Let's talk






Banking & withholding tax
We support banks, asset management companies (KVGs) and custodians through the rollout of systems for withholding-tax reclaim, relief at source and reporting obligations — from the as-is analysis until the team trusts the system. For vendors of banking and tax software we deliver the product UX.
Beabstracto is a UX studio in Munich for complex expert systems. For banks, asset management companies (KVGs) and custodians we support the rollout of software for withholding-tax reclaim, relief at source and reporting obligations such as MiKaDiv: as-is analysis, functional concept, UAT, go-live and hypercare, including key-user enablement and training at the desk. For vendors of banking and tax software we deliver product UX: research with expert users, status models, UI definition and usability testing. First findings from the as-is analysis are usually available within a few weeks. We only ever work for one side per project.
On a project we take the role of UX Consultant or Business Analyst with a focus on UX and rollout. We are independent of every software vendor and sell no licences.
Rarely because of the software, almost always at the handover into the business unit: the process from entitlement to claim is not mapped in the system the way the team actually runs it.
The licence has been running for months, but not a single reclaim goes through the system in production. The business unit is waiting for the interface, IT for the functional concept, the vendor for feedback.
The back office or fund accounting does not deliver positions and payments the way the reclaim process needs them. Nobody translates between business unit, development and vendor, and every query costs a week.
Certificates of residence, powers of attorney and tax vouchers are tracked in a parallel list because the system does not show status, deadline and next step. The list is the real expert system — without four-eyes approval or audit trail.
MiKaDiv applies to capital income received after 31 December 2026. The date is fixed, but who delivers which data from which level of the custody chain is still unresolved in-house — and the reporting module has not arrived in daily work.
Two days of training in a meeting room, then go-live. The key users know the functions but not the exceptions: partial refunds, expired certificates, corrections after submission.
On the vendor side: a new reporting or reclaim module is technically correct, but customers need long onboarding and report questions instead of bugs.
Regulatory dates and experience values we rely on — only with a verifiable source.
BZSt, MiKaDiv reporting procedure
Council Directive (EU) 2025/50, OJ L of 10 January 2025
McKinsey, “Losing from day one”, December 2021
Steam-turbine manufacturing case (Siemens)
Two services, cleanly separated: implementation support for banks, KVGs and custodians — and product UX for vendors of banking and tax software. We always work for the side that hires us; information from one engagement never flows into another.
We accompany the business unit — tax, fund accounting, operations — from the as-is analysis until the team retires the parallel list. We capture the processes as they are actually run, from entitlement to claim, write the functional concept in the language of both sides, prioritise by deadline and translate between business unit, IT and vendor.
We accompany the business unit — tax, fund accounting, operations — from the as-is analysis until the team retires the parallel list. We capture the processes as they are actually run, from entitlement to claim, write the functional concept in the language of both sides, prioritise by deadline and translate between business unit, IT and vendor.
For product teams whose users are tax specialists, fund accountants and operations teams. We make implicit domain knowledge explicit — personas, business processes, critical moments, failure mapping — and translate it into status models, information architecture and interface behaviour. We do not hide complexity; we make it manageable, auditable and safe.
The business unit’s — not an agency’s. These terms decide whether a concept holds up in the tax team.
The sequence is the same in every rollout; the scope depends on the institution, the system and the deadline. Every phase delivers a result for the business unit.
We sit next to the case handlers, not in a meeting room: which claims, which deadlines, which exceptions, which Excel lists. With the head of tax or fund accounting and the rollout project lead. Result: as-is process, parallel lists, media breaks, first risk list.
usually a few weeks
We map the target process from entitlement to claim — relief at source, quick refund, standard refund, correction — and phrase requirements so that business unit, IT and vendor mean the same thing. Whatever is deadline-critical first gets priority. Result: functional concept, status model, prioritised requirements list.
We translate queries in both directions, bundle change requests with context and write test cases from real transactions. In user acceptance testing the key users test, not us. Result: test concept, UAT log, acceptance decision.
