Grow or Die Connect my product

Integrations

Connect revenue, traffic and AI cost data

Grow or Die keeps three jobs separate: analytics counts people and product activity, billing records money, and the server SDK records model usage. Contribution only appears when the populations can be compared honestly.

AnalyticsWho used it
+
BillingWho paid
+
SDKWhat AI cost

What do I actually need to connect?

  1. Visitors: install your site's browser snippet. It observes browser activity, not server model usage. GA4 and PostHog are optional analytics sources, not prerequisites.
  2. AI cost: use the server SDK around an existing model call and verify that its final usage reaches the same GOD website. A browser snippet alone cannot see your backend's token usage.
  3. Revenue and customer contribution: connect a supported payment source, then link its payer ID to the same verified internal account ID used for AI calls. Revenue alone does not prove which customer incurred a model cost.

X mentions start from your website and do not require an SDK. For analytics, verify each source separately with the native SDK acceptance checklist; if costs are blank, use its no-data troubleshooting guide.

Choose a source

Read exactly what the calculation needs.

Each page documents the current connection method, data boundary, and limitations before you share a credential.

X mentions

Starts from your website

Find posts mentioning your website or verified product name without an SDK. Read the matching rules, search limits and saved-history boundaries.

Model usage

OpenAI

Capture provider-reported usage through the private server SDK. Prompts, responses, and API keys stay out.

Model usage

Anthropic

Measure Claude token usage, caching, retries, latency, and status from your own backend.

Revenue

Stripe

Verify a restricted read key and import observed payments, refunds, fees, and payer IDs.

Revenue

Lemon Squeezy

Import selected-store orders and subscription invoices, refund snapshots, and customer IDs. Processor fees are unavailable, not zero.

Product analytics

PostHog

Use observed visitors, events, acquisition dimensions, geography, and supported funnel definitions.

Product analytics

Google Analytics

Connect a GA4 property for visitors, sessions, acquisition, geography, and key-event context.

Revenue

Polar

Read one organization's paid orders and refund snapshots. Currency and unavailable fees remain explicit.

Search performance

Google Search Console

Read clicks, impressions and search queries for a matching property you authorize. This is not a search-volume estimate.

Revenue basis depends on the source

Revenue minus priced AI usage is observed contribution, not accounting net profit. Missing processor fees are unknown, not zero. Never compare different currencies or subtract a refund twice.

Stripe

The restricted read connection can use balance-transaction fees and dated refund events. Contribution still requires complete imports, the same reporting window and usable model prices.

Stripe permissions and setup

Lemon Squeezy

Reads the selected store's paid orders and subscription invoices, avoiding duplicate initial invoices. Revenue uses USD amounts after tax and refunds, before unavailable processor fees. It is not a payout ledger.

Provider refund time is used when available; an updated snapshot or order-time fallback is approximate. Truncated imports or incomplete evidence cannot support confident contribution decisions.

Lemon Squeezy boundaries

Polar

Paid order amounts are adjusted for tax and observed refunds. Processor fees are unavailable. Refund dates may be inferred from modified order snapshots; currencies remain separate, with no automatic FX conversion.

Polar setup and limitations

A concrete boundary example

If imported revenue after tax and refunds is $100 and priced AI usage is $12, the observed difference is $88 before any unavailable processor fees and other expenses. Unknown fees do not prove $88 net profit. If a refund timestamp is approximate or an import stopped before all pages were read, the period is partial—not a complete reconciled result.

Polar: organization-scoped revenue

Status: implemented connector. In the website's integration settings, choose Polar, select production or sandbox, and provide an organization access token with organizations:read and orders:read. Verification resolves the organization; sandbox data is not live revenue.

The encrypted connection reads paid orders for that organization, customer IDs and available refund amounts. It does not create charges or change subscriptions. Pagination is bounded, so coverage can be partial. Missing fees and inferred refund timing are reported as limitations; this is not a merchant payout reconciliation. Disconnect in settings and revoke the token in Polar when it is no longer needed.

See the Polar order reference and customer identity requirements before combining revenue with server usage.

Google Search Console: authorized search evidence

Status: implemented OAuth connector. Open Google Search Console in integration settings and authorize Google access. The connector resolves an accessible property matching your website. The connection requests Google's webmasters.readonly scope; it does not grant site ownership or submit indexing requests.

Clicks, impressions, CTR, position and available query/page dimensions come from the selected property. Google reporting lag, privacy filtering and row limits mean this is not every query or real-time traffic. Search Console dates use Pacific Time; analytics may use another reporting time zone. Disconnect in settings or revoke access in your Google account.

GSC is separate from GA4 visitors and public DataForSEO keyword estimates. Read the official Search Analytics query limitations.

A safe first result

Start site-wide. Add customer links when you are ready.

Revenue per visitor and site-wide AI cost do not require sending customer identity. Customer-level profit needs one explicit, server-side identity link.

See customer profit