Rant: 'Human Excel' Is Not a Database, Stop Making It One

Stop Painting Your Data

We need to have a very direct conversation about how you treat your spreadsheets.

I was recently asked to automate an invoicing process. The operations team proudly shared their “Master Tracker.” It was a disaster. They had used thick borders to draw visual boxes around client groupings. They had merged cells to create hierarchical titles. They used a specific shade of blue background to indicate that a client had paid via wire transfer, and a slightly lighter blue to indicate a credit card payment.

The operational cost of this formatting is absolute gridlock. When you build a spreadsheet for human eyes rather than machine readability, you sever the data from the rest of your tech stack. Automation tools, CRMs, and reporting engines do not have eyes. They cannot see that the cell is blue. They cannot interpret a blank row as a “section break.” When the automation script hits a merged cell, it crashes.

The Chaos: Spreadsheet abuse is born from a desire for comfort. We want our data to look like a document. But a spreadsheet is not a Word document. When you treat it like one, you create The Chaos. Data becomes trapped. If you want to know how many invoices are unpaid, you can’t run a simple pivot table or write a COUNTIF formula because the status isn’t written down anywhere—it’s just implied by a lack of green highlighting.

Telepathy is not a management strategy. Expecting your colleagues (or your software) to magically understand your personal color-coding system is arrogant. This is a hygiene problem. We do not paint; we document. If your data cannot be exported to a plain text CSV file and still retain 100% of its meaning, your data is fundamentally broken.

The System: We automate the boring work so we can do the real work. But you cannot automate chaos. Before we build a single Flow or Zap, we must enforce strict, uncompromising rules for machine-friendly tables. We transition from Human Excel to Machine Excel.

Here is the checklist. If you violate these rules, your spreadsheet will be rejected by the automation team.

The Unbreakable Rules of the Database

1. Rule 1: No Merged Cells, Ever.
   • The Fix: If a value applies to five rows, copy and paste that value into all five rows. A machine reads row by row. If row 3 is blank because it was “merged” with row 2, the machine assumes row 3 has no data.
2. Rule 2: Color is Not Data.
   • The Fix: If a cell is red to mean “Late,” you must create a column named Status and type the word Late. You may use conditional formatting to turn the cell red only after the word is typed. The text is the truth; the color is just an echo.
3. Rule 3: One Data Type Per Column.
   • The Fix: If Column E is for “Invoice Amount,” it must only contain numbers. Do not type $5,000 (pending approval). The moment you add letters to a number column, you break the math. Move the note to a separate Comments column.
4. Rule 4: Zero Blank Rows or Columns.
   • The Fix: Blank rows tell automation scripts that the dataset has ended. If you need “breathing room,” adjust your row height. Never leave an empty row in the middle of your dataset.

The same example as a layout and as records. “Yellow” is shown as an annotation so the comparison remains readable without colour.

BEFORE: MEANING HIDDEN IN LAYOUT
+--------------------------------------------------+
| Client A                  [merged across columns]|
+--------------------------------------------------+
| Invoice 101 | 5000 | [yellow; no status text]    |
|                                                  |
| Invoice 102 |  750 | [yellow; no status text]    |
+--------------------------------------------------+

AFTER: EXPLICIT FIELDS
Client   | Invoice | Amount_USD | Status
---------+---------+------------+--------
Client A | 101     | 5000       | Pending
Client A | 102     | 750        | Pending

One record per row. No blank spacer row.
Colour may echo Status; it does not replace it.

The Reward of Structure

When you strip away the aesthetic nonsense and build a rigid table, magic happens. You can instantly connect that sheet to Zapier. You can push the data into PowerBI. You can write a script that emails every client with the status “Late” in three seconds.

Process is not bureaucracy. Process is freedom. When you enforce these boundaries, your data can finally move without begging. Stop trying to make your database look pretty. Make it powerful. Clear is kind.

FAQs

Why are merged cells so bad? They make the headers look organized.

They make the headers look organized to human eyes. To a machine, a merged cell destroys the rigid grid structure required to index data. It breaks sorting, filtering, and script loops.

If I can't use colors to show status, how do I know what's urgent?

You add a column named 'Status' and you type the word 'Urgent' into it. Data must be readable as text so the computer can act on it.

Does this mean our spreadsheets have to look ugly?

They have to look functional. You can still use Conditional Formatting to change colors based on text values, but the text value must be the source of truth, not the color.