Customer data: what McDonald's knows and you don't

Can you say, right now, who was the last customer to stop buying from you, and why? Most business owners I know can't. Meanwhile, McDonald's knows when you'll order your next Big Mac. I'm not exaggerating. A Wired reporter asked the chain's loyalty program for a copy of his own data and got a 515-page dossier. That story says a lot about customer data. Mostly about the gap between companies that use it and companies that just collect it and forget it.

What was in the 515 pages

According to the report, the file had what you'd expect: orders, times, locations, amounts. But it also had things nobody expects to read about themselves. Behavior classifications. A churn risk score. And a prediction, in system language, basically saying this customer wasn't going anywhere soon.

Think about that for a second. A company looked at someone's fast food history and decided that person is loyal. Then it treats them accordingly. Fewer discounts, because they're not needed. More coupons for the people slipping away.

You could find that scary. I do, a little. But there's also a business lesson here, and it's far more useful to you.

If you understand your customers, you don't need to shout louder than your competitors.

Your business keeps a dossier too (nobody reads it)

Here's the part most people miss. You probably already have a dossier like this. It's just scattered.

It's in your sales system. In the CRM your sales team half fills out. In your WhatsApp history. In your email platform. In your ERP. In the spreadsheet Fernanda from finance updates every Friday.

Every SaaS you pay for collects customer data all day long. The problem is that each one keeps its own slice, and none of them talk to each other. McDonald's has 515 pages because it put everything in one place. You have 515 pages split across 8 tools, and none of them tell you anything.

Here's an example from a client, a small distributor:

• The order system knew a big buyer had cut volume by 40% over three months.
• Support knew he had complained twice about late deliveries.
• Sales knew neither.

The customer left. By the time the sales rep called, he had already signed with another supplier. All the information was there. What was missing was someone, or something, to connect the dots.

How to use customer data without becoming Big Brother

There's a fine line here. On one side, the company that knows you and serves you well. On the other, the company that knows you too well and gives you the creeps.

My honest opinion: most small and mid-sized businesses are nowhere near the second risk. It's not worth worrying about. Their real problem is the opposite. They know too little and react too late.

Still, a few rules help you stay on the right side:

• Use data to serve, not to exploit. Letting a customer know the product they always buy is running low is service. Raising the price because they "won't leave" is something else.
• Collect what you'll actually use. If you don't know what a field in your signup form is for, delete it.
• Actually comply with privacy law. In Brazil that's the LGPD. If someone asks for their data, you need to be able to hand it over. If that would take two weeks of digging through systems today, things are a mess.

That last point is interesting. The McDonald's customer only found all this out because he asked. If one of your customers asked today, would you know what to send?

Three signals already sitting in your systems

You don't need AI predictions to get started. You need three questions your data can already answer, if someone asks.

1. Who's buying less? Compare the last 90 days with the 90 before. Every customer who dropped more than 30% deserves a phone call. Not a promotion. A call.

2. Who missed their usual cycle? If a customer buys every 20 days and it's been 35, something changed. That's McDonald's "churn risk" in its simple form. One query in your sales system handles it.

3. Who complained and then went quiet? Silence after a complaint is rarely satisfaction. Usually it's someone shopping around.

None of these questions need a new tool. They need you to combine what you already have and look at it often. This is where automation makes a difference. A simple flow that cross-checks sales and support and sends a list to your sales team every Monday morning. It takes a few days to build and runs on its own after that.

And what do your SaaS vendors know about you?

Now flip it around. You're a customer too. And the tools you pay for are building a dossier on you as well.

They know how many users actually log in. Which features nobody uses. Whether your usage is dropping. Some know more about your business than your own partner does, because your orders, contacts and finances all pass through them.

Two practical takeaways:

• Read what you signed. Check where your data lives, whether it's used to train AI models, and how you export everything if you want to leave.
• Test the exit before you need it. Request a full export today. If you get a confusing, incomplete file, or it takes weeks, you're locked in. McDonald's predicted its customer would never leave. Some SaaS vendors don't even need to predict. They just make leaving hard.

That's the less talked about side of running everything on off-the-shelf tools. It's practical, it's cheap at first, and it's fine to use them. But the data is yours. You should be able to take it with you on a Saturday afternoon without asking anyone for a favor.

Where to start on Monday

If I were you, I'd do this, in this order:

1. List every tool that stores customer information. It's usually between 5 and 10.
2. Pick the two most important ones. Usually sales and support.
3. Ask the "who's buying less" question using only those two.
4. Call the top five on the list.

No big project, no pretty dashboard. Just a first return on something you already pay to store. If those five calls save one customer, the effort has paid for itself.

McDonald's spent millions to learn that a customer wasn't leaving. You can find out who is leaving with what's already in your drawer. Just don't let those 515 pages sit there unread.

If you want help pulling this scattered data together and building a routine that warns you before a customer disappears, tell me where your business stands today.

McDonald's knows when you'll order your next Big Mac. You don't know which customer stopped buying from you last week.

A reporter asked the chain for his own data and got 515 pages back. That included a churn risk score.

Your business has a dossier like that too. It's just scattered across 8 tools that don't talk to each other.

I watched a distributor lose a big buyer this way. The order system saw volume drop 40%. Support got two complaints. Sales never heard a thing.

You don't need AI to get started. You need one simple question: who bought less in the last 90 days? Call the top five on that list.

If you want a routine that warns you before a customer disappears, let's talk.

#CustomerData #Automation #Retention #SmallBusiness #Management