The Full Picture: How to Use AI to Make Sense of Your Business Spreadsheet

Most glazing businesses have at least one spreadsheet that has grown over time. It started as a leads tracker, or a job log, or a way to keep tabs on margins. New tabs appeared. Columns were added. And now it sits there, full of information that nobody quite has time to dig into properly. Right now, we need to know what’s happening in our businesses more than ever.

That spreadsheet you have probably has more to say than you realise. The challenge is that reading across multiple tabs, spotting connections between enquiry source and conversion rate, between product type and complaint volume, takes time that most business owners never find.

AI can do that work for you. In this edition, we walk through exactly how to upload your spreadsheet, which tools to use, what they cost, and the specific questions that surface the most useful insights.

What tools to use, and what they cost

Three tools are worth knowing about. All three allow you to upload a spreadsheet and ask questions about the data inside it.

• Microsoft Copilot in Excel – If your business already uses Microsoft 365, look for the Copilot button in the Excel toolbar. It has been rolling out across business plans. If it isn’t yet visible, check with your IT contact or Microsoft account. It works directly inside the spreadsheet, so there is no upload step.
• Claude.ai – Available free at claude.ai. The free plan allows basic file uploads. Claude Pro, at around £18 per month, gives you more reliable file handling and larger spreadsheet support. It can even then run inside Excel. For a one-off exploration it is a low-cost entry point. For regular use, the Pro plan is worth it.
• ChatGPT – Available free at chatgpt.com. File upload and data analysis is available on ChatGPT Plus at around £20 per month. The free version has more limited access to these features.

If you are already in Microsoft 365, start with Copilot. If not, Claude Pro is a simple and affordable starting point.

Prepare your spreadsheet first

You don’t need a perfect dataset. But three quick checks will improve the quality of the output significantly.

• Check your column headers. Every column should have a clear, consistent label in row 1. If columns are unnamed or use abbreviations only you understand, rename them. AI reads the headers as context for everything it analyses.
• Remove merged cells where you can. Merged cells cause confusion in AI analysis and can lead to misread data.
• Check for consistency within columns. Dates should look like dates. Product names should be consistent across rows. If a product appears as both "Bi-fold" and "Bifold", the AI may treat these as two separate categories.

Upload and confirm the AI has read it correctly

Once your file is ready, open your chosen tool and upload the spreadsheet. In Claude.ai or ChatGPT, use the paperclip or upload icon to attach the file. In Copilot, the file is already open in Excel.

Start with this question:

Tell me what data is in this spreadsheet. How many tabs are there and what does each one contain? Summarise the column headers on each tab.

This does two things. It confirms the AI has successfully read the file. And it gives you a clean summary of what you are working with, which is often useful in itself.

The questions that surface the most useful insights

Ask these one or two at a time. Review the response before moving on. If something surprises you, dig into it with a follow-up before jumping to the next question.

What patterns do you see in my enquiry data? Are certain months, lead sources, or product types consistently stronger or weaker?

Is there a connection between enquiry source and conversion rate? Which sources generate the highest volume but the lowest conversion?

Which product types or job types have the strongest average order values? Are there any that appear frequently in enquiries but rarely in completed jobs?

Looking across all tabs, what is the most significant pattern in this data that I might not be paying attention to?

Show me the top three things this data suggests I should look at or change in how we operate.

If your spreadsheet includes a complaints or issues tab, add this:

Looking at my complaints or issues data, are there any patterns? Particular product types, time periods, or job types that appear more frequently?

The AI doesn’t tell you what to do. It shows you what the numbers say. That is the point.

What the output typically looks like

AI reads across the data and connects things that would take you hours to spot manually. In a typical glazing business spreadsheet, you might see:

• A breakdown of which enquiry sources convert best, with average order values compared across each.
• A seasonal trend showing when certain product categories peak and when they dip.
• A product type that is converting well but hasn’t featured in recent follow-up or marketing activity.
• A connection between a particular enquiry source and lower-than-average job values or longer time to close.

An honest word about limitations

AI analysis is only as reliable as the data you give it. A clean, well-labelled file gives you clear patterns. A spreadsheet with missing values, merged cells, or inconsistent formatting gives you noise.

AI tools can also misread a column header or draw an observation from the wrong part of the data. If an insight surprises you, ask it to show you the specific rows it is drawing from. Treat everything as a starting point for a conversation, not a definitive report.

The output is a prompt. A prompt for a conversation you have with yourself about your business. Act on what resonates, verify what surprises you, and ignore what doesn’t fit.

What to do next

• Export a spreadsheet from your quoting system, CRM, or job management software. Anything with six months or more of data across multiple areas of the business is ideal.
• Run through the three preparation checks: column headers, merged cells, consistency.
• Open Copilot in Excel, Claude.ai, or ChatGPT and upload the file.
• Ask the opening question to confirm the AI has read the data correctly.
• Work through two or three of the insight questions above and note anything that surprises you.
• Dig into one finding with follow-up questions until you understand what the data is showing.

This is worth repeating quarterly. Patterns shift. The question that gives you one answer in January may tell a different story by June.

For more help and support getting started, visit https://thinkivity.co.uk/glazepro-ai/.