Every SAP team has the queue. Requests for a report that shows almost what an existing report shows, but grouped differently, or for last quarter, or excluding one plant. They arrive faster than they can be built, and the backlog is treated as a capacity problem.
It is not a capacity problem. It is a shape problem.
Why the queue never clears
A report is an answer, frozen. Somebody had a question, and the answer to that question was compiled into an object that will be maintained for years.
But questions are not stable. The person who asked “revenue by sales organisation, this quarter” asks, the moment they see it, “and how does that compare with last quarter?”, which is a different report. Then “which customers drove the difference?”, a third. The original request was one question in a chain, and the chain does not terminate.
So the queue is not a stock of work to be cleared. It is a flow rate, and it is a function of how many people have questions.
The three ways teams try to fix it
Build faster. Hire, template, industrialise the report factory. This lowers the cost per report and therefore raises demand, because the internal price of asking went down. The queue length stays roughly constant and the maintenance burden grows without limit.
Push it to self-service BI. Extract the data, put a tool on top, let the business serve itself. This works until month end, when the extract layer and SAP disagree and the reconciliation lands back on your team, now harder, because there are two versions of the truth and one of them has a dashboard.
Refuse. Tighten intake, demand business cases. The requests do not stop; they go underground into spreadsheets maintained by one person in each department, which is the same reporting estate with none of the governance.
What actually changes the shape
The queue exists because asking is expensive and answering is a project. The only structural fix is to make asking cheap, which means the business asks the system directly, and the answer is computed rather than built.
Three properties decide whether that is an improvement or a new incident class:
It has to read the real system. An extract is a second source of truth, and a second source of truth is what you were trying to avoid.
Every figure has to carry its source. A number without provenance is not an answer, it is an assertion, and an assertion at scale is worse than a slow report, because nobody can check it and everybody will act on it.
It must not be able to change anything without a person. The moment a system can write to your ERP autonomously, every governance conversation you have ever had reopens at once.
What this leaves your team doing
Not nothing, but different work. The catalogue, the authorisations, the business rules that decide what a figure means, the policy that decides who may do what. Those are the things that genuinely require your team, and they are currently the things that get least attention because everyone is busy building reports.
The backlog was never the job. It was what the job had been reduced to.