Throughout my career I've seen some of the best analysts in the world work. When I've led data teams at mid-market and enterprise companies, I noticed something uncomfortable: almost none of them were doing what I'd call analytics. They were doing reporting. Fast, polished, well-intentioned reporting, but reporting nonetheless.
The distinction matters more than most people want to admit. Reporting tells you what happened. Analytics tells you why it happened, what's going to happen next, and what you should do about it. The first is a rearview mirror. The second is a competitive weapon.
Most teams I speak to are firmly in the rearview mirror business. And they don't know it.
The Illusion of Sophistication
Here's how it usually goes. A company invests in Snowflake, DataBricks or Microsoft Fabric. They hook up Tableau or Power BI. They hire a couple of analysts and a data engineer. Leadership gets a weekly dashboard. Everyone feels like they've built an analytics capability.
They haven't. They've built a reporting infrastructure. Which is useful, don't get me wrong, but it's table stakes, not competitive advantage.
The tell is in the questions being answered. If your analytics function is answering questions like "what was revenue last quarter by region" or "how many customers churned last month," you're doing reporting. The real questions that find the $10M hiding in your data or that tell you which customers are about to leave before they leave, require something fundamentally different.
"The real questions — the ones that drive decisions, that find the $10M hiding in your data — require something fundamentally different."
They require predictive modelling. Causal inference. Segmentation that goes beyond pivot tables. Statistical rigour that most analytics teams simply don't have the capacity to deliver. Heck most don't even know the techniques.
Why Hiring Won't Fix It
The instinct when you recognise this gap is to hire your way out of it. Bring in a senior data scientist. Maybe a machine learning engineer. Upgrade the team.
I understand the instinct. I've had it myself. But it's mostly wrong, for two reasons.
First, the talent market for genuinely world-class analytics capability is brutally competitive. The people who can do real predictive modelling, who understand causality, who can build models that actually change decisions — they're at Google, at the top hedge funds, at the handful of companies paying genuinely exceptional packages. You're not getting them. You're getting someone adjacent to that, at best.
Second, even if you find someone great, one person doesn't make a world-class analytics function. They spend half their time on data cleaning, stakeholder management, and building dashboards they're too qualified to build. Then they leave because that work is frankly beneath them.
What world-class analytics actually looks like
At McKinsey, the questions we answered weren't "what happened last quarter." They were things like: which customer segments have the highest lifetime value potential that we're currently underserving, and what's the precise intervention that would move them? What's the causal driver of margin compression in this business unit, and how much of it is structural versus cyclical? If we change pricing in this market, what's the second-order effect on retention?
These aren't dashboard questions. They're modelling questions. Statistical questions. They require the kind of rigorous, hypothesis-driven analytical work that most teams don't know how to do.
The New Trap: Fast but Unreliable
Now there's a new version of this problem emerging, and I'm watching smart companies fall into it in real time.
The promise of AI-powered analytics tools is speed. Ask a question in plain English. Get an answer in seconds. No analysts required. And for basic ad-hoc querying, this works. It's genuinely useful. Non-technical leaders can get data without waiting for an analyst. That's real value.
But here's what nobody is saying loudly enough: speed without sophistication doesn't get you to world-class. It gets you to a new kind of mediocre. You're trading slow, manual analytics for fast, unreliable analytics. You're swapping one problem for another.
If the underlying analytical capability isn't there then you're just getting wrong answers faster. And wrong answers delivered at speed, with confidence, are more dangerous than slow right answers.
| What most teams have | What world-class looks like |
|---|---|
| Dashboards that show what happened | Predictive models that drive forward decisions |
| Ad-hoc queries that answer simple questions quickly | Causal analysis that finds the real levers |
| Analysts spending 80% of time on data wrangling | Sophisticated segmentation that changes how you operate |
| Models that took weeks to build and nobody trusts | Fast answers that are also statistically rigorous |
| Speed or sophistication — never both | Analytics that the business actually acts on |
The Only Path to World-Class
So what does it actually take? I've thought about this a lot, and I think the answer is uncomfortable for a lot of people.
You need speed and sophistication. Both. Not one or the other. Fast reporting isn't analytics. It's just fast reporting. Slow, rigorous analysis that takes weeks to deliver is analytically sound but operationally useless in most business contexts. You need the ability to move fast and get it right.
For most organisations, the honest answer is that this is now an AI problem. The teams that are pulling ahead aren't the ones that hired more analysts, they're the ones that figured out how to apply genuine analytical sophistication at machine speed. Models that would have taken a senior data scientist two weeks now can be built in minutes. Segmentation that would have required a team can be done on demand. The bottleneck isn't analyst time anymore. It's analytical ambition.
But only if the underlying capability is genuinely sophisticated.
Not just fast querying dressed up as AI analytics. Real predictive modelling. Real causal inference. Real statistical rigour applied to your actual, messy, real-world data.
That's the standard worth holding yourself to. Not "are we faster than we used to be" but "are we doing the kind of analytics that actually changes decisions and finds value that was previously invisible?"
Most teams aren't there yet. The ones who know they aren't are already moving.
Where to Start
Here's what I've found after working with dozens of data teams: knowing you need to get to world-class and knowing how to get there are two very different things. The gap between "we should be doing more advanced analytics" and "here is our roadmap to actually doing it" is where most organisations get stuck. It's not laziness. It's that figuring out which use cases will actually transform your business — versus which ones sound impressive but won't move the needle — requires a rigour that most teams haven't developed yet.
Done well, mapping your advanced analytics path looks like this: identify the decisions your business makes repeatedly that are currently based on intuition or incomplete data, understand which of those decisions are high enough stakes to justify a proper model, and sequence them in order of feasibility and impact. That's it. It's not complicated in principle. But it's surprisingly hard to do clearly when you're inside the organisation and too close to the day-to-day.
That's why we built a free guide to help teams work through exactly this — from understanding what advanced analytics actually means for your specific business context, through to identifying the use cases worth prioritising.
TrueState 360 strategy guide
Work through the full process: from business context to a prioritised analytics roadmap.
And for a limited time, we're also offering a free 30-minute strategy session where we'll do a first pass of this together, live. No slides, no pitch — just a working session to identify the two or three use cases that could genuinely transform what your team is capable of.
If you've read this far and recognised your team in any of it, that session is probably worth 30 minutes of your time.
Book Your Free Strategy Session →
The author is the founder of TrueState and a former McKinsey consultant specialising in analytics strategy. TrueState builds agentic analytics capabilities for data teams that want to move from reporting to genuine analytical sophistication.