Disband the analytics team

I mean, no, surely not. But maybe?

Eventually, the feds find you.

If you find yourself many years into running a Ponzi scheme, you have three choices:

You could try to keep the Ponzi scheme going! Keep raising more money from investors, and hope to die before you get caught. This isn't a great option—for a Ponzi scheme to sustain itself, it has to keep getting bigger, and raising more and more money is hard. Plus, the bigger it gets, the more likely it is that people might start asking questions and find out that it's a Ponzi scheme. And if it gets so big that the Securities and Exchange Commission finds out that it's a Ponzi scheme, they will arrest you and you will go to jail for 150 years. Not good. 

Second, you could stop doing the Ponzi scheme. This will probably upset your clients, because you will have to tell them that you don’t have any of their money, because you were running a Ponzi scheme. It will also probably upset the SEC, who will still arrest you and throw you in jail. But the judge might give you a lighter sentence if you stop doing the Ponzi scheme on your own, when it’s still a healthy Ponzi scheme and not a collapsing Ponzi scheme. I don't know. Seems risky. It might be better to hope that you don’t get caught. 

Finally, you could try to turn the Ponzi scheme into a real business? If your Ponzi scheme hasn’t imploded yet, you still have money in the bank. Do something real with it! Launch a startup. Trade crypto. Buy stocks that actually return a steady 11 percent a year. Open a coffee shop and sell it to a yogurt company for $900 million. This is hard to do, and it’s almost certainly better to do it without doing the Ponzi scheme first, but if you already stole the money, you might as well try to make the most of it. Best of all, if you pull it off, nobody will ever know that your business started as a Ponzi scheme.¹

Though the SEC won’t like this plan—you still ran a Ponzi scheme, and they will still want to arrest you for that—your investors might be ok with it. A thing about Ponzi schemes is that they’re great investments² so long as you get out before they blow up. Financially speaking—not ethically or morally, and definitely not legally—your early investors are doing great! They have been making a steady, above-market, risk-free return! They probably want to keep that. If you get arrested for running a Ponzi scheme, not only will you go to jail for 150 years, they will have to give back all of their returns. Bad for everyone.

So it seems like there's a trade here. You offer your existing investors a deal: If they don't tell the SEC that you've been running a Ponzi scheme, you’ll pivot your Ponzi scheme into a legitimate business. If it doesn’t work—if your coffee shop doesn’t make it—everyone loses, just as they would’ve anyway: Your investors lose their money, and you go to jail.³ But if it works, everyone wins: Your investors keep their previous returns, they get new returns from your successful company, and you don’t go to jail. It’s a graceful escape, if you pivot to the right thing. 

Data is a four-letter word

Well, not quite. Data is not a Ponzi scheme. Financial reporting and business intelligence are the only senses that many businesses have, and even the biggest cynics can’t contort themselves into saying that this stuff isn’t important. And operational telemetry, and the automated algorithms built on top of it, generates trillions in revenues in Amazon’s warehouses, on Google’s ad networks, and across factory floors and insurance companies and trading desks around the world. These are all real and enormously valuable applications of data.

But within the field of analytics—the practice of trying to make better strategic decisions with data and manual analysis—the track record is less compelling. Here, the animating narrative has long been the same: Data is full of useful insights. Companies that don’t find them are doomed to fall behind and die. Analysts are the experts who can extract those insights. Sure, we can’t live up to our full potential today—our data is too messy; our data is too unreliable; engineers are too busy to help keep it clean; our tools are broken; our workflows are broken; we are buried behind less important work; we distract ourselves; we don’t know how to organize ourselves; executives don’t support us; they treat us like second-class citizens; people don’t know how to work with us; people don’t invite us into the important rooms; people are illiterate.⁴ But on the other side of these problems, just over the horizon, there is an oasis of scholarship and shareholder value. 

Keep plugging, and we’ll finally clean up this mess.⁵ Keep plugging, and our potential energy will become realized energy. Keep plugging, and we’ll fulfill the long-told prophecy that data will become every company’s single most important asset. As data teams, this is our gospel: One day, when the glory comes, it will be ours; one day, when this war is won.⁶

As a philosophical point, I suppose this might be true. Though I have my doubts, it’s impossible to know what could be, given more time to build better tools, cultivate better cultures, educate more coworkers, and enlighten more executives. As a practical point, however, I’m not sure how much these hypotheticals matter. The question is not whether such a world exists, but if we will ever get there?

