Data hygiene for a small shop
Skip the enterprise advice
Read anything about preparing your data for AI and you will find enterprise ceremony: warehouses, lakes, governance committees. A ten-person shop needs none of it. What it needs fits on an index card, and the machine-readable payoff is nearly the same.
The goal is simple: when a person or a machine goes looking for a fact, there is one obvious place it lives, and it is actually there.
The index card
Three rules cover most of it:
- A naming convention: files carry the job number and date in the name, the same pattern every time, so JOB1482-quote-2026-04-10 finds itself
- One folder rule: every job gets a folder, everything about the job goes in it, no orphan files on desktops
- One source of truth per fact: the customer's address lives in one system, and every other place that needs it refers there instead of keeping its own copy
Why machines care
Automations can handle mess, reading scans and photos is routine now, but they inherit your ambiguity. When two spreadsheets disagree about a price, the machine faces the same question a new employee would: which one is true. It will flag it and wait, which is correct behavior, but every flag is time.
Clean-enough data is not about tidiness. It is about reducing the number of judgment calls per day to the ones that deserve judgment.
An afternoon, not a project
Adopting the three rules going forward takes a team meeting and a memo. Do not backfill history, it is rarely worth it; automations mostly work on current flow anyway. Start the convention today, and in three months the recent past, the part that matters, is clean. It also makes every future build faster, including the first one you pick from your audit.
The audit answers this for your business
Two weeks, $2,500 flat ($1,000 for the first three clients), and you get the map of your own automatable work with dollars on it.