Decisioning Studio

​

Overview

The Decisioning Studio is where you define everything your system can recommend and the rules that govern those recommendations. It contains all the building blocks that feed into Decision Flows — from business categories and offers to channels, qualification rules, and contact policies. Think of the Studio as your decisioning workbench: you configure the pieces here, and Decision Flows orchestrate them at runtime.

​

How the Pieces Connect

​

Studio Components

​

Business Hierarchy

You organize offers into a two-level hierarchy of categories and sub-categories. Categories define the structure — for example, “Credit Cards” with sub-categories “Premium Cards” and “Basic Cards”. Categories can also define custom fields including computed fields with formulas that personalize values per customer at decision time. Once your hierarchy is set up, you create offers within it.

​

Offers

An offer represents something you can recommend — a product, promotion, message, or next-best-action. Each offer belongs to a category and sub-category, and carries configuration for priority, budget, scheduling, and status lifecycle. Custom fields defined by the category are filled in per offer. Offers need channels to reach customers.

​

Channels

Channels define how recommendations are delivered. Each channel has a delivery mode (API for real-time, File for batch, Manual for human review) and optional provider configuration. Channels support multiple placements — named slots like “hero_offer” or “sidebar_ad” — for fine-grained content targeting. With channels set up, you create creatives that define the actual content each customer sees.

​

Creatives

Creatives are the content variants that customers actually see. Each creative ties an offer to a channel with specific content — subject lines, headlines, images, CTAs. Creatives support personalization variables (like {{customer.first_name}}), A/B test variants, and frequency constraints. Now that you have content, you need rules to control who sees it.

​

Qualification Rules

Qualification rules determine whether a customer is eligible for an offer. Hard rules are binary gates (pass or fail) — for example, age >= 18 or segment = premium. Soft rules adjust the offer’s score without eliminating it — for example, “loyalty members get a 1.2x multiplier”. Rules can apply globally or to specific categories. Rules control who sees an offer. To control how often, set up contact policies.

​

Contact Policies

Contact policies prevent over-contacting customers. They define 8 types of guardrails: frequency caps, cooldown periods, budget limits, outcome-based suppression, segment restrictions, time windows, mutual exclusion, and cross-channel limits. Policies can apply globally, per offer, per channel, or per creative. With rules and policies configured, you’re ready to build the pipeline that puts it all together.

​

Decision Flows

Decision Flows are visual pipelines that orchestrate the recommendation process. The composable pipeline lets you assemble 14 node types across 3 phases (Narrow, Score & Rank, Output) in a drag-and-drop canvas editor. Every recommendation passes through a Decision Flow.

​

Outcome Types

Outcome types define the vocabulary of customer responses — impressions, clicks, conversions, dismissals, and custom types. KaireonAI includes 10 default outcome types and lets you create custom ones. Outcomes are recorded via the Respond API and feed into behavioral metrics and model training.

​

Behavioral Metrics

Behavioral metrics aggregate interaction history into signals you can use in rules and scoring — conversion rate over 30 days, total revenue per customer, click-through rate by channel. You define metrics with an aggregation type, time window, and optional dimension. Up to 20 metrics per tenant.

​

Portfolio Optimization

When multiple offers qualify for a customer, portfolio optimization determines the final ranking. KaireonAI uses multi-objective scoring across revenue, margin, propensity, engagement, and strategic priority. You create optimization profiles that define the weight distribution — for example, 60% propensity + 20% revenue + 20% priority. The Optimize node in the composable pipeline applies these profiles at decision time.

​

Next Steps