Marcus and the 16 Competing Truths

Marcus is leaning so close to the monitor that the pixels are starting to look like a grid of tiny, glowing stained-glass windows. His face is washed in a pale blue hue, the kind that makes you look like you’ve been underwater for 16 minutes longer than humanly possible. On the screen, 16 different charts are competing for his attention. There are bar graphs that look like city skylines, pie charts that have been sliced into 56 unrecognizable slivers, and a heat map that suggests the entire Northeast is currently on fire.

He has all the data. He has 106 tabs open across three monitors. He has the most expensive business intelligence suite that money can buy-a system so advanced that the implementation team spent 256 days just ‘aligning the architecture.’ And yet, Marcus is paralyzed. He’s been asked a simple question by the board: ‘Which of our products is actually profitable?’ Not ‘which has the highest revenue,’ not ‘which has the most engagement,’ but which one is actually keeping the lights on. He looks at the 16 charts again. None of them agree. The marketing dashboard says Product A is a superstar because it has 406% year-over-year growth in ‘brand sentiment.’ The inventory report says Product A is a nightmare because it has a 76% return rate. The finance ledger, buried in a 1,006-page PDF, suggests that after shipping and storage, Product A is actually losing them 6 dollars for every unit sold.

The Illusion of Control

The data is correct. Every single one of those reports is pulling from a verified source. But the answer is wrong. Or rather, the answer is missing, drowned in a sea of ‘correct’ data points that refuse to speak the same language. This is the modern corporate purgatory: we are starving for clarity while drowning in information.

The Ghost of Human Action

Most BI tools are built by engineers for engineers. They are built to move data, not to deliver wisdom. They are built to be ‘flexible,’ which is usually just code for ‘we didn’t want to make a choice, so you have to.’

Take Jackson R., for instance. I’ve known Jackson for about 6 years. He’s a prison education coordinator, a job that requires a level of patience I can’t even begin to simulate. Jackson is responsible for tracking the progress of 456 inmates across 16 different vocational programs. The state recently moved to a new reporting system-a high-tech marvel that tracks everything down to the second.

The New System Deployment

Jackson showed me a report last week. It was 156 pages of ‘correct’ data. It showed that the inmates were spending an average of 46 minutes per day on the literacy software. It showed that ‘interaction events’ had increased by 86% since the new tablets were introduced. On paper, it was a triumph.

‘Look at this,’ he said, pointing to a graph showing a massive spike in engagement at 2:06 PM every day. ‘The guys have figured out that if they just keep tapping the ‘Next’ button rhythmically, the software records it as ‘high-intensity participation’ and the guards don’t bother them because they look busy.’

The data was 100% accurate. The button was being pressed. The engagement was occurring. But the answer-are they learning?-was a resounding ‘no.’ Jackson R. is fighting a war against the ‘correct’ data every single day because the people who designed the system forgot that a data point is just a ghost of a human action. If you don’t know the human, you can’t read the ghost.

Confusing the Map for the Territory

We have confused the map for the territory. Worse, we’ve started building maps that are more complex than the territory itself. I’ve seen companies with 16 different versions of ‘Total Revenue.’ One for the sales team (gross), one for the finance team (net), one for the tax man (depreciated), and one for the investors (adjusted EBITDA). All of them are correct. None of them help the CEO decide whether to hire 56 new developers or cut the marketing budget by 16 percent.

The retreat to ‘gut feeling’ is a direct result of this data exhaustion. We need a bridge between the technical precision of the data architect and the pragmatic needs of the decision-maker. It’s why I find myself gravitating toward teams like Debbie Breuls & Associates who don’t just hand you a hammer and call it a house.

Stripping Down to the Core Truth

I remember a project where I was asked to help a retail chain understand their customer churn. They had a report that was 86 columns wide. It tracked 106 different variables, from the color of the customer’s shoes (I kid you not) to the weather on the day of their last purchase. The data scientists were very proud of it. They had used a machine learning model that took 36 hours to run on a supercomputer cluster.

Variable Reduction: From 106 to 6

They told the VP of Operations that the primary driver of churn was a ‘multivariate correlation between Tuesday afternoon rain and a decrease in loyalty point redemption.’ The VP looked at them for about 16 seconds, then turned to me and asked, ‘What do I do with that? Do I buy umbrellas?’

We spent the next 46 days stripping the report down. We focused on the 6 things that actually mattered: Did the product work? Was it in stock? Did the cashier smile? Was the price fair? Did the store smell like old broccoli? And finally, did the customer come back within 26 days?

The answer-that the technology was driving customers away-was hidden under a layer of technical optimism. The 106-column report had ‘corrected’ for the scanner failures by labeling them as ‘manual override events,’ which the model interpreted as ‘high-touch customer service.’

AI: The Faster Way to the Wrong Answer

Jackson R. tells me that the most dangerous man in a prison is the one who has learned how to game the metrics. He says the same is true in business. When we reward people for ‘increasing engagement,’ they will find a way to increase the number of button presses, regardless of whether those presses create value.

We are currently obsessed with Artificial Intelligence, believing it will finally solve the ‘Answer’ problem. But if you ask an AI to maximize profit, it might suggest you fire all 456 of your employees and sell the office furniture. Technically correct. Strategically suicidal.

The Single Sentence Over 156 Pages

We need to stop asking ‘what data do we have?’ and start asking ‘what do we need to know to act?’ Those are two very different questions. One leads to a 156-page report that no one reads; the other leads to a single sentence that changes the trajectory of a company. I’ll take the sentence every time, even if it doesn’t come with 16 color-coded charts.

The glare from Marcus’s monitor is still there, but he’s finally turned it off. He’s sitting in the dark now, the only light coming from the streetlamp 46 feet below his window. He’s realized that the 15 charts were just a way to avoid making a hard choice. He doesn’t need more ‘correct’ data. He needs the courage to look at the 6 people on his leadership team and ask them what they see when they aren’t looking at their screens.

The ghost in the ledger isn’t a technical error. It’s the human element we tried to automate away. And until we bring that back, the data will always be correct, and we will always be wrong.