The deterministic illusion of digital marketing

Digital marketing sold everyone the same seductive lie: that the relationship between what you spend and what you get is linear, traceable, and clean. Change the ad copy and the click-through rate moves. Change the button colour and conversion moves with it. Attribute the sale to whatever happened right before checkout, put it in a report, and present it on Monday as cause and effect. It feels like science because it comes with numbers attached, and numbers feel like proof.

That's the trick. It's a very convincing illusion of determinism sitting on top of a system that was never deterministic to begin with.

Cause and effect are not the same as measurable and unmeasurable

Multiplication is a simple linear system. Cause goes in, effect comes out, and the relationship between them never changes no matter how many times you run it. A market is nothing like that. It's a network of people, influenced by things that happened weeks ago, conversations with people who never clicked anything, a competitor's price change, a mood, a memory of a bad customer service call two years back. Michael Mauboussin has a line worth sitting with here: complex adaptive systems effectively obscure cause and effect. That's not a soft, poetic point about marketing being unpredictable, it's a structural fact about the kind of system a market actually is.

Short-term digital metrics don't obscure cause and effect the way the market does, they do something worse, they replace it with a fake, tidy version that looks exactly as convincing as the real thing. Last-click attribution hands 100% of the credit to whatever happened right before the sale, and 0% to the twelve things that happened before that and actually built the trust the last click cashed in on. The report is clean, and the story it tells is mostly fiction.

Why marketing keeps falling for it

This isn't a stupidity problem, it's an incentive problem. A weekly dashboard needs a number that moves, and brand trust, reputation, and long-term demand don't move weekly, they accumulate quietly and then show up all at once, months later, with no clean line back to the campaign that built them. So marketing teams optimise for the thing that's visible on a Tuesday, because that's the thing that gets you through the Tuesday meeting, and the thing that's actually building the business gets starved because nobody can put it in a slide.

John Gall's warning about complex systems applies almost word for word here: in complex systems, malfunctions may not be detected for long periods, if ever. A brand can be quietly eroding for a year while every short-term metric in the dashboard is green, because the dashboard was never built to see erosion. It was built to see clicks.

The network doesn't care about your attribution model

Cut a brand-building spend line to hit a quarterly efficiency target, and the cost doesn't vanish, it moves. It shows up eighteen months later as rising acquisition costs, because the pool of people who already trusted you got smaller and now every new customer has to be persuaded from zero. The finance model says the cut saved money. The market shows the cost just moved house and came back with interest.

This is the exact mechanism behind a lot of the bad automation and bad service design I've written about before. Someone optimises one clean, countable metric, staff hours, cost per click, conversion rate, and it looks completely rational in isolation. Nobody accounts for what happens to the rest of the system once the thing they cut stops being free.

Big companies and small companies aren't playing the same game

Small companies live month to month. That's not a strategic choice, it's survival, cash flow forces the horizon down to whatever keeps the lights on next quarter. A short-term, metrics-chasing tactic makes sense for a company that genuinely might not exist in six months if this quarter goes badly.

Big companies don't have that excuse, and that's exactly the problem. A company with the balance sheet to think in years instead of quarters, that borrows the small company's short-term playbook anyway, isn't being lean. It's playing a game built for a different horizon than the one it's actually operating on. Over the medium to long term, that mismatch shows up as underperformance, brand erosion, and marketing that technically hits its quarterly numbers while quietly getting worse at the thing marketing is actually for. And because the biggest players in a category set the tone for what "good" looks like, once they start optimising for next quarter instead of next decade, the whole category's marketing effectiveness drifts down with them. Everyone ends up fishing in the same shrinking, short-term pond, including the companies that never needed to.

This is basically the problem military strategists have known forever: a purely logical strategy stops working the moment your opponent can predict it, because everyone doing the "rational" analysis, on the same short-term inputs, arrives at the same short-term move. The advantage was never in reacting faster than everyone else to this quarter's numbers. It was in being the one player whose next move wasn't derivable from the same spreadsheet everyone else was reading.

There's a scene in Airplane! that nails this better than most strategy books, precisely because it's a joke. As the plane comes in for its emergency landing, someone in the tower suggests turning on the searchlights to help guide it down. The old military hand in charge shuts it down instantly: no, that's exactly what they'll be expecting. It's funny because there's no "they." There's no enemy plotting around the searchlights. It's a landing, not a battle, and applying combat logic to a situation that doesn't need it is absurd. But the joke lands precisely because everyone in the audience has seen the real version of that scene a hundred times, the tactic that's smart specifically because it isn't the obvious, metric-optimal move.

That's the move most short-term marketing strategy has forgotten how to make, and it's a different mistake depending on who's making it. For a small company, chasing the short-term metric is often just staying alive. For a big one, it's giving up the one advantage size was supposed to buy: the room to play a longer, less predictable game than everyone forced to live month to month.

Most of it is wasted, and that's fine

There's an old, half-joking line about advertising: half the money is wasted, nobody knows which half. It's usually told as a complaint, but it's actually a clue.

Marketing effectiveness doesn't follow a neat bell curve where most campaigns perform reasonably and a few outliers do a bit better or worse, it follows a fat tail. The large majority of what you spend, most weeks, most channels, most creative, does close to nothing measurable. Then a small fraction of it does something wildly disproportionate, a piece of work that reshapes how a whole category thinks about a brand, a campaign that keeps paying back years after the budget line closed, a single decision that outperforms the rest of the annual plan combined. That's a Pareto distribution, not an average, and it behaves nothing like the tidy, linear world a monthly dashboard assumes.

This is exactly why short-term, per-campaign metrics are the wrong tool for judging whether marketing spend is working. A metric built to reward consistent, predictable weekly performance will always flag the fat-tail bet as inefficient, because in isolation, most of the spend feeding that bet genuinely does look wasted. Cut the "underperforming" 80% to protect this quarter's efficiency number, and you don't just save money. You also remove the exact conditions that let the outsized 20% happen at all, because the outlier rarely announces itself in advance. You don't get to keep the win and cancel the noise around it. The noise is often the cost of admission.

This is the fat-tail version of everything the rest of this piece has been circling. A short-term metric can only see the average, it was never built to see the tail, and the tail is where almost all of the actual value in marketing lives.

What to actually do about it

You can't fix this by finding a better dashboard. That's the same trap wearing a nicer interface. The fix is treating any short-term metric as a flashlight, not a map. It shows you one thing clearly and leaves the rest of the room dark, and the job isn't to trust the flashlight, it's to remember there's a whole system out there you're not measuring, and it's the part that actually decides whether the business is still standing in three years.

The brands that get this right aren't the ones with the tightest attribution model. They're the ones who've accepted that the thing that matters most will never show up cleanly in a weekly report, and who keep investing in it anyway.