Clean Data You Can Actually Trust: A Monthly Checklist
You finally have one home for your numbers.
So you open it expecting to trust whatever it says.
Then you look closely, and the same client shows up twice.
This is step four of setting up business data. A single home does not stay clean by itself. Here is the light, recurring check that keeps it that way.
A home is not the same as clean
Having a home does not mean the data is right.
Three weeks ago you chose your shortlist. Two weeks ago you fixed how each number gets captured. Last week you gave every number one home with one owner, one home per number. That solves where the data lives.
It does not solve what lives there. Duplicates creep in, fields get left blank, someone types a date as text, and two colleagues quietly disagree about what "closed" even means.
None of this looks dramatic from a distance. A dataset can look complete and organised while still being wrong in ways that only surface the moment you try to use it. That is the gap this step closes.
The four part data quality checklist
You do not need a data quality program.
You need a short data quality checklist instead, four checks, run lightly and often.
The full version of this idea, used across data teams, names six dimensions of data quality: accuracy, completeness, consistency, timeliness, validity, and uniqueness. For a founder checking a shared sheet once a month, four of those six do nearly all the work.
The same client, lead, or invoice entered more than once, usually because two people added it separately or an import ran twice. Most spreadsheet and CRM tools have a built in duplicate finder. Run it monthly, not once.
A required field left blank: an invoice with no due date, a lead with no source. Check the two or three fields from your Part 1 shortlist that actually feed a decision, not every field in the sheet.
A date typed as text, a phone number with the wrong country code, a price entered without knowing whether it includes VAT. Use a dropdown instead of free text where you can, and scan for anything that looks out of place.
The most common SME failure. Two people using "revenue" to mean two different things, one including VAT and one not, one marking a deal "closed" at signature and another at first payment. Write the definition down once, in one sentence, next to the field.
Timeliness and accuracy still matter, but they are mostly covered already if the other four hold and the data is captured at the source, which is what Part 2 was for.
Why the 15 minute version beats the big cleanup
Catching an error late does not just cost more. It costs a lot more.
A well known guideline in data quality management, the one to ten to one hundred rule, puts a number on this. Verifying a piece of data at the point of entry costs roughly one unit. Correcting it later, once it is already sitting in a system, costs roughly ten, because someone has to find it, check it, and fix it by hand. Leaving it unaddressed and letting it cause a wrong decision, a wrong invoice, or a failed report costs roughly one hundred.
The same mistake gets more expensive the longer it goes unnoticed.
Catching it at entry is cheapest. Catching it once a month is still cheap. Catching it when the bank balance does not make sense is not.
The exact multiples are illustrative, not a formula you can apply to your own business. The direction is not. This is also why this step cannot replace Part 2. Capturing data clean at the source stops most errors before they start. This step catches what still gets through.
What this looked like for the agency
The Eindhoven agency’s pipeline sheet is clean now.
Seven people, around EUR 40,000 a month in revenue. One shared pipeline, owned by the account manager, the old private copies deleted since Part 3. In week four, she finally trusts it enough to look closely.
The account manager runs the first monthly check. Three leads appear twice, added separately by her and a new hire during a busy week. Two closed deals have no close date, so last month's win rate cannot be calculated. And "closed" turns out to mean two different things: she marks a deal closed at signature, the founder had been marking his own deals closed once the first invoice landed, a habit from before the single home existed.
None of it is dramatic. All of it would have quietly wrecked the win rate number the first time anyone tried to report it. She deduplicates the three leads, pulls the two missing close dates from the email thread, and they agree in one sentence what "closed" means from now on. 15 minutes, and the number can finally be trusted the way Part 3 assumed it already could be.
Your 15 minute monthly habit, and what comes next
This only works if it stays small enough to actually happen.
Pick one fixed day a month, the first Monday or the last Friday, whatever you will actually keep. Then run the same short routine every time.
Run the duplicate finder on your shortlist data. Spot check the fields that feed a decision.
Do they still match what everyone actually means, now that a new person may have joined?
Small corrections, made monthly, so the next dashboard update (Part 5) starts from numbers you trust.
That is the whole habit. 15 minutes, once a month, on the numbers that matter, not an audit of everything. Next week is step five: turning this clean, single home of data into the dashboard that actually tells you something.
Not sure your numbers would survive a close look?
If you suspect your data has quietly drifted since it went into its single home, book a free clarity call. No pitch, no obligation, just a clear look at where the gaps most likely are and how to check for them.
Book your free clarity callDo-Creates works with SME founders on financial clarity and business strategy. This post does not constitute financial or legal advice.