By Jeffrey Shear, J Shear Consulting

It's Monday's leadership meeting. Sales reports revenue at one number. Finance reports another. Operations has a third, pulled from a spreadsheet someone updated last Thursday. The next twenty minutes are spent arguing about whose number is right instead of deciding what to do about it.

If that sounds familiar, you're not alone — and you're not short on tools. Most mid-sized companies I work with already own more reporting software than they use. The problem is the data underneath it. In my last article, Build, Buy, or Wait?, most of the AI ideas that landed on the "wait" list were there for the same reason: the data wasn't ready. Gartner predicts that through 2026, organizations will abandon 60% of AI projects that aren't supported by AI-ready data.

The good news is that fixing it is more about discipline than technology, and it costs far less than another platform.

Five ways your dashboards are lying to you

• Same word, different meanings. Ask three departments to define "customer," "revenue," or "active account," and you'll often get three answers. Every dashboard built on those definitions is quietly inconsistent.

• Stale data dressed up as live. A polished dashboard fed by a manual export from last week looks current. It isn't.

• Duplicate and incomplete records. The same customer entered three ways, missing fields, orphaned transactions. Totals inflate or deflate, and nobody can explain why.

• Measuring what's easy, not what matters. Many reports exist because the data was available, not because anyone makes a decision with them.

• Logic hidden in one person's spreadsheet. A critical number depends on a formula only one analyst understands. When that person is on vacation, so is your reporting.

Why buying more BI won't fix it

When reporting feels broken, the instinct is to shop for a better tool. But a new BI platform connected to the same inconsistent data just displays the same wrong numbers more attractively. The same is true for AI: a model trained on contradictory data produces confident, contradictory answers.

Tools amplify whatever foundation they sit on. Fix the foundation first.

Data before dashboards: a five-step plan

1. Pick the numbers that actually run the business

Not a hundred metrics. Ten to fifteen. The ones your leadership team uses to make real decisions about pricing, hiring, inventory, cash, and customers. Everything else is secondary until these are right.

2. Write down what each one means

Create a one-page metric dictionary: the definition, the source system, and a named owner for each number. This single document resolves more Monday-morning arguments than any software purchase.

3. Name one source of truth for each

Every metric should come from one system, not be reconciled across three. When systems disagree, the dictionary decides which one wins, and the others get corrected or integrated.

4. Clean what matters, at the source

Don't launch a company-wide data cleansing project. Clean the data behind your core metrics, then stop the problem from recurring: required fields, validation at entry, and clear rules for who can create or change key records. Fixing data downstream in spreadsheets just guarantees you'll fix it again next month.

5. Then automate and visualize

Now the tools earn their keep. Connect your dashboards directly to the sources of truth, schedule automatic refreshes, and retire the manual exports. You'll likely find the BI tool you already own is more than enough.

A composite example

Consider a 200-person regional manufacturer — a composite of situations I've seen. They had three reporting tools, more than 140 reports, and a leadership team that no longer trusted any of them. Month-end reporting took eight business days, mostly spent reconciling.

Over about ten weeks, they agreed on twelve core metrics, wrote a metric dictionary, assigned owners, and fixed the customer and product data behind those numbers at the source. Roughly a hundred reports were retired because nobody missed them. Month-end reporting dropped to three days, the Monday meeting became a decision meeting, and a demand forecasting idea that had sat on their AI "wait" list became viable within six months.

Is your data ready for AI? A quick check

• Your leadership team agrees on the definitions of your core metrics.
• Each core metric has a single source of truth and a named owner.
• Key records (customers, products, vendors) aren't duplicated across systems.
• Reports refresh automatically rather than from manual exports.
• You could explain to an auditor where any number on a dashboard came from.

If you can check four or five of these, you're in a strong position. If you can check one or two, that's not a failure. It's your roadmap.

Where to start this week

At your next leadership meeting, ask everyone to write down, independently, how they define "customer" and "revenue." Compare the answers. If they differ, you've found your first project — and you've found it for free.

If your team spends more time debating numbers than acting on them, I help companies build a reliable data foundation and get real value from the reporting and AI tools they already own.

Want to discuss how to improve your data? Book a Consultation