Case study: how Northwind Reviews recovered 32% of lost affiliate revenue in six weeks

Note: Northwind Reviews is an illustrative example built from patterns across several real Tracklume customers. Numbers are representative, not from one specific publisher.

Northwind Reviews is a six-page-a-week review site in the home-office niche. Two writers, one editor, no engineers. By the time they showed up in our inbox they had about 1,800 affiliate links across 380 posts, and a quietly collapsing affiliate cheque.

The owner, who I'll call Priya, had noticed three things in the same month:

• Conversion rate from clicks to sales had dropped roughly a third year-on-year.
• About 9% of their links were returning 404s or redirecting to the merchant homepage.
• A growing share of European traffic was hitting the US storefront, abandoning, and never coming back.

None of that is dramatic on its own. Added up, it was costing them about 32% of what the same traffic used to earn.

Week one: stop the bleeding

Before changing anything strategic, we ran the broken-link probe across all 1,800 links. 164 were dead. Of those, 71 were on posts that still pulled meaningful organic traffic — those were the priority.

The boring fix: replace each dead URL with the current product URL, redirect through a branded /go/ short link, and set a fallback destination on the offer so that if the merchant ever dropped the SKU again, the link would route to the category page instead of a 404. Two evenings of work for the editor.

Recovered revenue in week one: about 11% of the total uplift. Not because the fixes were clever — because the bleeding had been quiet for months.

Week two and three: geo routing

The single biggest leak was geo. Northwind wrote in English, ranked globally, and pointed every link at amazon.com. About 38% of their clicks were coming from outside the US, and the conversion rate on those clicks was about a fifth of US conversion.

We set up three geo rules per major merchant:

• US → amazon.com with the US affiliate tag
• UK + IE → amazon.co.uk with the UK tag
• DE + AT + CH → amazon.de with the DE tag
• everything else → the original US link, with a tiny disclosure tweak

This is unsexy work. There is no clever model. The win is that you stop wasting a third of your clicks on a storefront the visitor can't actually buy from.

By the end of week three, EPC on European traffic had roughly tripled. Total EPC across the site was up about 18% on the previous month — and we hadn't touched a single piece of copy.

Week four: rules for device and time

Two smaller leaks were left. Mobile users on certain merchants were hitting a desktop-optimised landing page that didn't render well; we set a device rule to route mobile traffic to the mobile-friendly URL. And a particular tool merchant ran a flash sale every other Tuesday that diverted traffic to a campaign page; we set a time-windowed rule on that offer so that during the sale window, links pointed at the campaign page instead of the regular product.

Both are exactly the kind of thing you'd do in a hand-built CMS if you had the engineering bandwidth. The point of routing rules is that an editor can ship them in five minutes without touching the post.

Week five and six: hygiene that stays fixed

The last fortnight was the part most teams skip and then have to redo a year later.

• Every offer got a default fallback destination so future merchant changes can't 404.
• Every link got a clear disclosure label so the FTC checks pass automatically.
• Broken-link scans now run nightly. Priya gets one email a week with the diff, not a live firehose.
• Click analytics by country and device are reviewed monthly, not whenever something feels wrong.

That last part is the difference between "we fixed it" and "it stays fixed".

The numbers

Six weeks in, Priya's monthly affiliate revenue was 32% higher than the month before they started, on essentially flat traffic. Three drivers:

• Geo routing: ~18% of the uplift
• Broken-link recovery: ~11%
• Device + time rules: ~3%

There was no traffic gain. No new posts. No new merchants. The same content, routed properly.

What I'd do differently

If we ran this again I'd do the broken-link probe and the geo rules in week one, in parallel. The week-two reveal of the geo problem cost them another two weeks of underperforming European clicks they could have kept.

I'd also start the nightly scan on day one, before fixing anything. It gives you a baseline you can show the owner when the dust settles.

What to copy

Three things, in order of payoff per hour of editor time:

1. Probe every affiliate link for status codes. Fix the 404s on posts with real traffic first.
2. Add geo rules to the merchants where you have meaningful non-home-country traffic. Start with Amazon if you use it.
3. Put a fallback destination on every offer so the next merchant change is invisible to your readers.

None of this is novel. The reason it works is the same reason most teams skip it: it's repetitive, it doesn't feel like growth, and it pays for itself the day after you finish.

That's the whole case study.