SAP reporting sits on four layers: data extraction, a small data warehouse, a central semantic model, and a BI tool like Power BI, SAP Analytics Cloud, or Tableau. The semantic model decides how much the setup costs to maintain.
A small warehouse and one trusted semantic model help keep reporting consistent and maintenance manageable. Business users can work with familiar names and agreed measures instead of navigating raw SAP tables. The choice of reporting platform matters, but the architecture behind it matters more.
Last week a client sent me a question I haven’t been able to stop thinking about.
“What platform are you using on top of SAP for the modelling and dashboards? We want something lightweight, low cost and low maintenance that covers our reporting, but still gives a less technical person room to run something ad hoc.”
It’s a fair question. I’ve also noticed that most people answer it by naming a tool. Power BI. SAP Analytics Cloud. Tableau. Whatever the vendor on the last call recommended.
My view is that the tool is the least interesting part of the answer. What decides whether SAP reporting stays cheap and easy to run is the bit in the middle, between SAP and the dashboard. And that’s the bit almost nobody budgets for.
So here’s roughly what I told them.
How SAP Reporting Usually Goes Wrong?
I have seen this play out more times than I can count. A team buys a BI tool, connects it more or less straight to SAP, and starts building reports. The first ten look great. Everyone’s happy.
Then you get to report forty. “Revenue” is now calculated four different ways. Someone hard-coded a company code filter into a measure six months ago and nobody remembers why. A column called VBELN has made its way into a report the CFO opens every Monday. And there’s exactly one analyst who understands how it all hangs together, and she’s just handed in her notice.
The BI tool did not cause any of that. Skipping the modelling layer did.
The Stack We End Up Recommending Most Often
For a mid-sized company running SAP that wants something light, this is usually where we land. It’s not exciting. That’s sort of the point.
Five-step low-maintenance SAP reporting stack
Get the Data Out the Right Way
CDS views, table or CDC-based extraction, or ODP OData. Pick the one your team can actually support, and write down why you picked it.
Worth checking now: if your current pipeline uses a third-party tool that pulls data through ODP-RFC, review it. SAP Note 3255746 prohibits that route into non-SAP targets, SAP has started blocking it technically, and the temporary opt-out ends with 2026. I’ve watched teams find this out the morning their nightly load didn’t run.
Build a Small, Clean Warehouse
You don’t need a huge platform for this. Fabric, Snowflake and Databricks all do the job, and honestly a well-run Azure SQL database is fine at this size. Most companies need four to six star schemas: sales, finance, inventory, procurement, maybe production.
This is also where SAP’s field names get turned into language people use. KUNNR becomes “Customer”. MATNR becomes “Product”. Business logic gets written once, here, and nowhere else.
If you only remember one thing from this post, make it this: “low maintenance” comes from this layer. Skip it and you’ll pay for it every month.
Publish One Semantic Model People Can Trust
One model per subject area. Somewhere around 20 to 30 measures that have been agreed and signed off. Names a finance manager would recognise without a glossary, and one definition of revenue. Every dashboard reads from it. Nobody builds their own logic on the side.
Give People Two Ways In
Dashboards handle the routine stuff, which in my experience is about 80% of what people need: the monthly pack, the sales review, the inventory check. For the other 20%, the one-off “can you just pull…” questions, people go to Excel connected live to the same model, or ask questions in plain English on top of it. Copilot-style tools get a lot more useful when they’re pointed at a clean model instead of raw SAP tables.
For most of our clients the front end ends up being Power BI. Not because it beats everything else, but because they already pay for Microsoft 365 and their people already live in Excel. Going with what people already use counts for a lot.
Set Up Monitoring Before You Go Live
Keep an eye on the things that can quietly break: failed loads, missing records, schema changes, refresh delays and unexpected spikes in data volume. Set up alerts so someone knows when a pipeline fails, rather than finding out when a dashboard shows yesterday’s numbers. Also, track refresh duration and data warehouse costs from the start. These are much easier to manage when you have a baseline to compare against.
When Staying SAP-Native Makes Sense
To be fair to SAP: Analytics Cloud, Datasphere and now Business Data Cloud are the right answer for some companies. Especially if planning and forecasting matter as much as reporting does, or if your landscape is almost entirely SAP and you’d rather have one vendor to call.
