I built a voucher portal. The wild thing isn’t the codes or the affiliate plumbing — it’s what’s not on the page.
If you open the network tab on a typical coupon site and scroll for ten seconds, the count of third-party requests blows past anything else you’ve visited that day. Tag managers loading more tag managers. Identity providers stitching cookies across domains. Behavioral pixels firing on every hover. Browser fingerprinting modules whose only job is to recognize you when you delete cookies. All of it dressed up as “personalization.”
It made me wonder: does any of that actually help me get a working coupon?
The honest answer is no
Voucher discovery is one of the most intent-driven actions on the open web. You don’t need a profile of me to know what I want — I literally typed the brand name into the URL bar. The ML magic that surfaces “the most exciting deals first” is almost always optimizing for the publisher’s revenue, not the visitor’s savings.
So I designed a portal that does the boring thing on purpose: no analytics, no marketing cookies, no behavioral profiles, no algorithmic personalization. Same visitor, same homepage, every time. The only ranking signals are editorial: trust score, recency, time to expiry. Things a human can verify by hand.
The site is gutscheinhub.de — a German voucher hub, in production since this month. You can verify the privacy claims in your terminal:
$ curl -sI https://gutscheinhub.de | grep -i cookie
# (nothing)No Set-Cookie headers, no tracking scripts in the HTML, no consent banner because there’s nothing to consent to. The architecture cannot build a profile of you, because it doesn’t persist anything that would let it.
Privacy by architecture, not by policy
“We respect your privacy” is the most over-promised phrase on the web. Every privacy policy starts with it; almost every site behind those policies still ships dozens of trackers. The phrase has been laundered into meaninglessness.
The version I care about is privacy by architecture: data that is never collected cannot be leaked, sold, or weaponized — not today, not in five years when the regulatory wind shifts, not in ten when the system is acquired by someone with worse incentives. A no-tracker portal is not a polite request. It is a structural constraint.
This is the same principle behind a small browser extension I maintain that strips tracking parameters from links before they hit the clipboard. And it’s the throughline of my recent academic work on provenance and verification — different surface, same root question: can you trust what you’re looking at, and can the system prove its claims to itself?
What about the affiliate part?
This is where most “privacy-first” projects get evasive. A voucher portal has to monetize somehow, and the realistic option is affiliate revenue: when a visitor activates a code and buys something, the merchant pays a small commission via a tracking network.
That involves a cookie — but it is set by the merchant’s network, after the visitor’s deliberate click on a code, and only for the purpose of attributing that one purchase. There is no behavioral build-up, no cross-site profile, no pre-emptive identifier. The visitor pays exactly the same price either way. It is the difference between a referral and surveillance.
The honest framing is: the site itself collects nothing, the affiliate network handles one purchase attribution after an explicit click, and we say so on the page. That’s the entire model. There are no “third-party partners,” no “analytics for service improvement,” no “personalized recommendations.” Anything more would be a betrayal of the point of the project.
Editorial, not algorithmic
The other half of the bet is the ranking. Big voucher sites lean on opaque ML to surface “the deal you’re most likely to redeem.” That sounds visitor-friendly until you remember the optimization target is revenue per session, not value per purchase. The two diverge constantly.
The portal’s ranking is three signals, none of which depend on knowing anything about you:
- Trust score — assigned by hand when a code is verified, decays if the code starts breaking.
- Recency — newer codes float, because they have the longest remaining shelf life.
- Time to expiry — codes that are about to die get pushed down, so the homepage stays useful tomorrow.
You and I see the same order. That’s a feature, not a limitation. It also makes the system auditable: I can show anyone exactly why a code is in position one. With a personalization model, that explanation collapses into “the model thought you’d like it.”
Verified, not crowdsourced
The third design choice: codes are checked before they are published, and re-checked on a schedule. When a code stops working, it is removed automatically, not voted down by users who happen to be online. Community voting feels democratic; in practice it amplifies whichever segment of the audience is most active, and rewards spam tactics. Editorial verification is slower, but it is the only way to make “we checked this” mean anything.
What I’m watching for
A few things will be honest tests of whether the “no tracking” stance holds up:
- Conversion. Conventional wisdom says you cannot run a content site without analytics. Server logs and post-hoc affiliate reports give a lot of what dashboards give, without per-visitor data. We will see.
- Indexing. Search engines mostly use crawlers, not on-page beacons, so there should be no penalty. Submission to the major search consoles is already done; the next month will tell.
- Newsletter, if it happens. Open and click pixels are tracking. A newsletter without them is technically straightforward and almost never offered. That seems worth doing properly if at all.
Why this matters beyond coupons
Voucher portals are a tiny corner of the web, but they are an interesting one because the standard playbook is so aggressive. If the playbook can be replaced here — by an editorial site that collects nothing and still works — it is replaceable in a lot of other places too. Most of the “we need tracking to function” claims, when examined closely, turn out to be “we built a business model that needs tracking to function.” Those are different problems with different fixes.
The portal will keep growing — more brands, more categories, more editorial guides. I will write more once enough data exists to say something honest about traffic and conversion under these constraints. For now: take a look at gutscheinhub.de, run curl -I on it, and judge for yourself.
