A guide to navigating dashboards (and their blind spots), consumption, and billing
There is a question that is giving quite a few IT departments a hard time this year: how much is artificial intelligence in SAP costing and what will it cost? But not only that: for what usage and by whom?
The answer exists but is distributed across different tools, each with a slice of truth and a rather precise blind spot. And those who govern the budget therefore find themselves in the most uncomfortable position that exists: a lot of data, little answer. In this article we try to bring order by seeing together where these slices are and what needs to be done to reconstruct the pie.
For those who do not deal with SAP licensing on a daily basis, let's start with the basics.
The SAP commercial model separates Base AI (included in the standard cloud subscription, usable without limits and at no additional cost) from Premium AI usage, powered by AI Units.
AI Units are therefore the virtual currency with which everything in SAP bearing the "premium artificial intelligence" label is paid for, regardless of the tool. They are purchased annually, end up in a central pool that can be spent cross-functionally across various services, and expire after twelve months. We also talked about it here.
The currency is unique, but the wallet is only one per tenant. The balance lives at the subscription level, not at the user level, not at the department level, not at the cost center level. No native link between "this unit was burned by Marco from controlling during the September closing".
To better understand the mechanism, think of a tourist resort's prepaid wristband. You load an amount, you deduct it on very different services, each with its own price, and at the end of the period what you haven't spent does not come back. With an additional difficulty: the wristband in the example is only one for the whole company. The bill arrives aggregated and the fourteen extra mojitos do not have a name on them.
Let's now understand it by putting together "the receipts", i.e., the data available in the various tools.
Tool 1: The Business AI Consumption Dashboard in SAP for Me
It is the financial dashboard, the one that answers the question "how many AI Units are we consuming as an organization". It is present in SAP for Me in the Finance & Legal area.
What it shows you:
What it doesn't show you:
In summary: you know how much you spent and on which feature, but not who to send the bill to.
Two points of attention:
Tool 2: The Joule Analytics Center
Here the register changes completely. The Joule Analytics Center (accessible with the analytics_admin role) is an interactive dashboard that measures adoption. It is the right tool to understand if Joule is really being used or if it is yet another ignored icon, to discover which scenarios work and where there are knowledge gaps.
What it shows you (filtered by interaction type, product, scenario):
What it doesn't show you: consumption in AI Units.
No, this is not a joke: SAP puts it in black and white in a dedicated KBA, where it confirms that to date Joule usage data is not displayed in SAP for Me's AI Units consumption tracking (source: SAP KBA 3604251, "How to monitor Joule Consumption").
To summarize the situation with a certain brutality: you have one tool that tells you how many messages have passed and another that tells you how many units have left the pool, and the two do not talk to each other.
Tool 3: BTP Cockpit
If your AI perimeter includes services on BTP, the BTP cockpit adds the operational layer, with monitoring of costs and usage by global account, directory, and subaccount, thus with the possibility (if you have structured the subaccounts well) to separate development, testing, and production.
It is the most granular tool of the three as well as the one that brings you closest to a sensible internal allocation.
However, in the periodic use of this tool (both in BTP Cockpit and in SAP for Me) we have noted a delay between 24 and 72 hours in consumption data.
We were not the only ones to notice it, but there is no official information regarding this. However, we recommend not considering this information as real-time because in case of consistent batch loads the balance you saw this morning may describe a situation that no longer exists.
The fourth tool: The Feature Estimator
Then there is a final piece, which is on the other side of the timeline. The SAP Business AI Feature Estimator (indicated by SAP in KBA 3580451 precisely as an answer to the question "how many units do I need") does not look at the past but tries to tell you how much a feature will consume before you turn it on, feature by feature, action by action.
A tool (like others) certainly to be used before signing the contract, but it is an estimate built on volumes that you declare, so it is worth exactly as much as the quality of your assumptions.
The picture, at this point, speaks for itself: SAP provides you with a tool to estimate in advance, one for ex-post accounting, one to understand adoption, and one to monitor the infrastructure. None of the four attributes consumption to a department.
The summary, in short
Four partial answers which, added together, do not yet make a chargeback because a decent internal chargeback requires three pieces of information simultaneously: how much was consumed, by which feature, and on behalf of which business function.
SAP delivers at most two, never the same two in the same place, and with a temporal misalignment between the balance updating on the backend and the monthly report finalized on the client side.
The concrete risk is not knowing consumption until the bill arrives. And when it arrives, the discussion with business functions opens on an aggregate number that no one can contest or claim. Ideal situation for arguing, but terrible for deciding.
How to patch it up (seriously)
Countermeasures are not exotic, they are just boring, and that's why almost no one implements them before receiving their first surprise bill.
What is needed?
All very nice... but who does it?
If you want, WEGG can do it for you, it is exactly the perimeter we work in.
Applying a FinOps methodology to AI consumption in SAP means giving every consumed unit an owner, a trend, and an alarm threshold, instead of a monthly number to endure.
If you want to avoid turning three dashboards into a three-month internal project, our FinOps consultants do that for a living: transparency on "how much" and "who" exists.
Only that in this case, unfortunately, it must be built by hand by cross-referencing sources that SAP keeps separate and it is not data handed over to you ready-made.
Sources:
SAP Community, Joule Analytics Center: Your Go-To Tool for Tracking SAP Joule Usage
SAP Help Portal, Joule Analytics Center
SAP Help Portal, Reviewing Consumption Summary of AI Units (SuccessFactors Platform)
SAP Support, Business AI Tab (SAP for Me)
SAP KBA 3604251, How to monitor Joule Consumption
SAP KBA 3580451, How to monitor AI Units Consumption for Premium AI features in SAP SuccessFactors
SAP Learning, Evaluating the Commercial Model Article by Jary Busato, MBA , SAM/ITAM&FinOps Consultant at WEGG
Insights
OUR OFFICES
OUR OFFICES
PADUA
Via Arnaldo Fusinato 42, 35137
MILAN
Viale Enrico Forlanini 23, 20134
ROME
Viale Giorgio Ribotta 11, 00144
Copyright © 2025 WEGG S.r.l. • P.I 03447430285 • C.F. 02371140233 • REA 311023
Azienda Certificata ISO 9001:2015 – ITA / ISO 9001:2015 – EN
