Audience Loop (iCustomer)

Category: Lead Gen/Data

Warehouse-native audience building, callable from Claude.

Audience Loop is the self-serve entry point to iCustomer rather than a separate product, so expect the wider platform language once you are inside. The job it does is building and activating audiences on top of your own data warehouse, Snowflake, Databricks or BigQuery, without copying data out. Named components include OneSource for identity resolution, a Signals Hub unifying first, second and third-party signals, and FIRE scoring across fit, intent, recency and engagement, with a signals waterfall behind it. Audiences push out to the ad platforms and sales tools already formatted, and the pitch is far higher match rates than pushing a raw list. Around that sit role-based agents for marketing ops, lifecycle and analysis, an orchestrator with guardrails, packaged reusable campaign plays, causal incrementality testing with holdouts, and a per-decision audit log. Pricing is credit-based with visible burn and caps, no per-seat charge and unlimited users even on the entry tier.

Our verdict

The MCP story is the reason this is here. The server exposes audiences, scores and decisions as governed tools any agent can call, with documented paths for working from Claude Code or Codex, querying audiences, drafting decision loops and deploying via CLI with everything versioned in git. For a marketer who has moved into that way of working, no other audience platform offers it. Warehouse-native with no data copies is also the right architecture, and SOC 2 Type II plus the cloud partner badges suggest it is built properly. Two things to set expectations on. The naming is confusing, as iCustomer says plainly that Audience Loop is not a separate product, it is one of three ways into one platform, so do not expect a standalone tool. And the enterprise tier needs a warehouse, which puts it out of reach for most smaller UK teams; the $99 and $499 self-serve tiers are the realistic entry. The vocabulary is heavily proprietary, so budget some time working out what each named component actually does.

Best for: Data-literate marketing teams who already have a warehouse and want audience building, scoring and activation governed in one place rather than exported between tools.

Pros

• MCP server exposes audiences, scores and decisions as governed tools any agent can call
• Documented Claude Code and Codex workflow, deploy via CLI with everything versioned in git
• Warehouse-native on Snowflake, Databricks and BigQuery with no data copies
• FIRE scoring across fit, intent, recency and engagement, with a signals waterfall behind it
• No per-seat charge and unlimited users even on the self-serve tier
• Causal incrementality testing with holdouts, and a per-decision audit log

Cons

• Audience Loop is a tier of iCustomer, not a standalone product, which the naming obscures
• The enterprise tier needs a data warehouse, so it is out of reach for most smaller teams
• Very heavy proprietary vocabulary, so expect a learning curve working out what is what
• Match-rate claims are vendor figures with no published methodology
• Early-stage funding against enterprise-scale positioning

Pricing: Free to start, then from $99/mo (Freemium)

Rating: 4/5