How scoring works
Scoring in Productised isn't a decision tree or a separate node — your AI Agent reads the whole conversation and produces the score, tier and analysis live, for each respondent.
There is no separate "scoring node" in Productised. Scoring is done by the AI Agent — the same node that writes your result. It reads the full pattern of answers and produces the score, the tier label, the per-dimension breakdown and the written analysis, generated for each end user in the moment.
This is the core of what makes a Productised assessment different from a quiz builder: the outcome is reasoned, not selected from a shelf of pre-written results.
The AI Agent is the scoring engine
When you build a scored product, the AI Agent is configured with output fields — for example overall_score, tier_label, dim1_score, dim1_insight, action_1. You set these by describing what you want in plain language; Claude wires them up (set_agent_fields).
At completion, the agent:
- Reads every answer — including free text, tone and the tension between responses.
- Applies your scoring objective and framing.
- Writes each field — the score, the tier, the insights and the next steps — specifically for that end user.
Those fields then populate your result page through {{ object:* }} tokens (e.g. {{ object:overall_score }}, {{ object:tier_label }}).
Because the score is reasoned, two nearly identical end users can receive meaningfully different results when one detail changes the picture — something a fixed decision tree can't do.
Two ways to score
AI reasoning (default). The agent judges the answers against your objective and assigns the scores. Best when the signal is qualitative — clarity, readiness, fit — and you want the result to read like a consultant wrote it.
Deterministic points (optional). For any scale, yes/no or choice question, you can switch on scoring and assign points per answer plus a weight. Those combine into a fixed, auditable score. Best when you need the same answers to always produce the same number (lead-qualification gates, compliance).
You can use either, or both together — the deterministic numbers anchor the headline score while the agent writes the analysis around it.
Tiers and segments
The agent's tier_label (e.g. "Ready", "Stabilising", "Not yet") is your end user's segment. It shows on the result page, and it powers your segment breakdown analytics — so you can see how leads are distributed without configuring anything separately.
To tailor the outcome by tier, give the agent guidance on how each tier's result and call-to-action should differ. The single result page then adapts its message and CTA to the tier your end user lands in.
What this replaces
Most quiz tools use conditional logic: hand-written if/then rules that route people through a fixed tree to one of a few pre-written outcomes. Productised doesn't. The AI Product node already adapts its questions conversationally, and the AI Agent generates the outcome — so there's no branching tree to build or maintain, and no shelf of pre-written result pages to match against.