An empty dance floor
Awkwardness kills data-driven decision making
Do you remember those moments, like a wedding, or a company party, with a DJ who is trying hard to get the party going. The dance floor is empty, the music is playing, and everyone is standing around the edges holding drinks, glancing at the scary open space, waiting for someone to break the ice. Going out there means being the weird one, or the very confident one. Everybody waits, and the waiting becomes its own justification for more waiting. The DJ keeps playing banger after banger, but the floor stays empty.
This is a good metaphor for what goes wrong between Data teams and the rest of the business. Ask anyone why Data teams struggle to demonstrate value and you will get some variation of the same answer: “They don’t explain things in simple terms, they use a lot of technical terms to describe things to build”. Data people are too technical, too jargon-heavy, too in love with their methods and tools. If they just learn to speak business, everything would be amazing.
On the other hand, Finance is full of jargon: EBITDA, amortization, deferred revenue. Legal is even worse, it’s not even English, it’s a different language altogether. Marketing has its own dialect of attribution models, incrementality, 4Ps, brand lift.
Executives interact with these functions without much friction, and they don’t question their value. Every CXO knows that they need a finance department, a legal department, a HR department, even a marketing department.
The difference with Data is not that the jargon is harder, but that most executives have a working mental model for what finance or marketing are for. They have built an intuition over multiple years of exposure, they know roughly what a P&L tells them, even if they cannot build one from scratch. With Data however, that mental model often does not exist, but admitting that feels dangerous.
Doing Data in corporations is like sex for teenagers. Everyone talks about doing it. Everyone claims they are doing it. Very few are actually doing it for real. Being “Data-driven” is the correct thing to say since 2015, like saying things like “innovation” or “customer-centric”. It signals that you are in the race, you live in the right century, even if your data infra is a guy named Dave running queries in a spreadsheet.
The problem with this is that when everyone claims fluency, nobody knows who actually understands the function for real. The VP who funded the Data org with a bunch of ML engineers to build a recommender system to radically change the way their org operates sits in the same meeting as the VP who has not opened PowerBI since the training session two years ago. They both nod along when the Data team presents something, they both say the right things at face value.
There is a coordination problem at play here. Both sides would benefit from honesty, but neither moves first because they cannot be sure the other will follow. Game theorists call this scenario “Stag Hunt”, when there is a conflict between safety and mutual cooperation, modeled by two hunters choosing to hunt a stag (high reward, requires both) or a hare (low reward, safe). In this game trust is crucial, as each player’s fears of betrayal can lead to an inferior, non-cooperative outcome.
Imagine player one as a data practitioner that is about to present findings about a LTV model to senior stakeholders. She could start with a 101 explanation of the concept of LTV, what is it for, how it is estimated, how the accuracy is evaluated. But what if the stakeholders already have a good understanding of all that? Starting too basic from fundamentals in a “teaching” way risks coming across as condescending. So more often than not we skip the basics and dive into the results, hoping that the context is understood.
Player two is a business stakeholder sitting in the meeting, looking at a chart he doesn’t fully understand. He could ask for clarification, but he has been saying “data-driven” in every all-hands for two years. Asking “wait, what does statistically significant mean?” feels like confessing he has been faking it (spoiler: he did). You need a good level of self-confidence to pull off a Jeremy Irons move.
The resulting sub-optimal equilibrium is that both players refuse to cooperate. The data practitioner presents results assuming a high level of familiarity, the stakeholder nods without understanding a thing. The meeting ends, everyone says it went well, but nothing gets done after that.
How do you fix this?
The stakeholder move is showing vulnerability. Ask for that 101, say “I want to make sure I’m reading this correctly, can you walk me through the methodology in simple terms?”. Frame it as due diligence, not ignorance.
The data practitioner move is empathy. Offer to give that explanation as an act of alliance-building, removing the stigma of not knowing.
I once had a VP1 stop a presentation about marketing mix modelling and say “I’m going to be honest, I don’t know what a channel elasticity is. Can we start there?”. The room was relieved. Two other people admitted they were confused about the same thing. The meeting ran longer than planned but it was much more productive. The cost of that moment of vulnerability was a few seconds of awkwardness, in exchange for long term value from better collaboration and trust.
Both have to jump into the empty floor and dance like nobody is watching.
As Data lead, I have a problem with the term “data literacy”, because the flip side is “illiteracy”, which is quite harsh as a term. Call it “data office hours”, “data for everyone”, something that signal no judgment, no mansplaining, no testing. You are building a shared language, not grading people.
The other thing that helps is inverting how you present. Most data teams present like academics: methodology first, results at the end. Flip it, and start with the business implication. In the LTV example, open with “we can identify which customer cohorts will generate the most value over their lifetime, and use that to prioritize acquisition spend, product investment, and experimentation”. That sentence lands because it maps to things the stakeholder is already accountable for. Once they are bought into why it matters, you have an opening to explain all the building blocks that are needed to make this happen, so that they understand is not costless wizardry, but there is work to do and fund, and also limitations and caveats.
It takes two to dance, but the awkwardness dissolves as soon as it is shared.
Do you like this post? Of course you do. Share it on Twitter/X, LinkedIn and HackerNews
Of course, a VP is in a better position to afford that move, compared to a mid-level manager asking basic questions in front of their boss