← Back

2026-07-20

The Intermediary Commission Problem: Why BES Agent and Distribution Channel Data Is the Most Underreconciled Layer in Pension Reporting

Everyone in pension operations talks about fund NAV accuracy, contribution reconciliation, and regulatory reporting to EGM. Almost nobody talks about the commission layer. That is strange, because in every BES portfolio I have worked on, agent and distribution channel data is the single most inconsistent slice of the warehouse — and the one that quietly poisons downstream reports for years before anyone notices.

The reason is simple. Commission is treated as a finance output. It is actually a lineage problem.

The Assumption That Breaks Everything

Most BES pipelines are built on an implicit assumption: a policy has an agent, and that agent stays with the policy. This is wrong on day one and gets worse every quarter.

A participant's assigned agent changes for reasons that have nothing to do with the participant:

Each of these events is legitimate. Each one silently rewrites the attribution logic for future commission accruals — and sometimes retroactively for trailing commissions. If your pipeline stores agent_id as a column on the policy table with no effective-dated history, you have already lost.

What Actually Needs To Be Tracked

A correct commission lineage model treats the agent-policy relationship as a slowly changing dimension with typed events. Minimum viable structure:

That last point is where most implementations collapse. In Turkish BES, initial commission (giriş) and trailing commission (aidat/yönetim over the accumulation phase) frequently follow different rules. The originating agent may keep the trailing commission for X years even after being replaced as the servicing agent — unless the participant executes a fund switch that triggers a re-attribution clause in the intermediary agreement. Now you have three parties with a claim on the same policy's future commission stream, and your data model has one column.

Where Reconciliation Fails

The failure pattern is predictable. I have seen it in three separate BES portfolios:

  1. Month-end commission accrual matches finance. Everyone is happy.
  2. Quarter-end intermediary statements start diverging by small amounts. Ignored as rounding.
  3. An agent disputes their statement. Ops runs a manual query and finds forty policies where the assigned agent no longer matches the historical commission recipient.
  4. Someone tries to build a retroactive fix. They discover there is no reliable way to know who the agent was on any given accrual date, only who the agent is now.
  5. A reconciliation project is opened. It never closes, because the source system overwrote history on every update.

The root cause is always the same: the operational system was designed to answer "who is the agent today" and the warehouse inherited that limitation without questioning it. Nobody built the pipeline to answer "who was the agent on 14 March 2022 when this contribution was booked, under which channel agreement, at what commission rate, and who was entitled to the trailing portion."

Why This Layer Stays Broken

Commission data does not get the reconciliation attention it deserves because of how responsibility is split:

So the layer sits in the gap between four teams, none of whom own the lineage question. It only surfaces when an intermediary lawyers up over a disputed statement, or when EGM asks a question that requires historical channel attribution and the answer takes six weeks to assemble.

What To Build Instead

If you are designing or refactoring a BES data platform, three concrete moves close most of this gap:

None of this is exotic. It is standard temporal modeling. The reason it is rare in BES pipelines is that the original systems were built when the portfolios were small, agent turnover was low, and channel structures were stable. None of those conditions hold anymore.

The Cost Of Ignoring It

Unreconciled commission attribution is not a cosmetic problem. It shows up as:

Every one of these has a real lira cost. The commission layer is not glamorous, and rebuilding it will not get anyone promoted. But it is the layer most likely to produce the phone call nobody wants to receive — the one where an intermediary's lawyer explains, in detail, what your data cannot prove.