GHAN— by singular Get started

GHAN vs. affiliate networks for desktop software

An affiliate network pays a commission when someone buys, usually attributed by a cookie and usually alongside a discount code. A cross-promotion network pays a fixed amount when someone installs and keeps using your software. The difference in what is being bought changes almost everything downstream.

The core difference

Affiliate networkCross-promotion (GHAN)
Paid onA saleAn install still in use at 48 hours
Typical cost20–40% of revenue, recurring for some programsFrom EUR 12–50 once, floor price
AttributionCookie, last click, 30–90 day windowSingle-use signed token, 15-minute expiry, two signatures
Who writes the creativeThe affiliateNobody — templated card
DiscountingUsually expected, often requiredNever involved
Who the traffic isCoupon sites, reviewers, YouTubers, communitiesUsers of other desktop apps, at install time
Brand riskReal — affiliates say what they likeStructurally low — no partner copy exists
Common fraudCookie stuffing, last-click hijack, brand biddingInstall farms, emulators, token replay
You can pay inMoney onlyMoney, or attention you already have

The cost nobody puts in the spreadsheet

A 30 percent commission on a EUR 49 licence is EUR 14.70, which looks competitive against a cross-promotion CPI in the same range.

But for desktop software, most affiliate volume arrives through coupon and deal sites. That means a discount on top of the commission, so the real figure is closer to EUR 14.70 plus 20 to 40 percent off the price. And the discount does not end when the campaign does. Once "[your app] coupon" is a search result with answers, your effective price is the discounted one permanently.

The commission is the visible cost. The price anchor is the one that compounds.

Attribution: cookies versus tokens

Affiliate attribution is a cookie, and cookies have three well-known problems: they expire, they are blocked, and they can be written by anyone who can get a browser to load a URL.

That last one is why the most common affiliate fraud is not fake sales but stolen credit — a browser extension injecting an affiliate cookie at checkout on a purchase the customer was already making, or an affiliate bidding on your own brand name to intercept traffic that was already yours. The conversion is completely genuine. Only the attribution is theft, which makes it much harder to detect than a fake install.

GHAN's attribution is a single-use token that expires in 15 minutes and must be countersigned by the receiving application's own key. That does not make GHAN better than an affiliate program — it makes the fraud different. Ours is manufactured volume, and it is caught with retention, device signals and cohort shape. Theirs is stolen credit, and it is caught by auditing your own affiliates, which most programs do not do.

When an affiliate program is the right answer

  • You have real margin and can absorb a commission plus a discount.
  • Your buyers research before purchasing, so reviewers and comparison sites genuinely influence them.
  • You have a recognisable product that creators want to talk about.
  • You can afford to police brand bidding and coupon extensions, which is the actual work of running one.

When cross-promotion is the right answer

  • Your price point cannot support a commission plus a discount.
  • You would rather pay in attention than in cash.
  • You do not want anyone else writing copy about your product.
  • You want a published floor price rather than a percentage of a sale you have not made yet.

Running both

There is no conflict. Affiliates reach purchase intent; cross-promotion reaches people who are installing desktop software right now and were not looking for you.

Measure them on the same basis, which for desktop software means cost per retained customer — affiliate commission plus the discount given, divided by customers still active at your retention horizon, against the rate card, which is already a retained price. Comparing a commission percentage to a CPI directly will mislead you in whichever direction your instinct already leans.

Questions people ask about this

Are affiliate programs good for desktop software?

They work, and they cost more than the commission rate suggests. A 30 percent commission on a EUR 49 licence is EUR 14.70, which is comparable to a cross-promotion CPI - but most affiliate traffic for desktop software comes through coupon and deal sites, which means a discount on top of the commission and, more importantly, a permanent change to what your product appears to be worth. The commission is the visible cost. The price anchor is the real one.

How is cross-promotion different from an affiliate program?

An affiliate is paid on a sale; a cross-promotion network is paid on a retained install. That single difference cascades. Affiliate attribution is usually a cookie, which is fragile and contestable; cross-promotion here uses a signed single-use token. Affiliates choose their own creative and messaging, which is why brand-safety complaints dominate affiliate management; cross-promotion cards are templated. And affiliates are usually incentivised to discount, while a fixed CPI is indifferent to your pricing.

Which has worse fraud?

Different fraud, not obviously more or less. Affiliate fraud is mostly attribution theft - cookie stuffing, last-click hijacking, brand bidding on your own trademark, and coupon extensions that inject an affiliate cookie at checkout on a sale you had already made. Install-network fraud is mostly manufactured volume - install farms, emulators, token replay. Attribution theft is harder to detect because the conversion is genuine; only the credit is stolen.

Can I run both?

Yes, and plenty of apps should. They reach different people - affiliates reach purchase intent, cross-promotion reaches people installing desktop software right now - and neither requires exclusivity. Just measure them separately and on the same basis, which for most desktop apps means cost per retained customer rather than cost per conversion.

Machine-readable versions of this page: markdown · llms.txt · llms-full.txt · OpenAPI · AI catalog