Modernizing the finance ERP and data core: a CFO's execution playbook

August 23, 2026 · 5 min · Free, with full transcript

Most finance ERP modernization projects stall not because the technology fails, but because the sequencing is wrong and the data governance is an afterthought. This playbook gives CFOs a concrete, ordered set of moves to replace legacy infrastructure without breaking the close cycle or burning two years of goodwill.

Listen to the podcast

4 min

Read the full transcript

Welcome to the Leaders Insights Podcast.

Today's episode, Modernizing the Finance ERP and Data Core, a CFO's execution playbook.

Every CFO I know has a war story about an ERP replacement, the core financial software

that runs the books, that went sideways.

Why do these things keep failing?

Because people treat it like a technology purchase when it's really a sequencing problem.

The software works fine.

What breaks is the order you do things in and the fact that nobody owns the data until

it's already a mess.

That sounds like a convenient excuse.

The tool was great.

We just used it wrong.

No, it's the opposite of an excuse.

It's the part you actually control.

Take Revlon.

They filed for bankruptcy in 2022, and the SAP implementation the year before was so

botched they couldn't fulfill orders.

The system wasn't cursed.

They flipped the switch on everything at once and hoped.

So what's the durable principle here?

Strip away 2026 and the AI hype.

What have the good operators already internalized?

That you never modernize the close cycle and the data foundation in the same breath.

The close, the monthly process of shutting the books and reporting numbers, is sacred.

If you break that, the board stops trusting you in about six weeks.

So you freeze the close, migrate the data underneath it, and only then touch the process.

Give me the actual order then.

Pretend I'm a CFO about to sign a two-year contract.

First move.

Build the data model before you buy anything.

Decide what a customer is, what a cost center is, in one canonical definition.

Most companies have four versions of revenue living in different spreadsheets.

Reconcile those first.

That's the boring part everyone skips.

Everyone skips it because it's unglamorous and it surfaces political fights.

Sales counts revenue one way, finance another, and now somebody has to lose.

But if you migrate dirty data into a shiny new system, you've just bought an expensive

new place to store your confusion.

Second move.

Run the old and new systems in parallel for at least two close cycles.

Yes it's double work.

Yes it's painful.

But that's your insurance.

When Hershey rebuilt their systems years back, the disciplined ones always kept the old ledger

running until the new numbers matched to the penny.

How long does match to the penny actually take?

Give me a real figure.

For a mid-sized company, budget three to four months of parallel running.

Not the two weeks the vendor slide deck promises.

Vendor timelines, and I'll flag that's their number from their sales team.

Assume your data is clean.

It never is.

Where does AI fit?

Because in 2026 every finance conference is drowning in it.

Narrowly and late.

AI is genuinely useful for one thing here.

Reconciliation, matching transactions across systems to find where the numbers diverge.

A model can flag the 400 line items that don't tie out overnight.

Work a human would take a week to do.

But you point it at clean defined data.

Point AI at garbage and it gives you confident garbage faster.

So the machine doesn't save a bad sequence.

It amplifies whatever discipline you already have.

If your data governance, the rules for who owns which number, and how it's defined, is

an afterthought, AI just makes the afterthought louder.

Let me push you.

Isn't GoSlow running parallel to find your data just a way to burn 18 months and a lot

of budget?

It's cheaper than the alternative.

Gartner has put large ERP overrun rates north of 50% on budget.

And that's their research.

Revlon's rushed version ended in Chapter 11.

Slow is expensive, wrong is fatal, pick your poison.

If a CFO listening does one thing tomorrow, what is it?

Before you talk to a single vendor, get your finance team in a room and write down the

one definition of revenue everyone will use.

On paper.

If you can't agree in that room, you are nowhere near ready to buy software,

in jai, and you just saved yourself two years and eight figures.

One sentence, one definition.

That's the whole playbook.

That's the part nobody wants to hear, which is exactly why it's the part that works.

That's it.

The CFO Decision Tools are live at MBA-training.com.

Read and go deeper