AI tools promise to tell a business owner what is going wrong and what to do next. They can only work with what the business has recorded. If hours are not logged, costs sit in three systems and nobody can say which clients are profitable, an AI will either stay silent or, worse, sound confident about a guess.
The Data hygiene category of the free PLYNT business scan asks three questions: could you state last month’s profit within ±10% right now, how many systems hold your money data, and can you see anywhere which clients are profitable. The scan also calculates an AI data readiness score from five inputs: time tracking, the number of systems holding money data, visible client profitability, confidence in profit and the number of tools. Time tracking and money-data systems count twice. Below 40 is Low, 40 to 69 Medium and 70 or more High.
Why these inputs matter
- Time tracking. In a service business most cost is people’s time. Without it, no analysis can explain where money went.
- One place for money data. When invoices, payments and expenses live in different systems, every answer starts with reconciling them, and each copy can disagree.
- Visible client profitability. If the business itself cannot see it, an AI cannot either.
- Confidence in profit. Knowing last month’s profit within ±10% means the basic figures are recorded and trusted.
- Fewer tools. Every additional tool is another place where the same fact can be recorded differently.
A checklist
- One record per client. The same client should not exist under three spellings in three tools.
- Time on tasks and clients. Hours linked to the work they were spent on, approved by a person.
- Costs allocated to work. Direct expenses such as freelancers and media assigned to the client or project.
- Collected, invoiced and estimated kept apart. A figure is only useful if you know which of the three it is.
- Consistent names for stages and statuses. “Done”, “complete” and “finished” should not mean three different things.
- Access rules decided in advance. Decide which roles may see financial and personal data before any AI reads it.
- A named owner for the data. Someone who notices when records stop being updated.
What AI should and should not do
Useful AI in operations explains a change, shows the records behind a finding and suggests a next step that a person reviews. It should not invent missing numbers, hide its sources or act on important decisions without approval. A good test is simple: for every conclusion, can you open the entries it is based on?
Where to start
Most businesses do not need a data project. They need two habits: logging time against the work, and keeping invoices, payments and expenses in one place linked to clients. Three months of both is enough for meaningful analysis.
How PLYNT supports this
PLYNT keeps tasks, approved time, invoices, payments and expenses in one workspace, linked to clients and projects, and keeps collected cash, unpaid invoices and estimates distinct. Its AI capabilities are in development and are designed to work within each user’s access, show the records behind each finding and leave decisions to people. Nothing is presented as available before it is. See Intelligence and reports and analytics.
To see your AI data readiness score, take the free business scan. The answers stay in your browser, and no external AI receives them.