Costs
Part of Platform trends: what beginners should know
Platform trends platforms: what the good ones share
Platform trends platforms and the cost of moving between them: rebuild, relearn, reacquire, remeasure, plus how to price concentration before you commit.
Every platform shift has a price, and it is almost never the price people quote. The visible cost is the work of rebuilding something. The larger costs are the ones with no invoice: the audience you have to reacquire, the instrumentation that stops comparing, and the months during which you know less about your own operation than you did before.
The durability tests decide whether a shift is worth reacting to at all. This page sorts what the reaction costs, so the figure can be estimated before you commit rather than discovered afterward.
What to take away
- Four cost categories exist, and only the first one shows up in a plan. The other three are what makes the estimate wrong.
- Concentration is a cost you are already paying, whether or not you ever move. It just falls due all at once.
- The exit cost, not the running cost, is what you should be measuring when you decide how deeply to commit to anything.
The four categories
| Category | What it is | Why it gets missed | Rough shape |
|---|---|---|---|
| Rebuild | Producing the thing again in a new form | It is the only one that looks like work | Estimable, and usually estimated too low |
| Relearn | Finding out what works on unfamiliar ground | It feels like ordinary operating | Months, and it cannot be hurried |
| Reacquire | Building the audience again from a lower base | The old audience is assumed to follow | Largely unrecoverable |
| Remeasure | Rebuilding the ability to tell whether anything is working | Nobody budgets for instrumentation | The longest tail of the four |
Relearn is the category that surprises people most. Format conventions, timing, what an audience on a given surface will tolerate, what a normal week looks like: none of that transfers, and none of it can be read from a document. It has to be acquired by operating, which takes as long as it takes.
Remeasure is the one that does the most quiet damage. Your history is a series of numbers whose meaning depends on the surface that produced them. Change the surface and the series stops being a series. You are not starting from zero; you are starting from a position where you cannot tell whether you are improving, which is worse than zero because it invites confident nonsense in both directions.
Concentration, priced honestly
If most of what you do depends on one operator, you hold a concentrated position. That is not a mistake by itself. Concentration is often the correct call, because attention concentrates too, and spreading yourself thinly across six surfaces can leave you with a presence on none. Why attention concentrates is structural, and it is not going to reverse. What is a mistake is holding the position without pricing it.
The right frame is concentration risk: the exposure that comes from having your outcomes depend on one thing rather than several. It is a standing cost paid in optionality, and it is invisible right up to the point where it is the only thing that matters.
Related but distinct is vendor lock-in, which is about the cost of leaving rather than the cost of staying. The two come apart in a useful way.
You can be heavily concentrated with a low exit cost, if what you produce is portable and your audience relationship is not held by the operator. You can also be lightly concentrated with a very high exit cost, if the small thing you depend on has no substitute.
What to measure before you commit
Three numbers, all of which you can estimate today. They are the three columns worth adding to a dependency inventory.
The portability of what you produce. If your material exists in a form you hold, in a format that moves, the rebuild cost is small. If it exists only inside somebody's product, the rebuild cost is the whole of it.
The ownership of the audience relationship. If you can reach your audience without the operator's permission, reacquisition is a marketing cost. If you cannot, it is not a cost, it is a loss.
The substitutability of the dependency. How many things could do this job. One is a very different position from three, and it is worth knowing which you are in before the question is urgent rather than after.
The decision this actually informs
None of this argues for or against any particular move. It argues for holding the exit cost in mind while you build, because the cheapest time to reduce it is at the start and the most expensive time is when you need it reduced.
Practically, that means keeping a copy of what you produce in a form you control, keeping at least one channel to your audience that nobody else can switch off, and knowing which of your dependencies has no substitute.
None of that requires predicting anything, which is fortunate. A watch notices changes; it does not forecast them.
Where the cost is forced rather than chosen, the same categories apply with the timeline removed, which is why what happens when a dependency stops belongs in the same file as this.
Common questions
Is it cheaper to be on several surfaces from the start?
Cheaper in exit cost and more expensive in everything else. Running well on one surface is hard; running well on four is usually four times as hard rather than a quarter as risky. The honest answer is that this is a real trade with no general solution, and the right point depends on how portable your material is.
How do I put a number on the reacquire cost?
You mostly cannot, and pretending otherwise is worse than admitting it. What you can do is establish what fraction of your audience you could reach if the surface disappeared tomorrow. That fraction is the part that is not at risk, and it is usually much smaller than people expect.
Does a smaller operation face this differently?
The costs are the same in kind and the buffer is smaller, so the sequencing matters more. A large operation can absorb a bad quarter of remeasurement; a small one may not have four quarters. That argues for keeping the exit cost low rather than for diversifying early.
What about the cost of not moving?
Real, and much harder to see, because nothing visibly breaks. It shows up as a slow decline that gets attributed to everything except the surface, which is why the order to check in when your numbers fall puts the platform hypothesis last rather than first.
