One thing that really frustrates me about AI consultants - and software engineers - is when they say things like:
- "You need to know your data"
- "You need to have clean data"
- "You need to know where your data is"
- "You need to do a data audit"
But they never explain what data actually is.
So many businesses I talk to about AI don't have a clue what "data" means. And that's not their fault. "Data" is possibly the most bland, abstract, generic term we could use to describe something that's actually critically important.
Here's the thing: data isn't as complicated as all these geeks want you to think it is.
Data is just information of some kind. It's just words and numbers.
The data we usually care about is the information you need to do your job.
It can be numbers output by a machine, analytics, financial figures, and spreadsheets. But it's also emails, customer support tickets, meeting notes, site logs, and any quantifiable or qualitative information you work with.
It isn't this big, scary, abstract concept.
If you were asked to do your job tomorrow the same way you did it today:
- What information do you need?
- What information do you receive?
- Which decisions do you make?
That's your data.
It means the emails you get, the spreadsheets you open, the CRM you check, the website analytics you review, the documents you edit. Both the quantitative numbers and all the qualitative text. It's all data.
So when people say "Let's do a data audit," it sounds complicated, but it really isn't. You're just asking: What information does my team use to do their work, and where does it live?
Why This Matters for AI
AI thrives on data - on information. The more relevant, organised information you provide, the better it generates useful outputs.
One of AI's superpowers is turning unstructured data - like conversation transcripts, meeting recordings, or scattered emails - into structured data, like reports, summaries, spreadsheets, and SOPs. It can do so much of that heavy lifting so your team can spend time on work that actually requires human judgment.
But here's the catch: AI amplifies whatever you feed it. Give it clean, current, relevant information and it produces useful outputs. Give it messy, outdated, contradictory information and it produces confident nonsense.
That's why this section exists. Once you realise data is just the information you use every day, you can see all kinds of ways AI can make your workflow easier and cut down on the repetitive grunt work everyone hates. But you need to prepare that information properly first.
Your business data typically falls into these categories:
- Customer data: vorrespondence, service history, project records, feedback
- Operational data: meeting notes, site logs, process documentation, handover reports
- Financial data: invoices, contracts, supplier agreements, pricing records
- Technical data: specifications, standards, drawings, compliance documentation
Each of these represents potential fuel for AI - or potential poison if it's messy, outdated, or inconsistent. The rest of this section shows you how to prepare it properly.