Being Early Looks Exactly Like Being Wrong

Published 2015-06-16 · Updated 2026-06-23 · By Alan Wizemann

Topics: Digital Transformation, Product Strategy & Development, AI & Machine Learning

At Target we have been making a lot of bets on technology that is not quite ready yet, which is a thing I have learned to be comfortable saying out loud even though it makes some people nervous. We were one of the first major retailers to put real weight behind mobile payments, we are deploying beacons in stores to make the app aware of where a guest is standing, we built for the smartwatch the week it became a thing, and we are chasing indoor-location systems that promise to do what GPS simply cannot do once you walk through the front doors. Some of those bets will pay off, and some will turn out to be early in a way that, from the outside looking in, is genuinely hard to tell apart from having just been wrong. Right now, in the moment, I honestly cannot tell you which of them is which.

That gap, the one between early and wrong, is the entire subject I want to talk about, because it is precisely where most corporate innovation quietly goes to die. Here is the uncomfortable truth of it. At the moment you actually place the bet, a brilliant early call and a foolish one look completely identical to each other. Both of them involve spending money on something customers are not asking for yet, that the infrastructure does not fully support, that your finance team cannot model on a spreadsheet, and that might turn out to be three full years too soon. There is no tell, no reliable signal. You cannot distinguish "ahead of the curve" from "off the curve entirely" in the moment (I have tried, and the people I trust most have tried), and anyone who claims they can is mostly just remembering their winners and quietly forgetting all of their losses.

So if you cannot pick perfectly, and you cannot, then the discipline has to live somewhere other than the picking – it has to live in how you bet. The first rule I held to is that an early bet has to be cheap enough to be wrong. The mistake big companies make, over and over, is treating an emerging-tech bet like a committed initiative, with a multi-year plan, a big budget, and a confident revenue projection sitting right there on the last slide. Now the bet simply has to succeed, because far too much is riding on it, so you cannot kill it when the signal tells you that you should, and you have quietly turned a cheap little experiment into an expensive and very public embarrassment. As we put beacons in stores, the framing was explicit on purpose: stand up the capability in a limited footprint, move quickly, get the learnings, and let what we learn inform whether and how we go any further. The goal of phase one is not revenue, it is to find out whether the thing is even real. You can afford to be wrong a dozen times if each wrong answer is cheap, and you cannot afford to be wrong even once if you bet the whole building on it.

The second rule is that you measure an early bet on learning, not on return. Asking a brand-new technology to prove its ROI in year one is exactly how you guarantee that you will only ever fund the things that are already obvious to everyone, which is functionally the same as never being early at all. The right question for an experiment is not "what did this earn." It is "what do we now know that we did not know before, and does that change what we would build next." A beacon pilot that teaches us how customers actually behave with location-aware features in a real, messy store is a success even in the cases where the feature itself does not stick, because the learning outlives the experiment that produced it.

The third rule is to bet on the capability and not on the gadget, because the specific device is almost always the part that ages badly while the underlying shift usually does not. Mobile payments are a good example of this – the particular early implementation matters far less than the durable bet that people will eventually pay with their phones, and being early there means we come to understand the customer behavior before our competitors do. The same logic applied cleanly to location awareness inside the store, whichever specific technology happened to deliver it in a given year. When you bet on the capability rather than on the hardware, you get to keep all of the learning even when the gadget you started with turns out to be a dead end. The smartwatch app might not survive the decade, but what we learned about glanceable, in-the-moment interactions absolutely does.

The fourth rule is the one my dad taught me long before I ever worked a day in technology, which is simply to diversify. You place a portfolio of small bets rather than one large one, precisely because you cannot tell the early winners apart from the early losers up front. Some of them go, most of them do not, and that is not a failure of judgment on anyone's part, it is just the plain math of being early. The job was never to be right about any single bet. The job is to make sure that when one of them finally does hit, you were holding it in your hand at the time, and that none of the misses along the way were ever big enough to actually hurt you.

I want to be honest about failure here, because the rules above are exactly the things that separate a survivable miss from a genuinely damaging one. Plenty of early bets simply do not work, or they work years later in someone else's hands instead of yours, and that is fine. The ones that truly hurt are never the bets that just quietly fail to pan out. They are the ones that have been allowed to grow too big to kill cleanly, where someone has attached a revenue number and a personal reputation to what was supposed to be an experiment, so the team keeps feeding it well past the point where the evidence justifies another dollar. Early-and-cheap recovers gracefully and nobody much remembers it. Early-and-overcommitted does not recover, and everybody remembers it.

The reason most large companies are bad at this is not a lack of vision, and it is not a lack of budget either. It is that their entire machinery is built to fund certainty, and early technology offers exactly none of that, so they either avoid early bets entirely and get steadily out-maneuvered, or they dress an early bet up in the costume of a sure thing, overcommit to it, get badly burned, and teach the whole organization a lesson that amounts to never trying again. The way through, as far as I can tell, is to stop trying to tell early from wrong at the moment of the bet, because you genuinely cannot. Bet small, measure the learning, back the capability instead of the device, and spread your bets across many of them. Do that consistently, and being early stops being a gamble you have to win every time. It becomes a thing you can simply afford to do, over and over again, until one of them turns out to have been early in the very best way.