We enable key users to explain exceptions and error cases, and we train at the desk with real cases. The cutover checklist is built with IT and the vendor. Result: key-user concept, training materials, go-live checklist.
After go-live we stay alongside: first deadlines, first corrections, first report. We resolve queries with the key users or pass them to the vendor in bundled form. Result: hypercare log, lessons learned, handover to regular operations.
At the start we agree on metrics — claims processed on time, queries to the vendor, remaining parallel lists — and check them at the end of hypercare.
From UX consulting for the vendor of a withholding-tax reclaim software and from years in steam-turbine manufacturing — no client data, clear lessons.
In an engineering-led organisation, domain knowledge sat with developers, product owner and management — implicit and unprioritised. Only a glossary, taxonomy and regulatory mapping made personas possible; personas made business processes visible; processes revealed critical moments.
Expert users in withholding tax and reporting are specialists, not casual users. They need validation logic, audit trails and evidence visible — not an interface that smooths everything over. The goal is auditable and safe, not “simple”.
As long as a reporting module is discussed as a collection of screens, it stays unclear what it is optimising for. Only the whole path — find, validate, correct, approve, track — shows where the system must not fail.
In steam-turbine manufacturing the core problems were not found in a workshop but after months next to the engineers: printed parts lists, highlighters, copy-paste of old documents. In withholding tax the paper list is called Excel — just as invisible until you sit next to it.
Stakeholders at tier-1 banks took a TypeScript prototype with realistic data for a finished product — because it showed filters, validation, bulk actions and audit trail the way daily work needs them. Conversely, the business unit only believes the system once it finds its own case in it.
Teams keep Excel next to the system because the system does not show status, deadline or next step — not out of defiance. Ban the list and you get a hidden list. Take it seriously and you get the most precise requirements list there is.
Product UX in the withholding-tax environment and process work with experts in manufacturing. No names of banks, KVGs or depositary banks.
Withholding-tax reclaim software · vendor side
New MiKaDiv reporting module for reporting to the BZSt plus UX modernisation of the existing platform: personas, process mapping, information architecture and a TypeScript prototype that tier-1 stakeholders treated like production software. External consulting — not a partner, not part of the vendor’s team.
→Industry · SAP · manufacturing
Months next to industrial engineers in steam-turbine manufacturing: paper parts lists and copy-paste became a searchable interface with a template library and Excel-to-SAP integration. We apply the same method — observe, map, prototype — in the business units of banks.
→Document management platform
Work on a document management platform from the vendor’s perspective: filing, workflow approvals, four-eyes principle.
We orient our work on these standards, studies and authors — without claiming any certification.
ISO/TC 159/SC 4 · 2020 (2nd edition)
Seven interaction principles, from suitability for the task to robustness against use errors — our review grid for claim screens.
ISO/TC 159/SC 4 · 2019 (2nd edition, confirmed 2025)
Human-centred design as a process; conformance is self-declared, which is why we say “aligned with”, never “certified”.
Kate Kaplan, Nielsen Norman Group · 2020
Five layers of complexity — integrative, informational, intentional, environmental and institutional. Withholding-tax systems hit all five.
Nielsen Norman Group · 2025
Legacy, legend and learner users: experienced users fear losing productivity, not change — which is why we enable key users first.
Teresa Torres, Product Talk · 2021
Weekly conversations with users and the Opportunity Solution Tree — how vendor teams stay close to the business unit.
Jeffrey M. Hiatt, Prosci · 2006
Awareness, Desire, Knowledge, Ability, Reinforcement. According to the Prosci benchmark, visible sponsorship is the top success factor.
McKinsey & Company · 2021
Fewer than one third of respondents report full success — instead of the unsubstantiated “70 % fail”.
Federal Central Tax Office (BZSt) · 2026
Technical specification of the reporting procedure under Sections 45b, 45c EStG — the yardstick for every status model of a MiKaDiv report.
An initial call without a sales pitch: you tell us where the parallel list is still running. We tell you honestly whether we can help.
Book an initial call




