The Three Losers

A note on what's actually broken in affiliate marketing, and why I think the fix is a category, not a tool.

---

There's a market of more than twenty billion dollars a year where everyone loses.

Publishers lose commissions. Webshops lose customers. Consumers lose trust. The money that should flow through the system instead leaks out of it, quietly, all day, every day. Nobody is responsible. Nobody is even watching. The losses don't show up on a dashboard because there is no dashboard for them. They show up as flat months, dead pages, returning customers who don't return.

I've spent the last few years close enough to this to see it from all three angles. What I keep coming back to is that this isn't a tooling problem. It's an infrastructure failure. And the reason nobody has fixed it is that nobody owns it.

This piece is the long version of what I think the fix looks like. The shorter version is at the end.

This is an infrastructure failure, not a missing tool

When a system breaks and nobody is accountable, the failure mode is silence.

Affiliate marketing runs on a chain of redirects across networks that don't talk to each other. A publisher writes a review, drops in an affiliate link, and from that moment on the link is a dependency on infrastructure they don't control. The merchant can change the URL. The product can be discontinued. The network can strip tracking parameters during a redirect. The link still resolves to something, so it looks fine in a casual check. The commission just stops registering.

None of these failures throw an error. None of them fire an email. The link doesn't turn red. It turns invisible. It still gets clicked, sometimes thousands of times a month, and the publisher quietly stops earning on it.

I've written about why the affiliate networks themselves don't surface this. The short version is that the incentive structure makes silence cheaper than honesty. The networks aren't being malicious. They're optimizing for what they're measured on, and link health isn't on the list. So the layer that should have been monitored from day one became the only layer in modern web infrastructure that nobody monitors.

Compare this to anything else that the web runs on. SSL certificates expire and you get pinged six weeks in advance. DNS records change and a dozen tools yell at you. Server response times degrade and your dashboards turn yellow. Affiliate links rot and the only feedback is a shrinking line on a P&L statement three months later, which you'll probably attribute to seasonality.

The world changed in 2026 in ways that make this worse, not better: cookie deprecation pushed more value onto first-click attribution, AI-generated content multiplied the surface area of affiliate links by an order of magnitude, and network consolidation means failures cascade further when they happen. The tooling didn't keep up. It barely kept up with 2018.

This is the gap. Not "there's no scanner." The gap is that an entire category of infrastructure is missing.

The three losers, one at a time

Here's the part that turned this from a personal annoyance into something I think matters.

The publisher loses the most visibly. Every broken affiliate link is a direct hole in the bucket. Industry estimates put the loss at one to two percent of total commissions across the affiliate ecosystem. On a twenty-billion-dollar market, that's two to four hundred million dollars a year leaking out, conservatively. For an individual publisher running a six-figure affiliate site, it tends to land somewhere between three and eight percent of revenue, because rot concentrates on older content that still gets traffic.

The publisher's loss is the easy story to tell. It's the one I told in The Quiet Leak two weeks ago. But it's only one third of the picture, and arguably not the largest third.

The webshop loses next, and they don't even know it. When a publisher's affiliate link points at a discontinued product, the visitor doesn't see "404, sorry." They see a polished merchant page that says "this product is no longer available, here are some alternatives" or, worse, gets dumped onto the homepage. From the merchant's analytics, this looks like a normal visit with a high bounce rate. From the visitor's perspective, this looks like a broken promise from the publisher they trusted.

The merchant just paid for that traffic. Most affiliate programs pay per click contribution to a sale, but the merchant is also paying in customer acquisition cost across every other channel that brought a visitor to the publisher's review in the first place. When the publisher's link sends a high-intent buyer to a dead end, the merchant doesn't get a sale, doesn't get a chance to retarget, and often doesn't get the visitor back. They lose the most expensive kind of traffic, the kind that was already convinced.

The merchant has no way to detect this from their own systems. It happens upstream of their analytics. They'd need to monitor every publisher who links to them, which they don't, and shouldn't have to.

The consumer loses too, and this one is the slow corrosive damage. A reader trusts a review site. They click a link to buy the recommended product. The product is gone, or the page is broken, or the link redirects to something else. They don't get the thing they came for. Sometimes they buy a substitute they didn't research. Sometimes they buy nothing. Sometimes they go back and read the review more skeptically, then notice that the publication date was three years ago, then close the tab and never come back.

The compounding effect is a slow erosion of trust in the entire review-and-recommendation ecosystem. People stop believing that affiliate sites are giving them current information. That distrust isn't fair to the publishers. Most of them aren't being lazy; they're losing the visibility war against link rot. But the consumer doesn't know that. From the outside, the system looks unreliable.

