How Broken URLs Bias Your Marketing Attribution

Lucas: So here's something that keeps me up at night - and I think it should keep any marketing analyst up at night. We talk a lot about attribution models, incrementality, view-through windows. But there's a much more basic problem that quietly distorts every single number in your dashboard: broken URLs. Luna: Like, links that just 404? That seems almost too simple to be a big deal. Lucas: It sounds simple, but the scale is shocking. I was looking at a mid-size retail campaign from late 2025 - about 2 million dollars in ad spend across paid social, search, and email. They ran a full attribution audit and discovered that roughly 12 percent of their tracked conversions came through URLs that had broken at some point in the customer journey. Luna: 12 percent? That's massive. How do you even lose a conversion to a dead link? Lucas: Let me walk through how it happens. Say you launch a paid social campaign with a utm tagged link pointing to a landing page. The page performs well, so the product team redesigns the site and moves that page to a new URL. They set up a redirect - usually a 301 - but they forget to preserve the UTM parameters. So anyone who clicks that original ad now lands on the new page, but all the attribution data is stripped. The conversion gets credited to 'direct traffic' or, worse, to the wrong channel. Luna: So the attribution model sees the conversion but assigns it to whatever the last click was - which might be a branded search or an email open - and the paid social campaign gets zero credit. Lucas: Exactly. And that's the best-case scenario where the page still loads. In the retail campaign I mentioned, nearly 4 percent of all tracked clicks hit a full 404 error. The user bounced, maybe searched for the brand, came back through organic, and converted. Last-touch attribution gave the credit to organic search. The paid campaign looked like it underperformed, so the marketing director cut its budget the next quarter. Luna: That's a direct, measurable budget decision based on bad data. And the team probably never knew the root cause was a broken link. Lucas: They didn't. And that's the scary part. A lot of attribution bias gets blamed on model choice - last touch versus multi-touch versus algorithmic. But the underlying data quality is often the real culprit. If your click logs are full of broken redirect chains or missing UTM parameters, the most sophisticated model in the world will give you garbage output. Luna: So what's the fix? Do you just crawl your own site every week? Lucas: Pretty much, yeah. There are tools - Screaming Frog, DeepCrawl, even simple Python scripts - that can audit every URL your ads have pointed to in the last 90 days. You check for four things: the HTTP status code should be 200, the redirect should be a 301 not a 302, the redirect should preserve UTM parameters, and the final destination should load within two seconds. Luna: And how often do teams actually do that? My guess is almost never. Lucas: In my experience, maybe one in twenty marketing teams has a regular URL audit. Most assume that if the link worked on launch day, it still works. But e-commerce sites change constantly - products go out of stock, pages get consolidated, campaigns end and landing pages get pulled. A link that worked in September can be dead by November. Luna: Let's talk about the email channel specifically. I've seen email campaigns where the same link is used across multiple sends, and after a few months, the destination page has changed entirely. Lucas: Right, and that's a classic source of attribution drift. A customer clicks a link from an email in January, but the page they land on in March is a different product category. The email attribution still gets credit, but the actual experience is misaligned. Over time, your attribution data starts to reflect a journey that never really happened. Luna: I've also seen cases where the redirect chain itself introduces latency. A 301 that goes to a 302 that goes to a 200 - each hop adds time, and if the page takes more than three seconds to load, you lose a chunk of mobile traffic. Lucas: That's another hidden cost. Google's own data shows that as page load time goes from one second to three seconds, the probability of bounce increases by 32 percent. So a broken redirect chain doesn't just distort attribution - it destroys conversions outright. And the marketing team blames the ad creative or the audience targeting, not the technical infrastructure. Luna: So what should a marketing analyst do tomorrow morning? What's the first step? Lucas: Export all the destination URLs from your ad platforms for the last 30 days. Paste them into a free HTTP status checker - there are browser extensions and online tools. See how many return anything other than a 200. I'd bet good money that at least 5 percent have some issue. Then fix them and measure what happens to your reported ROAS over the next two weeks. Luna: That's a low-effort, high-impact test. And it's completely independent of whatever attribution model you use. Lucas: Exactly. You could be using last touch, first touch, or a custom algorithmic model - if your underlying click data is tainted by broken URLs, every output is suspect. Cleaning up link rot is probably the single cheapest optimization available to any marketing team spending more than a few thousand dollars a month. Luna: It's one of those fixes that feels too boring to be important. But the numbers don't lie. Lucas: Speaking of things that are boring but important - and I mean this genuinely - if these marketing conversations have sparked something you've actually used in your own work, buying us a coffee makes a real difference. A couple of dollars a month is genuinely what keeps these going. That's buy me a coffee dot com slash fexingo. Luna: Yeah, it's a small thing that adds up. And it keeps the show ad-free, which I love. Lucas: Alright, back to the data. There's another layer to this problem that's even more insidious: the link looks fine to a human, but the tracking parameters are corrupted in a way that the ad server doesn't flag. Luna: Can you give me an example? Lucas: Sure. Some ad platforms append their own click identifiers to the URL - like '?fbclid=' for Facebook or '?gclid=' for Google. If your landing page is set up to only read certain UTM parameters, the platform's appended ID might get dropped during a redirect. So the conversion event fires, but the platform can't match it back to the original ad because the click ID is gone. Luna: So the platform reports fewer conversions than actually happened, and the advertiser thinks the campaign underperformed. Lucas: Exactly. I've seen campaigns where the discrepancy was over 20 percent between what the ad platform reported and what the internal analytics showed. The marketing team blamed the attribution model, but the real issue was that the click IDs were being stripped by a misconfigured redirect on the landing page. Luna: So the solution there is to test the full redirect chain - not just the final page, but every intermediate hop - with the exact UTM parameters and platform click IDs that your ads will send. Lucas: Bingo. And you need to do it on every device type and browser. Sometimes the redirect behaves differently on mobile than on desktop. One retailer I worked with found that their mobile redirect dropped the UTM parameters only on iOS Safari. That single issue was costing them about 8 percent of their iOS paid search conversions. Luna: That's the kind of thing that would never show up in a standard QA check. Most teams just click the link on their desktop once and call it done. Lucas: Right. And the bigger your campaign, the more complex your redirect infrastructure becomes - multiple landing pages, A/B test variants, personalized URLs. Each one is a potential point of failure. I'd argue that link rot is one of the largest uncounted sources of ad waste in digital marketing today. Luna: And it's completely fixable. It's not like trying to model incrementality or measure view-through - it's a straightforward technical audit. Lucas: Absolutely. The irony is that teams spend thousands on attribution software and consultants, but they won't spend an hour a month checking their own links. If you take one thing from this episode, it should be: go check your URLs. You might find an immediate 10 percent improvement in your reported performance. Luna: And if you do find something, let us know. We'd love to hear real-world examples - maybe we'll do a follow-up episode on the most common link rot patterns. Lucas: That's a great idea. I've got a feeling we'll get plenty of stories. Until next time.