Export published geo, user-level, and platform test results into your warehouse, BI, notebooks, and client reporting. Lift, iCAC/iROAS, intervals, diagnostics, and model agreement move with the result through a read-only REST API.
Read-only by design. It exports results and never writes, launches, or changes anything. Permission-controlled portfolio keys can be rotated or revoked anytime.
GET · api.shakostats.com/api/v1/agent (read-only)
{
"results": [
{
"product": "geo",
"name": "Meta Prospecting · Q2",
"channel": "Meta",
"lift_percentage": 11.4,
"icac": 41.2,
"iroas": 3.6,
"p_value": 0.02,
"published_date": "2026-05-18"
},
{
"product": "user_level",
"name": "Lifecycle Email Holdout",
"channel": "Email",
"lift_percentage": 6.1,
"icac": 12.8,
"iroas": 9.4,
"published_date": "2026-05-02"
}
],
"count": 2
}Illustrative response — the same numbers your dashboard shows, returned as clean, read-only JSON.
Shako Stats runs the measurement models and computes every result. The API hands those published results back over plain HTTP, so the tools your team already runs can read them directly.
Published test results belong in your warehouse, BI, notebooks, and client reporting, not locked behind one vendor login. The API exports Shako's computed results so your team can analyze, share, and operationalize them anywhere.
A single endpoint returns a flat feed across geo, user-level, and platform tests with lift, iCAC, iROAS, intervals, p-values, dates, and model agreement already computed. No manual export stitching.
Read results into pandas, sync them to BigQuery, Snowflake, or Redshift, or wire them into Looker, Tableau, Power BI, and Sheets. Plain JSON over HTTP fits the workflows your team already runs.
The API exports results. It never writes, launches, or changes anything. Designing tests, running results, and publishing stay in the app, so a pipeline cannot touch your measurement workspace.
Personal keys can follow a member's authorized organizations; account service keys use a fixed child-organization scope. Both can be narrowed to companies. Every request selects an organization_id, and broad keys stay in a trusted backend.
The same key can power the MCP server, so your AI agent and your data pipeline read the same approved Shako context. Pull results into your stack and ask questions about them in plain language.
Create a key, make a call, and load results wherever your reporting lives. No new tools to learn; it's HTTP and JSON.
Organization members with Use Data API permission can generate a personal key under AI MCP / API Access. Scope it by product, organization, and company. Authorized billing-account integration managers can issue a separate account-owned service key for durable automation.
Add your key as a Bearer header and pass organization_id for the tenant you are reading. Optionally pass company_id to narrow collection feeds to one brand. Plain HTTP works with curl, requests, fetch, or a scheduled job, with a typed OpenAPI spec for generated clients.
Get a flat feed of published results across every method, with lift, iCAC, iROAS, and dates. Or read a single test in full, down to designs, diagnostics, and model agreement.
Land it in your warehouse, refresh a BI dashboard, drop it in a notebook, or fold it into a client report. Schedule the sync and your reporting always reflects the latest tests.
No proprietary SDK, no connector to wait on. If it can make an HTTP request and read JSON, it can read your measurement results.
The /published-results feed returns one flat row per result, ideal for a warehouse table or a BI dashboard.
Exporting results shouldn't mean handing over the keys to your workspace. The API is built to pull published results out safely and predictably, over and over.
Every route is a read. The API never creates, launches, or mutates anything. Designing tests, running results, and publishing stay in the Shako Stats app, so your pipeline cannot break your workspace.
Personal keys follow the owner's organization permissions; account service keys use a fixed child-organization scope. Both can be narrowed by product and company. Keep broad credentials server-side, and rotate or revoke anytime.
The API reads the exact published results your workspace computed. It gives the app, your AI agent, and your pipeline one source of truth. Read a result in full and the model evidence travels with it, including intervals on every estimate.
Stable JSON, consistent pagination, clear error codes, and a token-gated OpenAPI schema for generated clients and scheduled ETL. Built to run unattended, not just to demo once.
Brands, consultants, and agencies who run their own numbers. Whoever owns the warehouse, the BI dashboard, or the client report gets the results in a form they can build on, with no exports to chase and no login to share.
A proper read-only interface over your measurement workspace: the same numbers your dashboard shows, served in a form your pipeline can reason over on a schedule.
Your geo, user-level, and platform tests all live in one workspace and feed one central database. The API reads across all of them, so a single pull gives you a complete, cross-method view of what your spend actually caused, ready to load wherever your reporting lives.
Inform. Attribution gives the MMM a fast, directional read on what's working between tests.
Calibrate. Incrementality tests anchor the MMM to causal ground truth, not just correlation.
Validate. Holdout tests check that attribution and the model agree with what really happened.
Prefer to ask instead of pull? The same Shako context powers your AI Marketing Scientist. Connect any AI agent through MCP/API with one key and ask about test design, power, metric definitions, iCAC, marginal CAC, and model agreement in plain language, grounded in the exact results your pipeline reads.
Meet your AI Marketing ScientistOne-click OAuth for Claude & ChatGPT, read-only, anyone on the team can ask
The models and tests matter, but the workflow around them matters too. Shako Stats is designed to become the operating system around experiment planning, metadata, documentation, and cross-test learning.
Centralize datasets, mappings, and historical records so experiments and models always start from the same source of truth.
Organize tests by audience, creative strategy, bidding logic, or business objective so learnings remain searchable and reusable.
See what tests are planned, in-flight, or completed so overlapping interventions and measurement conflicts are easier to manage.
Turn methodology, definitions, and experiment design guidance into an internal operating system instead of leaving them scattered across decks.
Use the docs and examples to get your first pull landing in your warehouse or dashboard.
Paid B2B plans built for brands & agencies.