Three parties, three different shapes of loss. None of them have an incentive or a tool to fix it on their own. The publisher could in theory audit links manually, but it's a job that takes days, has to be repeated forever, and feels like maintenance work nobody enjoys. The merchant could in theory monitor inbound traffic for dead-end patterns, but they don't have visibility into the publisher's content. The consumer can do nothing.

This is the textbook profile of an infrastructure problem: it harms everyone in the system, but no single party has a clean reason to own the fix.

What I built, and why this shape

I built LinkPulse because the missing infrastructure won't appear by itself. It's a WordPress plugin that detects affiliate links across sixteen networks, follows the redirect chains, surfaces what's broken, and quantifies the revenue impact. It's free for the first five hundred links, which covers most small and mid-sized affiliate sites entirely.

The interesting part is not that I built it. Several scanners exist. The interesting part is the four design decisions that I think define the category, and that I'd defend against the obvious alternatives.

One source of truth, not six dashboards. Every serious affiliate publisher works across multiple networks. The default state is six tabs open across six interfaces, each showing a slice of the same underlying reality. This is not work. It's tab-switching dressed up as work. LinkPulse pulls from WeCanTrack, which already aggregates across networks, and presents a single coherent view. The reason isn't that one dashboard is prettier than six. The reason is that attention is the most expensive resource in this work, and any system that demands six contexts to answer one question is a system that ensures the question stops getting asked.

Delegate to agents, don't automate with scripts. This is the design choice I expect to be most contested. Affiliate link work is judgement work. A dead link sometimes wants a substitute product, sometimes wants a removed paragraph, sometimes wants a redirect to a category page, sometimes wants a note to the editor that the whole review needs an update. A cron job can't make those calls. An agent with context can. LinkPulse exposes its scanning and link-replacement tools through an MCP server so that a real model (Claude in my case, but the protocol is open) can sit on top of the data and act on it the way an experienced editor would. You ask once a week. The agent checks, proposes, you approve. Twenty minutes of work that used to take three days.

No revenue cut. The dominant business models in this space (Skimlinks, Sovrn, Viglink before it folded) charge twenty to twenty-five percent of commissions in exchange for managing the affiliate relationships. That model made sense in 2012 when those relationships were genuinely painful to manage. It does not make sense now, because most serious publishers already have direct relationships with their networks. Charging a quarter of revenue forever to sit between the publisher and their existing partners is a parasite model. I think it's also the main reason the category has been frozen for a decade. When the dominant players have a structural reason to keep things complicated, simpler tools never get built. LinkPulse charges a flat subscription for the work the software does. No cut, no insertion into your link flow, no equity in your commissions.

Agent-native first, dashboard second. Most of the visible affiliate tools assume the human will sit inside a dashboard for hours. I think that's the wrong default in 2026. The dashboard is for occasional review. The day-to-day work happens where you already are: in a chat, in your editor, in your existing workflow. LinkPulse ships an MCP server before it ships dashboard polish, which is unusual sequencing, but it reflects the bet that AI-native workflows will replace dashboard workflows for this kind of operational maintenance work. The full tech-stack argument for this is here.

These four decisions sound like product choices. They're really claims about what kind of category this is.

Why this is a category, not a feature

The hardest part of writing this is resisting the urge to make it sound bigger than it is. So let me be precise.

I don't think LinkPulse is going to "disrupt the affiliate industry." That phrase is meaningless. What I think is happening is narrower and more specific: the layer of infrastructure that monitors and maintains affiliate link health is going to become its own category, the way observability became its own category once people stopped pretending logging was enough.

That category has a shape. It's monitoring infrastructure operated by agents on behalf of publishers, sitting on top of unified data, charging for software work rather than taking a cut of the underlying revenue. Link rot is the headline problem, but the same infrastructure layer eventually handles attribution drift, network consolidation impact, commission rate changes, and a dozen other operational concerns that currently don't have a home anywhere.

When I say I want LinkPulse to be a category leader, this is what I mean. Not biggest. Not most-funded. The thing that defined the shape of the category early enough that everyone else who shows up has to position relative to it.

What this means for your work on Monday

Even if you never use what I built, three things change once you start seeing this clearly.

First, stop accepting that link rot is invisible. It isn't. It just hasn't been worth surfacing for the players who could have surfaced it. Audit your own site once. Whatever you find will be worse than you expected, and that's the point.

Second, stop looking at your affiliate dashboards as separate worlds. They're slices of one underlying reality. If your tooling forces you to context-switch across six tabs to answer one question, your tooling is the problem.

Third, when you find yourself building automation, ask whether the work actually wants automation or whether it wants delegation. Automation is for rules. Delegation is for judgement. Most of the maintenance work that consumes a publisher's time is judgement work. It wants an agent, not a cron job.

The shift, when you see it, is small and obvious. Which is usually how shifts feel right before they happen.

---

This is LinkPulse. Free for the first 500 links, no account needed. linkpulse.dev.