More pointedly, it's this: Would you, as a reader of this blog and presumably someone who is at least vaguely optimistic about this whole analytics endeavor, bet your entire bank account⁷ that we will—collectively, in general, on average—reach that promised land, in which data is democratized and insights are liberated and businesses are empowered and various other nouns are actioned? Or would you take the other side, and bet that most analytics teams slog along, heaving slowly forwards through the same briar patches that we’ve been stuck in for years? A decade from now, will most analytics teams—not your team, of course I have faith in you—be seamlessly driving critical corporate decisions? Will they be plucking new business opportunities out of haystacks of data? Will they be invited to weigh in every key debate, and have a seat around every key table? Will they be a competitive advantage?

Or will they continue to toil on the fringes, responding to queues of feckless questions, adding bits of value here and improving some marginal thing there, while continuing to say, to others and to themselves, that better days are ahead?

If my life savings were on the line, my answer is that our hope is a mistake, and most analysts will never escape their current lot. As we’ve talked about many times before, analytics might be an inherently flawed enterprise. Data is fundamentally messy and fundamentally biased. Turning this data into useful information requires rare analytical skills; turning information into better business results requires courage and a knack for persuasion. Unless we can teach these skills at scale, which we probably cannot, there will be little demand for the everyday analytics team’s work, and little advantage for the everyday company to invest in analytics. 

In short, my answer is that analytics—not as an industry or as a technology ecosystem, but as a discipline—might not work. The average company may never be able to make better decisions by hiring a team of average analysts. We can make dashboards and be operational accountants. But the fun, exploratory, “valuable” work may always be an indulgent, empty dessert, and never the entrée we want it to be. 

Which is an awkward answer, because in some rough way, my life savings is on the line. For many of us who work in data, we were hired to build this data-driven, insight-rich, collectively shared future. If we don’t, the whole apparatus—our jobs, our conferences, our startups, our academies, our statuses, all of it—will fall apart. 

Though no, that’s not exactly right. People in the analytics profession haven’t bet their life savings on building that future; we’ve bet our life savings on other people believing in that future. As long as people have faith in its potential, the machine can stay upright. Our executive overlords, many of whom probably feel that they aren't getting their money’s worth out of what we are, can mentally reclassify our salaries as down payments for what we will be. Vendors can keep promising future insights. We can keep trotting out explanations—data quality, data literacy, executive support, the distractions from AI, whatever—for why we aren’t yet a true analytics team. Our Ponzi scheme can stay solvent, until I’ll be gone and you’ll be gone, and another generation can worry about what comes next.

The walls close in

Honestly, it might work. People unquestionably accept that data is valuable; it wouldn't take much to keep that faith alive. 

But what if we assume we can't perpetuate it anymore? Suppose money gets tight, and the SEC (Stingy, Efficient CEO) runs out of patience. They demand to know how we plan on making our work worth what we’re paid. They’re onto our scheme, and our usual explanation—that it’ll all get better, once we complete a few eternal prerequisites—will no longer fly. This sounds like a Ponzi scheme, the SECEO will say, and in years of efficiency, Ponzi schemes get fired. 

One obvious answer is to stop doing the Ponzi scheme. Dissolve the analytics team. Trade in our big promises for something smaller. Instead of getting fired, take a reduced sentence: Twenty years to life of building reports and maintaining operational infrastructure. Become more like IT or HR—invaluable components in the corporate machinery, but with fewer delusions of grandeur. Accept our lot in life as bean counters, concede that our “insights” aren’t business-critical, no matter how much we beg people to use them, and count the beans.

I mean, maybe, but man, that is unsatisfying. People make decisions all the time, those decisions are rooted in facts, and facts are rooted in data. This basic logic seems so sound. Surely there’s a way to make this work? Surely there’s something more legitimate, and more reliably useful, that we could gracefully pivot into?

The adversarial analytics team

Go to any data conference, and there will almost certainly be a few talks about how data teams can be better partners with their business counterparts. It would be funny if someone gave a talk on how we could be more adversarial instead. Rather than showing people a dashboard of KPIs, use some black box algorithm to blend them all into a single composite. Rather than giving someone a day-by-day time series or a scatter plot with individual data points, make every chart a trend line. Rather than telling anyone the exact value of a metric, always round it to the nearest Fibonacci number: “Two months, we had 610 orders. Last month, about 610 again. This month, we’re up to about 987. We can’t share any more detail than that.”