They usually aren’t the cheapest or the lightest option, though. If cost and simplicity are top of your list, run the numbers carefully before you sign anything.
Here’s how the two approaches compare:
| Comparison | Lightweight stack (e.g. warehouse + Power BI) | SAP-native (SAC, Datasphere, BDC) |
|---|---|---|
| Works best when | Reporting is the main need and people already work in Microsoft 365 and Excel | Planning matters as much as reporting, or the landscape is nearly all SAP |
| Strengths | Lower licence cost, familiar tools, flexible ad hoc access | Tight SAP integration, business context carried over, one vendor |
| Watch out for | Extraction compliance, and the discipline to keep one model | Licensing complexity and cost at mid-market size |
Ad Hoc Reporting for People Who Aren’t Technical
Self-service BI has been “just around the corner” for about 15 years. Mostly it’s disappointed, and I don’t think the tools are to blame. Companies tend to give business users either far too little (a locked PDF) or far too much (a few hundred raw tables and good luck).
What works is somewhere in between: a tidy model with plain names, sensible defaults and a short list of numbers everyone trusts. When someone can drag “Net Sales” and “Region” into a pivot table and get the same figure the CFO sees, the tickets dry up.
And please don’t fight Excel. Your finance team is going to use it whatever you roll out, so connect it to the governed model and let them work the way they already do.
What “Low Cost” Actually Means
Licences are almost never the big number. The big number is the hours spent keeping everything alive. Fixing extracts that broke overnight. Working out why two reports that should match don’t. Rebuilding a dashboard because the person who made it left. Maintaining 60 reports when 12 of them get opened.
So when someone asks me for a low-cost stack, I don’t ask which tool is cheapest. I ask how few moving parts we can get away with.
If I Were Starting from Scratch Tomorrow
I would avoid starting with a full SAP reporting overhaul. It’s tempting to move everything at once, but that usually means more dependencies, more testing and more things to fix. I’d rather get a small setup working well and build from there.
- Pick one or two subject areas. Usually sales and finance.
- Build the extraction, the model, and the semantic layer properly for just those.
- Ship five to eight reports that people will actually open.
- Give it a couple of months and watch how much rework they need.
- Not much? Add the next subject area. A lot? Fix the model before you build anything else on top.
It’s not glamorous, but the goal here is not to get every SAP report into a new platform as quickly as possible. It’s to build something people can rely on without needing constant fixes or a growing maintenance team. It’s not glamorous, but it’s the approach I’ve seen still standing years later.
Conclusion
SAP reporting doesn’t have to mean investing in another expensive platform or rebuilding your entire data architecture. For most mid-sized companies, the better starting point is a reliable extraction process, a clean warehouse, a trusted semantic model, and reporting tools people already know how to use.
Get these foundations right, and you can expand your reporting setup as business needs grow, without making maintenance a full-time job.
What are you running on top of SAP today? If you have made Datasphere or Business Data Cloud work on a mid-sized budget, or found a lighter setup that copes well with ad hoc requests, I’d like to hear how it’s going.
Frequently Asked Questions
What’s the best BI tool to use on top of SAP?
There isn’t one best tool. For most mid-sized companies already on Microsoft 365, Power BI on top of a small modelled warehouse works well. SAP Analytics Cloud, Datasphere, or Business Data Cloud make more sense when planning is central, or when almost everything you run is SAP. Either way, the modelling layer matters more than the front end.
Can we still extract SAP data with third-party tools after SAP Note 3255746?
Yes, just not over ODP-RFC into non-SAP targets. CDS view and table-based extraction, CDC approaches and ODP OData are all still on the table. SAP has started blocking unauthorised ODP-RFC calls and the temporary opt-out ends with 2026, so it’s worth checking what your current pipelines use.
How can non-technical users run ad hoc reports on SAP data?
Give them a curated semantic model with plain names and a short list of agreed measures. Then let them use Excel connected live to that model, or ask questions in plain English on top of it. Don’t hand them raw SAP tables.
What really drives the cost of SAP reporting?
Usually not licences. It’s maintenance: broken extracts, the same logic copied into lots of reports, reconciling numbers that don’t match, and keeping dashboards alive that nobody uses.