Last quarter I ran a consulting session with the CFO of a mid-sized logistics company in Muscat. She had three years of shipment data in an Excel file and a vague feeling that they were leaving money on the table. We spent 45 minutes looking at the data together. By the end of the session, I'd found two pricing anomalies and one route optimisation opportunity worth a combined $200,000 annually.
She didn't need a data scientist. She didn't need an AI model. She needed someone to ask the right questions about data she already owned.
Here's how to run that same process yourself — before you spend anything on AI tools or consultants.
Step 1: Map your data assets (20 minutes)
List every source of structured data your business produces. This includes: invoices, sales records, CRM data, support tickets, inventory logs, delivery records, employee time sheets. For each source, note: How many records? How far back does it go? Who owns it? Is it in Excel, a database, or a SaaS tool?
Most businesses have at least 5–7 data sources they've never analysed. The question is not whether the data exists — it almost certainly does — but whether you've looked at it.
Step 2: Ask three questions about each source
For every data source you list, ask: What does the average look like? What are the outliers? What changed over time?
You don't need a statistical model. You need a pivot table and 10 minutes of attention. Outliers are where hidden value lives: customers who buy at unusually high margins, products with unusually high return rates, routes that cost twice what they should, employees who close deals at twice the team average.
Step 3: Look for asymmetries
An asymmetry is when two similar things produce very different outcomes. Similar customers paying very different prices. Similar products with very different profit margins. Similar markets with very different growth rates.
Asymmetries are almost always opportunities — either to expand what's working or stop what's not. Most businesses have at least one large asymmetry hiding in their data that they've never noticed because they look at averages, not distributions.
Step 4: Find one thing you could change this month
The audit is only valuable if it produces action. Once you've found an outlier or an asymmetry, ask: What would it take to change this? If you have customers paying 20% less than average for the same service, can you raise their prices? If you have one salesperson performing at 3× the team average, can you study their process and teach it?
You don't need AI for this. But AI makes it 10× faster once you know what to look for.
What AI adds
Once you've completed this manual audit, you have something much more valuable than a vague AI project: a specific question. 'We think customers in segment X are underpriced — can an AI model tell us the optimal price point?' That's a question a model can answer in hours, not months.
The businesses that get the most value from AI are not the ones who bought the technology first. They're the ones who understood their own data first, then used AI to answer specific questions faster.
If you want help running this audit on your actual data, that's exactly what the 45-minute Deep Dive session is designed for. I'll bring the questions; you bring the spreadsheets.