This is obviously kind of dumb, probably kind of fraud,⁸ and…might kind of work? Because whenever we share metrics with people, this is mostly what they do anyway: Squint, decide if the line is going up a lot, up a little, flat, down a little, or down a lot. Though small statistical details matter for operational optimization problems, the analytical decisions that executives make are mostly based on these sorts of directional vibes. The line is wiggling; is it noise or is something changing? Customer retention is declining; let’s hire more account managers. Everyone is talking about AI; let’s send dump trucks of cash to the houses of famous AI experts. We struggle to sell to women; let’s make a horror movie about it.

This is part of why customer interviews are often more valuable than analyzing a bunch of data, even though the latter is the stuff of scientific inquiry, and the former is gossip. In so many cases, gossip is what people want.⁹ CEOs want to know what their customers are thinking. Behavioral data isn’t “truth;” it’s an observable proxy, the input to a kind of analytical alchemy that attempts to turn individual outcomes into generalizations about intentions. In some sense, the careful rigor of data analysis is a red herring—it’s what we were trained to sell, but not the main thing our customers want to buy. 

Perhaps that’s actually what’s held us back for so long: numbers, more or less. Though most business decisions are driven by numbers, those numbers matter because they define people’s loose mental models for how the world works, not because people need to know about the often-meaningless tedium of things like statistical significance. By talking primarily in the language of exact facts and figures, we make ourselves dismissible, because nobody really cares about the precise amount some metric went up or down. At the end of the day, executives care if the world still works the way they think it does. Everything else is just bluster and bikeshedding.

Disband or rebrand—that may be our choice. Analytics teams don’t have to be bound to data teams, and that association might be our problem. We might be better off finding other departments to operate in, and advertising ourselves as majors in that field with a minor in analytics, rather than the other way around. It might not be the fun thing we imagined, but it’s the useful thing we would get paid and promoted for. Because after flirting with a life of crime, we might need to start keeping our backstory a secret, and learn to blend in with everyone else.

1. I mean, they might be suspicious. If, say, CryptoFX stopped offering “‘risk free’ and ‘guaranteed’ crypto and foreign exchange investments,” and started selling cold brew, the SEC will probably have some questions. But still. When they show up, it would help if you had a real business and not just a Ponzi scheme.

2. issuing correction on a previous statement of mine, regarding Ponzi schemes. you do not, under any circumstances, "gotta hand it to them"

3. Sure, it’s more complicated than this. If your Ponzi scheme gets raided by the SEC, your investors will lose their illegitimate returns, but they might recover some of their principal. If you spend their money on a startup, they might lose all of it on your business’ expenses, like employees’ salaries and coffee beans and 350,000 H100 Nvidia graphics cards. Plus, if they make a deal with you and you get caught, I assume they might also go to jail for “allowing the fraud to continue unabated.” But that doesn’t change the overall point: Your investors would prefer that your Ponzi scheme becomes a legitimate business, and would not prefer that Ponzi scheme blow ups. 

4. “I can’t read Liz Lemon!”

5. All of our messy data sources will neatly write into a centralized warehouse. Each pipeline will be actively governed by a tightly written data contract. Data marts will be documented with encyclopedic completeness and precision. People will know where to go to ask questions. Answers will be delivered in-line and on-demand. People will be skilled in its language and its limits. Errors will promptly trigger alerts. And so on and so forth. 

6. Yes, fair, on one hand, this is a melodramatic and appropriative link. On the other hand, it’s a banger of a performance and everyone should watch it. I’m aware that our “struggles” as analytics teams are, relatively speaking, quite silly.

7. If you are a generational baseball talent who throws 100 miles an hour, hits absolute tanks, and steals 20 bases a season, do not bet on this! Do not bet on anything! Please! You are our Shohei! Our only Shohei! You'll never know how much we love you, please don't take our Shohei away!

8. “I’m sorry Mr. Gensler, but the most we can say about our revenue is it either rounds to 102,334,155 dollars or 165,580,141 dollars. We don’t keep track of what happens to the 63,245,986 dollars in between. No, you cannot look at my personal bank statement.” 

9. Gossip is the beginning of moral inquiry, as they say.