# AI monetization tutorial

Everyone talks about monetizing AI, here is how to implement it.

This package gives you some concrete examples of how to monetize your company's new AI based features and solutions using Zuora. The intended audience are developers working in IT or engineering who are responsible for connecting the data pipes from your services and solutions to Zuora for billing and revenue recognition.

If you're unsure about pricing your company's AI features and products, we have [a useful step by step guide to help](https://ai-monetization-simulator.zuora.com/). With pricing decided this tutorial and the companion artifacts show how to bill for your company's AI features.

Once completed you will understand:

* How Zuora Billing can reliably bill consumption related events at scale.
* How other Zuora customers are using Zuora Billing to monetize the new AI features of their products and how to replicate their pricing and packaging.
* How Zuora Mediation and Billing can convert raw source consumption event data into verifiable invoice item charges.
* Determine if you need Zuora's Mediation add-on for your use cases.
* How to build custom Zuora Meters using Mediation for your unique use cases.


**Author's note**

This tutorial was written by humans for humans, though there are a couple of AI generated graphics. The "Measured add-on" use case has also been turned into an interactive, step by step wizard that you can run on your personal computer (requires node/npm), as well as a Jupyter Notebook for you to experiment with (requires python). The text in these artifacts was written by humans but the code came out of Opus. These artifacts are available in a zip file you can download from [this Zuora Community post](https://community.zuora.com/discussion/ai-monetization-implementation-kit?ReturnUrl=%2fcommunities%2fcommunity-home%2fdigestviewer%3fcommunitykey%3de2a932b4-50c4-4019-a3e8-362e38714df3). While we've included some AI prompts, you do not need AI or MCP to run or replicate any of these use cases, manual step by step instructions and video tutorials are included.

## How others monetize their AI features

Let's review what other customers are doing, today, to monetize their AI features. Below we'll detail how to implement each of these approaches, [Measured Add-On](/docs/get-started/tutorials/ai-monetization/measured-add-on), [Commitment](/docs/get-started/tutorials/ai-monetization/commitment-billing) and [Prepay](/docs/get-started/tutorials/ai-monetization/prepay-with-drawdown). We're skipping 'Baked In' as there is no billing impact, but you still need to discuss this approach with your Revenue team.

table
thead
tr
th
Approach
th
Examples
th
Description
tbody
tr
td
Baked In
td
n/a
td
See the earlier 
a
Revenue discussion
. No usage or mediation required. Often a place holder until the business figures out the COGS and the customer's perception of value. Used by some SaaS and Business Process Automation customers.
tr
td
Measured Add-on
td
$2.50 per conversation or $25/month per user with 1000 tokens included, then $0.10/token overage or $75/seat per month recurring charge
td
There's either a simple usage charge added, e.g. $2.50/conversation or an included amount of some AI consumable, for example, 'tokens', 'conversations', or 'answers', with the overage being billed for. Customers perhaps pay in advance for the included amount, overage is paid in arrears. Mediation needed if AI source data doesn't include billing identifiers such as account number. Used by many of our SaaS customers but a common example are US cell phone plans where the following month's bill includes the overage you incurred the previous month. Included amount is use or lose. Can be difficult to justify the implied equal value for all your users, you may experience push back and discounting
tr
td
Commitment
td
$100 commitment each quarter of AI feature use, measured as 'conversations'
td
Customers commit to a specific dollar amount, say $100 per quarter and if their actual usage of conversations at $3/conversation doesn't exceed that $100 commitment in that period a true up charge is added automatically at the end of the quarter. Mediation needed if AI source data doesn't include billing identifiers such as account number. Used by SaaS as well as PaaS customers. Can be similar to 'measured add-on' except there is a predefined customer specific minimum amount each customer will be charged regardless of their actual usage. Note billing is still in arrears, the true up comes at the end of the specified period.
tr
td
Prepay
td
Prepays $1000 each month for up to 100,000 tokens, optionally charge a dollar for every additional 150 tokens
td
Customers 
strong
pay up front
 for a wallet that stores cash or credits and Zuora tracks the spend. You either cut them off once the prepay is consumed or keep a payment method on file to bill them automatically for any excess. Mediation needed if AI source data doesn't include billing identifiers such as account number. Use automatically triggered feedback from Zuora when customer specific consumption thresholds are neared or exceeded. Used by Infrastructure, PaaS customers or other businesses that have a tradition of consumption billing.
## Zuora Mediation, Billing, and Revenue

### Zuora Mediation

**Zuora Mediation Pipeline - Event Data In - Money Out**

(inside one of our data centers)

Mediation simplifies the loading of consumption 'events' into Zuora. Source records are defined as **events** that pass through a **Meter** that has multiple '**Processors**' that filter, dedupe, aggregate, enrich and transform the incoming events into canonical Zuora Billing Usage records that are then consumed by our billing engine and **rated,** meaning quantities are converted to money, and placed on an invoice.

