I paid $500K for filthy books

Hi {{first_name | Reader}},

Over the last three years, poor bookkeeping has cost me more than half a million dollars.

I want to be precise about what that means. I didn't embezzle it. I didn't miscalculate a tax bill. I lost it in the slow, invisible way.

Hours spent cleaning client data before I could analyze it. Reports I couldn't ship because the numbers didn't tie. Client decisions I couldn't support with confidence because the underlying ledger was a swamp. Projects I walked away from because the cleanup alone would have taken longer than the engagement.

Half a million dollars of my time, my attention, my ability to deliver. All gone to fixing other people's books before I could do my actual job.

So when I say I loathe filthy bookkeeping, I mean it as a professional injury, not as an opinion.

Here is what took me years to admit out loud.

Confidence in client presentations is a data quality artifact, not a personality trait.

When the numbers are clean, I sound like a CFO. I make calls, I take positions, I give advice that moves the business. When the numbers are dirty, I hedge. I qualify everything. I use words like "approximately" and "subject to reconciliation." The client hears the hedging and treats my advice the way they would treat a weather forecast.

The same data dictates whether your KPIs mean anything. You can build the prettiest dashboard in the world. If the underlying entries are misbooked, the trend line is noise. You're looking at the wrong KPI at the wrong time, and you don't know it.

Clean data is the floor. Everything else stands on top of it.

Last week's send was about two spreadsheets that disagreed on the same number. This week's is about why they disagreed in the first place.

I built a diagnostic tool that drills through every line in a client's general ledger. It grades the books across five dimensions. It flags anomalies, one by one, with severity levels. And it tells me, before I write a single slide, whether the data is trustworthy. Or whether I need to clean it first.

I run it on every new client now. And every Monday morning on the ones I've already onboarded.

This week I ran it on one of my clients. 113 accounts. The ledger balanced to the penny, 50.7 million debits matching 50.7 million credits exactly.

Here is what the scanner showed me.

At the very top, a green banner: Ledger balanced · 50,724,796 dr = cr. Debits and credits reconcile to the penny. By every traditional test, the one most CFOs run, these books pass. They add up.

113 accounts total. 88 clean, 25 flagged. 86 raw anomalies, 11 of them at high severity.

If I close the scanner here, I write a normal quarterly report. I'd hedge on a few ratios and move on.

That's the trap. The balanced ledger test is 1950s bookkeeping. It tells you that debits equal credits. It tells you nothing about whether those debits and credits are in the right accounts. Or the right periods. Or the right descriptions. Or the right sizes.

A balanced ledger is the first test, not the last.

The scanner grades the books across five dimensions the balance test doesn't see:

The weighted average is 75. Two dimensions scored near 100. Two scored in the 80s and 90s. One dropped to 69 and dragged the grade down.

This is how every "Strong" score happens. A couple of mediocre dimensions hidden behind a couple of perfect ones. The overall number is mathematically correct and diagnostically useless.

Under the dimensions, the scanner lists what it found:

Out of 113 accounts, 112 had at least one journal line booked into a period that was supposed to be closed. Every account except one. That's not rounding. That's systemic.

The first three flags alone represent over 260 account-flag pairs on a ledger that balances perfectly.

Every trend line this client has built in the last decade runs through data with these flags. Nobody caught them because every traditional test passes. The balance test. The cash reconciliation. The bank tie-out. The errors are structural, not arithmetic.

260 problems hiding behind a score that looked fine.

Here is the part I can't show you for free.

You've seen the scan. You've seen the 75. You've seen the five dimensions and the five flags. You've seen that a perfectly balanced ledger can hide 260 structural problems that corrupt every number built on top of it.

What you haven't seen is which account broke. Or the three findings on that account. One of them changes how this client reports revenue forever, and I almost shipped a quarterly report without catching it.

Starting today, that's where the newsletter splits.

What you get for €19/month

Every Saturday, the full build behind the story.

For this week:

The story stays free. Forever. The principle stays free. Forever. What moves behind the wall is the how. The code, the rules, the templates, the exact steps you can copy into your own CFO practice.

I lost half a million dollars learning all of this the hard way.

Now the scanner runs every Monday at seven. It costs me nothing. It saves me from myself.

That is the whole point of automating yourself. You build the system that stops you from repeating the most expensive mistake of your career.

Reply and tell me: what would your books score? I'll tell you what the scanner usually flags first. For free.

Or take the Scalable Finance Wheel if you want a broader diagnostic on the finance function, not just the books.

See where your finance function stands →

Until Saturday,

Samer