Desktop distribution is broken, and the auction is why

The number that ends the conversation

	One-time licence	Monthly subscription
Price	EUR 29 once	EUR 29 per month
Gross lifetime value	~EUR 29	~EUR 350 at 12 months
Affordable customer acquisition cost at 30% margin	~EUR 9	~EUR 105
Clicks per customer at 3% conversion	33	33
Affordable cost per click	EUR 0.27	EUR 3.18

Both are bidding on "best screen recorder for mac". One can pay nearly twelve times more for the identical click. There is no creative, no landing page and no bid strategy that closes a twelve-fold gap, because the gap is not in the advertising. It is in the pricing model.

An auction is a mechanism for discovering who values a thing most. It works exactly as designed. The uncomfortable conclusion is that for a one-time-licence desktop app, working as designed means you lose.

What people try instead, and why each one stalls

Content and SEO. Genuinely works, and takes eighteen months. It is also increasingly mediated by AI answers rather than blue links, which changes what "ranking" means without removing the eighteen months.

Communities. Works once. Launch on Hacker News, get a spike, and discover that the spike is not a channel. Communities reward novelty, and your app is only novel for one week of its life.

Affiliates and deal sites. Works, and prices your product at 50 percent off forever. The discount is not the cost; the anchor is.

Curated bundles. A real distribution surface, and the only browsable shelf desktop software has. You give up the customer relationship for it, which for some apps is the right trade and for others is the whole business.

The app stores. The Mac App Store is a sandbox tax and a discovery surface people do not browse. The Microsoft Store is worse.

The asset nobody counts

Every desktop app has an audience of people who have proven, within living memory, that they will download and install desktop software. That is an extraordinarily specific behaviour. Roughly nobody in the general population did it this month, and everybody in your user base did it at least once.

Advertising exists to find those people. You already have them.

The reason this has not been turned into a channel is not that nobody thought of it. It is that every previous attempt monetised the wrong moment, paid on the wrong event, and gave partners the wrong amount of control — and got the entire category classified as malware for it. That history is the subject of the next piece.

What has to be true for the trade to work

For an app to be willing to show another app's card, three things have to hold:

1. The moment cannot be hostile. After install, inside the product, at a moment the host app chose. Never inside an installer, never in competition with a Next button.
2. The unit has to be honest. Paying on the install event means the seller's incentive is volume regardless of fit. Paying on retention at 48 hours means an irrelevant card earns nothing.
3. The count has to be checkable. If the network reports the numbers and both sides have to believe it, the network will eventually be wrong in its own favour. Two independent signatures per billed line removes the question.

Those three constraints are the whole of GHAN's design, and they read as arbitrary until you know which specific failure each one is a response to.