Agent Insights and settings
Measure the conversations, orders and revenue your shopping agent drives, and control its data, access rules and audit log.
Agent Insights shows what your shopping agent is doing: conversations, orders from AI assistants and the revenue they bring. Agent Settings controls what the agent keeps, who it serves and how far it can go.
Available on plans that include the Agent module. Every member can read these pages; only Owners and Admins can save changes.
How it works
Agent Insights
Open Agent › Agent Insights. Data comes from the agent's call log and your Shopify orders, which Opttab checks every 5 minutes (with a 30-day backfill when you connect).
Tabs: Overview, Performance, Live Activity, Outcomes and Promotions.
The Overview tab shows This week:
| Metric | What it means |
|---|---|
| Conversations | Shopper conversations with your agent. |
| Orders from assistants | Paid orders that came through a checkout handoff. |
| Conversion | Share of conversations that led to an order. |
| Agent revenue | Merchandise value of those orders, less discounts and refunds, excluding tax and shipping. |
Below it: Needs you today, Recent conversations, Conversations and orders, by week, Where customers meet your agent and Customer journeys.
- An order counts as Direct when it carries the reference from the agent's checkout handoff.
- By default you see live calls only. Use the Origin filter to include test and preview calls.
- Export downloads the open tab as CSV (up to 10,000 rows).
If nothing shows yet: "Your agent hasn't been called yet. Share its URL with an AI assistant to see activity here." To measure orders, connect your store.
Agent Settings
Open Agent › Agent Settings. Select Save changes after editing. Members can read the settings but not change them.
| Tab | What you set |
|---|---|
| General | Merchant (your workspace name), Currency (used for the order ceiling) and your role. |
| Data & Measurement | Call payloads on or off (off keeps only the call rows), and Call payload retention from 1 to 365 days (default 90). |
| Access Rules | The Effective permission matrix (unknown vs verified callers), Order ceiling (larger carts get no checkout link), Allowed markets (from Brand Hub) and Pause new agent effects. |
| Decision Simulator | Run simulation to test a request against your saved rules. Each run is recorded in the audit log. |
| Audit Log | Every change and simulation, with CSV export. |
Pause new agent effects stops journeys from activating or sending and stops promotions from starting. Your agent keeps answering shoppers.
Journeys and Playground
- Journeys (
Agent › Journeys) follow up after an agent conversation — for example, Recover a cart when a shopper leaves checkout, or Welcome a customer. Each journey must pass a test before you can Activate journey. - Playground (
Agent › Playground) lets you try your agent as each assistant serves it, with your connected store. Reads are real, writes are checked but never run, and no credits are spent.