For FP&A and Finance · monthly

Your monthly customer profitability review, rebuilt for AI products.

AI cost per customer isn’t on any invoice, and it moves every time a prompt, model or agent step changes. AI Margin Review joins usage, rates and revenue, flags customers below your margin target, and explains what changed, with the evidence saved for the meeting.

Set up your first margin review See how it replaces the spreadsheet ↓

Self-service setup · 14-day free trial · No credit card

ILLUSTRATIVE REVIEW / ACME01 OF 4 FLAGGED

Acme is below your margin target.

What needs attention: 58% contribution margin, below the 65% target.
What happened: $50,000 revenue − $9,000 recorded AI cost − $12,000 other included delivery costs = $29,000 contribution.
What explains the AI cost change: Executions rose from 100,000 to 150,000 and cost per execution rose from $0.03 to $0.06. Of the $6,000 increase, $1,500 came from volume and $4,500 from cost per execution.
What remains uncertain: The records don’t show why cost per execution rose, or whether the extra cost improved outcomes.
What should happen next: Product owner inspects the workflow change before deciding whether implementation or contract terms need attention.

Saved with source records and calculation method · Example only

Made for the monthly customer profitability reviewFinance leadsFP&AAI product ownersOnly for companies that ship AI products and features

Why your current review breaks

Profitability reviews assume cost per customer is stable.
For AI products, it isn’t.

No invoice shows cost per customer.

Providers bill by model and month. Someone has to join usage to customers by hand before the review can start.

Cost changes without a price change.

A new prompt, a model swap or an extra agent step can double what a customer costs to serve. Their plan price stays the same.

“Why did it go up?” has no answer in a spreadsheet.

A higher bill could mean more usage or more cost per use. Those need different fixes, and the spreadsheet can’t tell them apart.

The review you already run

You bring the data.
Ganivra does the prep.

Keep your financial judgment. Hand over the repeatable checks, calculations and evidence assembly behind it.

Today, by handWith AI Margin Review

Check whether the spreadsheet is complete

A readiness checklist finds missing customer attribution, prices and revenue, and links to the fix.

Recalculate contribution for every customer

Contribution and the gap to your target for each customer, using the period and costs you approve.

Rebuild the explanation for a cost change

Split each change into volume and cost per execution when comparable history is available.

Assemble the review and chase its evidence

Save findings with calculations and source records, then record owners, due dates and decisions.

Everyone in the room

One review. A different answer for each person at the table.

FP&A

The numbers, already prepared

Contribution per customer, readiness gaps closed, calculations saved.

CFO

Margin against target

Which customers are below target, by how much, and on what cost scope.

Product owner

What drove the change

Volume or cost per execution, and the workflow behind it, with an owner and due date.

CEO · Founder

What to do about it

Which customers to reprice, which workflows to fix, which contracts to revisit.

Getting set up

Start with one period.
Finish with a review.

01
Define the scope
Customers, period, revenue basis, included costs and a contribution target.
02
Connect your records
Use the SDK or API and financial imports, or reuse data already in your workspace.
03
Resolve the gaps
Follow readiness checks for attribution, pricing and revenue. Acknowledge any cost-scope limits.
04
Review and act
Generate the review, inspect the evidence, and record decisions, owners and due dates.

What you bring

Model and tool usage linked to customer and execution IDs.
Cost rates or provider-reported charges.
Customer revenue or configured pricing for the same period.
Additional delivery costs, or an explicit decision to exclude them.

What’s not included

Scheduled delivery, external task routing, ERP posting and automatic usage blocking. Recorded contribution is not accounting net profit. Missing required data blocks generation; finding no customers at risk is a valid result.

AI Margin Review

Walk into next month’s review
with the answer, not the spreadsheet.

Create your account, connect your records, resolve the readiness checks and generate your first review, all in your workspace. No onboarding call required.

Set up your first margin review ↗Already have an account? Sign in →