The disconnect tax: what it costs to run a fragmented finance team, processes and stack

From $100M to $5B in revenue, finance teams spend 40–60% of their hours on work that exists only because the systems are disconnected. Here's what that costs, who pays for it, why it compounds, and what changes when the architecture is right.

Finance teams are not slow because the people are slow. They are slow because the architecture is wrong. I spent twenty-five years inside the Office of the CFO, across companies from roughly $100M to $5B in revenue, and I want to be precise about a cost that almost nobody puts a number on — because the number is uncomfortable and the people paying it have stopped noticing they pay it.

Call it the disconnect tax. It is the share of your finance team's time that exists only because your systems cannot agree with each other. Not work that creates value. Not judgment. Not analysis. Just the labor of moving data between tools that were each built to optimize their own square and were never built to be a stack.

How the stack became a tax

The ERP came first. NetSuite, Business Central, SAP — the general ledger was the job. Then procurement got bolted on with Coupa or Ariba. Then AP got bolted on with Bill.com or Tipalti. Then the close got bolted on with BlackLine or FloQast or Numeric. Each tool was excellent at its own square. Each arrived with its own data model and its own integration story.

The integrations worked enough to demo. They never worked enough to remove the human in the middle. So every Controller, CAO and CFO I know has someone — sometimes a whole team — whose entire job exists because the systems cannot talk. They reconcile AP to the GL. They re-key journal entries that already exist in the AP system. At month-end they explain variances that exist only because two systems rounded differently. None of it is anyone's fault. It is the architecture.

Adding another point solution makes the disconnect worse, not better. Every new tool is one more square to reconcile against all the others.

The ERP was never built to do this

Here is the part the market keeps getting wrong. The ERP is a system of record — it was designed to hold data: the general ledger, consolidations, intercompany, eliminations, revaluation. It is very good at being the place the numbers live. It was never designed to be a system of work.

But the ERP has spent two decades trying to grow into workflows it was never architected for — procurement, AP, close orchestration — because it holds the data and the data is where the gravity is. The result is an ERP straining at tasks that aren't a natural fit for it, and a finance org that still has to buy a separate tool to do the thing the ERP can't. It holds enormous amounts of data, and yet you still need Workiva to do your financial reporting on top of it. That gap — the system that holds the data is not the system that does the work — is the disconnect tax in its purest form.

The number

At a mid-market company, the typical finance team spends 40 to 60 percent of its hours on work that exists only because the systems are disconnected. That is not cynicism. That is me looking at my own teams over twenty-five years and adding it up. At a large finance org, the reconciliation and alt-tab burden is the equivalent of many full-time roles — roles that would simply not need to exist if the data were connected.

The number is the headline. The grinding indignity of the work is the worse story. Smart, ambitious accountants did not sign up to push data between systems. They signed up to do the work that requires a brain. The disconnect tax takes that work away from them and replaces it with the work a machine should never have made necessary in the first place.

Why it compounds — and why you fix it early

Here is what makes the disconnect tax different from an ordinary inefficiency: it does not scale linearly. It compounds. A $100M company with one entity, one currency, and a dozen vendors pays a small version of this tax — small enough to ignore, which is exactly the trap. Because the architecture you tolerate at $100M is the architecture you inherit at $1B.

Every new dimension multiplies the seams. Add headcount and more hands touch each record. Add revenue and transaction volume climbs. Add a second entity, a third currency, a fourth location, an acquisition — and the reconciliations don't add, they multiply, because every system now has to agree with every other system across every dimension. The team that spent 20% of its time reconciling at $100M is spending 50% at $1B, and the CFO who could have changed the shape for almost nothing early is now facing a re-platforming project at the worst possible time.

This is the real argument for fixing it early. The disconnect tax is cheapest to eliminate when it is smallest — before the headcount, the entities, and the volume have hardened the wrong architecture into place. A company that connects its data layer at $100M never pays the compounding tax on the way to $1B. A company that waits pays it the whole way up, and then pays again to unwind it. The best time to remove the tax is before you can feel it. The second best time is now.

Who actually pays

The Controller/CAO pays first. The reconciliation lands in their lap because they own the number that has to be right. The CFO pays in headcount. The only lever that ever worked was to staff up — another senior accountant, another AP specialist — and the work expanded to fill the hands. The audit committee pays in fees. When the audit trail lives in eight systems, reconciling the trail is the auditor's actual job, and you are billed for it.

And everyone pays in time. The close that drags on for days, if not weeks. The variance meeting that exists to explain rounding. The close meeting that exists to bring it all together. The integration project that is never quite done.

Why it persisted — and why that's ending

The CFO has known about this tax for years. They accepted it because the alternative did not exist. You could not buy your way out; buying more tools has made it worse. So the tax was treated as a fixed cost of running finance at scale.

Two things changed. First, the agentic AI generation actually works — models can now read a contract, draft a journal entry, run a reconciliation, and explain themselves. We were promised this in 2017 and got nothing; promised it again in 2022 and got something half-there. The threshold finally moved. Second, and more important: finance teams hit the wall. Adding tools stopped helping, headcount could not keep pace with revenue, and the tax started showing up in audit findings and turnover. The Controllers I talk to have stopped believing the next tool will fix it. That exhaustion is the precondition for an architectural answer.

Prevention beats reconciliation when the architecture allows it. For the first time, it does.

What “connected” actually means

The answer is not a better integration. It is the absence of the need for one. One connected data layer: procurement, AP, and the close on the same record. A purchase request becomes a PO — the same record at a different stage. An invoice arrives and matches that PO and the underlying contract, in real time, against the actual contract terms. The match generates a journal entry — the same record at yet another stage. Payment fires; the JE clears. Reconciliation is unnecessary because nothing ever got out of sync.

That is the whole idea, and it is why the tax disappears rather than shrinks. You do not reconcile a single record against itself. The work that consumed half your team's day was never real work. It was the cost of a wrong shape. Change the shape and the cost is gone — not optimized, not automated around, gone.

And to be clear about the posture: none of this runs unattended. AI proposes; your team confirms. Every action arrives as a proposal a human approves before anything posts, pays, or closes. The point isn't to remove the human from the close — it's to remove the tax, so the human spends the close deciding, not reconciling.

The tax was never a people problem. It was an architecture problem wearing a people costume. Take off the costume and you find what your finance team was actually hired to do, waiting underneath.