**Canonical Zuora Billing Usage Record (The 'Target'):**

* Unit of Measure - UOM, must be predefined in Zuora Billing, e.g. Token, Seat, User…
* Quantity - How much of the UOM was consumed
* Start Date - When was the quantity consumed and so what invoice to appear on
* Account Number - Billing Account Identifier
* Subscription Name - Optional, see note on Charge Numbers
* Charge Number - Optional since charges contain a UOM in their definition and Zuora will match UOM to charge automatically if this is unspecified. But if a single billing account has subscriptions that have two or more charges using the same UOM then the Subscription Name and Charge Number MUST be provided
* Additional Fields such as EndDate and Desc are available but have no impact on billing, you are welcome to use them for reporting or leave them undefined


While the example in the picture above uses Amazon's S3 as a source, in our examples here we'll use a simple csv file as the source so we can easily understand and test the meter. Other possible sources include Snowflake, Kafka or a streamable API endpoint to ingest events in real time.

### Zuora Billing

img
  
(A Zuora Billing Engine Room - the Zuora elves are on break)

Zuora Bill Runs pick up unprocessed usage records and generate invoices for the consumed usage. The customer's subscription defines the rating, or price charged for metered quantity. To be clear, consumption billing with Zuora does NOT require Mediation, you can load Zuora usage records without Zuora Mediation, but these usage records have to be in the canonical format defined earlier.

### Zuora Revenue

Even if you don't charge for your AI features, 'free' add-ons can incur revenue performance obligations that must be accounted for. Your Finance and Revenue team must know how you are planning on deploying or selling your AI products.

You may need to create new charges in the Zuora product catalog to capture sales of AI features and products. These charges capture flags needed to ensure revenue is recorded correctly. Again, you must involve your Finance and Revenue team to ensure these are configured correctly.

## Mediation decision tree

Decide if you need Mediation.

table
tr
td
Multiple raw usage sources to merge? (APIs, meters, IoT, CDRs)
td
Yes ➔
td
Yes, you need Zuora Mediation
tr
td
No ↓
td
tr
td
Need transformation, deduplication, enrichment, filtering or aggregation applied before billing?
td
Yes ➔
tr
td
No ↓
td
tr
td
Need near-real-time processing that scales automatically on our infrastructure instead of yours?
td
Yes ➔
tr
td
No ↓
td
tr
td
strong
Use Direct Usage Record Import
td
## AI monetization pre-launch checklist

Before diving into implementation, use this checklist to ensure you have everything in place:

**Have you identified:**

1. Is Mediation enabled in your chosen sandbox tenant? Look at the left hand menu in the UI, 'Mediation' should be an option. If it's missing contact your Zuora Administrator or Zuora Account Team.
2. A charge model, how will the captured quantities be rated, on a per unit basis? Tiered pricing? Flat Fee? Volume?
3. Variable vs recurring - will you charge the same each month or bill variable amounts based on the actual events recorded?
4. Pay in advance or in arrears? Will you collect the money up front as a credit or let your customer generate events and charge them later?
5. The unit of measure(s) involved?
6. Is Zuora Mediation needed? And what transformations are needed to convert the source event data to a canonical Zuora Usage Record?
7. Did you create a rate plan and a billing account and subscription with that rate plan?


## Ready to implement?

Once you've completed the pre-launch checklist, choose your monetization approach and start implementing:

- **Measured add-on** — [Start implementation](/docs/get-started/tutorials/ai-monetization/measured-add-on) or [See interactive demo](/docs/get-started/tutorials/ai-monetization/interactive-walkthrough)
- **Commitment billing** — [Start implementation](/docs/get-started/tutorials/ai-monetization/commitment-billing)
- **Prepay with drawdown** — [Start implementation](/docs/get-started/tutorials/ai-monetization/prepay-with-drawdown)


**All approaches need these first:** → [Basics for all examples](/docs/get-started/tutorials/ai-monetization/basics-for-all-examples)