GVGanivra

Multi-Tenant B2B SaaS

Your biggest users can be your smallest margins.

Two customers on the same plan can generate very different AI costs. Connect model and MCP usage to your existing tenant IDs to see which accounts earn their margin and which need a closer look.

The margin risk

Flat revenue. Uneven consumption.

A power user may run longer prompts, invoke more tools, or automate hundreds of tasks while paying the same subscription as a light user. Account growth can therefore increase AI spend faster than revenue.

Tenant IDModel & tool usageTracked costRevenue & margin

Attribute

Use the tenant ID you already have

Read the tenant or account ID from your trusted backend request context and send it as customer_id. Attach it consistently to model and MCP events across the same execution.

New IDs appear automatically in customer breakdowns. No separate customer registration is required, and a tenant ID is not a separate Ganivra workspace.

Inspect

Find what makes the account expensive

Select a customer in Cost breakdowns, then inspect its features, workflows, models, and executions. Distinguish high-volume use from unusually expensive calls before changing pricing.

Review missing pricing and attribution first. An unpriced call is not free, and unattributed usage can hide the cost of a tenant.

Connect

Add what the customer pays

Configure customer pricing in Profile & setup. Send the configured plan_key on executions or assign the tenant to that plan. Import confirmed revenue when you need billing actuals.

For credit-based plans, send the actual units deducted using the configured usage attribute. Keep rates in the pricing configuration rather than calculating them inside every model call.

Decide

Evaluate pricing that fits consumption

Use customer margins, cohorts, and pricing headroom to decide whether to retain a subscription, charge for a valuable action, or combine a base fee with usage charges.

Ganivra supports subscription, execution, token, unit, and hybrid revenue calculations. Implement checkout, included allowances, overage rules, and customer limits in your billing and application systems.

Illustrative customer comparison

The same $100 plan can tell two different stories.

CustomerMonthly revenueTracked AI costContributionMargin
Typical account$100$20+$8080%
Heavy automation account$100$140−$40−40%

Invented figures illustrate the calculation. The second account needs investigation: optimize an expensive workflow, revisit its commercial terms, or test an appropriate usage charge. Heavy use alone does not mean the customer should pay more; consider the value delivered and the cost drivers.

Integration essentials

Give each call the context finance needs.

Who generated the usage?customer_id

A stable pseudonymous tenant ID, populated by your backend. Use the same ID on related revenue records.

What generated the cost?application + feature + workflow

Separate products and AI features within one workspace. Keep labels consistent across model and tool events.

How does that customer pay?plan_key

The customer pricing plan configured in Ganivra. This is separate from your company's Ganivra subscription.

Compare like-for-like periods

Use the same date range when reviewing account economics. Check revenue coverage and reported costs before interpreting margins. Headroom is an estimate under its stated assumptions, not a forecast or an automatic pricing recommendation.

Connect customer revenue →

Start with one tenant

Follow its usage from feature to margin.

Capture real calls, verify attribution, and connect customer pricing. Expand to the rest of your accounts once the numbers make sense.

Open integration guide →