Reverse Engineering a Business

Nobody hands you the manual

Here is the thing about the decisions that matter most in business. You almost never get the manual.

A competitor is beating you and will not explain why. A platform changes its algorithm overnight and publishes nothing. A company you might buy hands you a data room that shows what they want you to see, not the levers that actually run the business. A market is clearly working for someone, and no one will tell you where the money really comes from.

In every one of those cases you have the same job. Figure out how the machine works from the outside, by watching what it does. That is reverse engineering, and it is most of what real research actually is.

I have been doing this since I was fourteen, taking systems apart to see how they worked. Then for twenty years I did it at scale, in businesses where the platforms and algorithms we depended on changed the rules without warning and never sent a memo. You learn fast that the people who win are not the ones with the manual. They are the ones who can rebuild the manual from the outside, faster than the system changes.

This is how we do it.

What reverse engineering actually is

Forget the word for a second. The method is simple to state and hard to do well.

1. Watch the outputs. What does the system actually do? Not what it says it does. What it does.
2. Form a theory of the machine. What would have to be true, underneath, to produce those outputs?
3. Find the cheap test. What small thing can you observe or try that tells one theory apart from another?
4. Keep the theory that survives. Then use it, but hold it loosely, because a model is not the truth. It is the best guess until the next observation.

That is it. The discipline is in refusing to skip step three. Most people jump straight from watching outputs to a confident story about the machine, and the story is usually the one that flatters what they already believed. Reverse engineering is the habit of making the story earn it.

Doing it to a market

Say a competitor is winning and you do not know why. Everyone has a theory. Better product. More funding. Got lucky. Those are stories, not machines.

Here is what watching the outputs looks like instead. Where do their customers actually come from? You can often see the shape of that from the outside. Search, paid, referral, partnerships, each leaves a different trace. What do they charge, and to whom, and what does the price ladder tell you about who they are really built for? How fast do they ship, and what does that say about their team and their tech? What do their own job postings reveal about where they are investing? Reviews, support forums and churn complaints tell you where the product actually hurts.

None of that is one answer. It is a pile of observations. The work is turning it into a theory of the machine. Maybe the real engine is not the product at all. Maybe it is a distribution channel they locked up early and everyone else is fighting for scraps of. If that theory is right, then out-building their product is the wrong move, because the product was never the constraint. The channel was.

That is the payoff of doing it properly. You stop competing on the thing you can see and start competing on the thing that actually decides the outcome.

Doing it to a business you might buy

Diligence is reverse engineering with a deadline.

The data room shows you the numbers the seller chose. Your job is to rebuild the real machine behind them. Which customers actually drive the profit, and which ones just drive the top line? Recompute it. What holds those customers, and is it something durable or is it one relationship, one contract, one channel that could vanish? What does the cost structure assume, and what breaks if volume doubles or the main supplier raises prices?

The seller will present a clean story. Growth, margin, a tidy reason for both. Reverse engineering is how you find out whether the story is the machine or just the paint. The gap between the two is exactly what you are being asked to pay for, and it is where most bad acquisitions are made.

Doing it to a platform

This is the one I have lived the most. When a platform you depend on changes, it does not tell you what changed. Traffic moves, rankings shift, the numbers go strange, and you get no explanation.

You reverse-engineer it the same way. Isolate what moved and what did not. Form a theory of what the change was optimizing for, because a platform change almost always has an intent behind it. Find the cheap test that separates your theory from the alternatives. Then adapt to the new machine before your competitors have finished arguing about whose fault it was.

The businesses that survive platform shifts are not the ones with inside information. They are the ones who can read the change from the outside and move first. I have had entire models repriced overnight, and this is the skill that got us to the other side. More than once.

The honesty that makes it work

Reverse engineering has a failure mode, and it is worth naming because most people fall into it. You find a theory that fits, you like it, and you stop. You stop testing because the story is comfortable. That is not analysis, that is a story with a chart attached.

The discipline is the opposite. Hold every theory loosely. Look hardest at the evidence that would prove you wrong. State what would change your mind before you go looking, so you cannot move the goalposts later. A reverse-engineered model of a business is always a guess. A good one is a guess that has survived every cheap attempt to kill it, and that you are still willing to update tomorrow.

That is the difference between someone who sounds smart about your competitor and someone who can actually tell you where to push.

Why this is the whole job

Finding the leverage point in a business, which is what we are really hired to do, is reverse engineering almost every time. The constraint that matters is rarely the one written on the wall. You have to infer it from how the system behaves, test the inference, and then aim there instead of at the obvious.

If you have a competitor you cannot explain, a target you are not sure how to value, or a platform that just changed the rules on you, that is the work. Tell us what you are trying to figure out, and we will start from the outputs.