or declines three recommended offers during a single session, the recommendation engine automatically shuts off. No nagging, no clutter.

  • Counts active declines in real-time session stores
  • Suppresses all triggers when decline limit threshold is hit
  • Resets automatically when sessions close or settle

Zod-backed trigger AST schemas

Rules are defined as structured data, never hard-coded scripts or messy string patterns. Using Zod-defined Abstract Syntax Trees (ASTs), triggers parse complex logical rules (e.g. cart value ratios, customer demographics, time limits, and location codes) safely. Evaluated deterministically at runtime with zero LLM query overhead.

  • Enforces strict parameter validation at schema compile time
  • Checks cart properties, time scopes, and user segments
  • Predictable execution results under sub-millisecond latencies

Proportionality price delta limits

Upsell suggestions must feel reasonable. Pitching a ₹5,000 item on a ₹1,000 cart is bad math. The engine checks active cart values and filters out any candidate recommendation whose cost ratio exceeds the configured proportionality cap (e.g. max 35% delta). Only relevant, proportional items reach the customer.

  • Configurable percentage caps prevent pricing mismatch
  • Checks target margins and baseline item values
  • Filters out out-of-bounds inventory candidates automatically

Append-only suggestion outbox ledger

Conversion metrics must be auditable. In SK Upsell, all recommendation impressions, acceptances, and declines are recorded to a transaction ledger. Writes are append-only. Edit and delete requests are dropped at the Postgres trigger layer. Verification sweeps prevent back-dating, providing transparent attach statistics.

  • Database constraints reject any UPDATE or DELETE operations
  • Enforces immutable logging of all customer choices
  • Powers feed rollup metrics directly from raw event lists

That's the core of SK Upsell.

Expected value calculation metrics, logical AST gates, session limits, and the append-only outbox ledger—every detail, documented.

Explore all features →

Smart suggestions that drive conversions.

SK Upsell operates as a core module in the Softknack platform: it queries live stock from Inventory, checks guest schedules in Calendar, reads client profiles from CRM, and records conversion invoices in Billing—instantly, without custom developer work.

See the platform →

Flat-rate suite pricing. No transaction cuts.

SK Upsell is included in the Core Suite. Billed as one flat subscription, allowing you to optimize checkouts, recommendations, and bundles without paying commission fees to intermediaries.

core suite: CRM + calendar + tickets + works + catalog + inventory + queue + upsell + notifications · billed in ₹, $ or €
See Core pricing →

Fair questions, straight answers.

Does this query LLMs in real-time during checkout?
No. Real-time recommendation routes use deterministic AST trigger matching and statistical matrix lookups in under 5ms, avoiding LLM latencies and costs. LLMs are only run offline in background worker tasks to derive and propose rules.
How does the decline back-off policy operate?
If a guest closes, declines, or skips three recommendations in an active session, the engine sets a suppression state. The recommendation endpoint will return empty arrays for the rest of that session, preserving a clean user experience.
Are the suggestion metrics safe from manipulation?
Yes. Recommendation event logging maps to append-only database schemas. Manual rows edits or deletes are blocked by PostgreSQL constraints, guaranteeing that all attach-rate stats represent genuine outcomes.

pricing details and the comparison matrices live on the pricing page

Optimize your checkout profit metrics today.

Enable the suite, configure your price delta thresholds, and let deterministic margin optimization boost your ticket sizes.

Start free  Compare with Rebuy & Linkello →