Some reporting failures arrive with sirens. This one arrived as a number that was merely a little too large to ignore. On the Let’s Talk Marketing podcast, revenue operations leader Mira Wise told Sophie Steffen the story of a direct-traffic spike that looked like a tidy privacy story and turned out to be a tracking bug hiding behind one. It is a useful case study in resisting the comfortable explanation.
The Convenient Explanation Was Wrong
The setup was a fresh HubSpot launch. HubSpot drops a tracking pixel on your website, the pixel sets a cookie, and the cookie is where all the source properties that tell you where a visitor came from get defined. Standard machinery. Then the team looked at the data and saw a really big portion of traffic sitting in the direct bucket.
Now, rising direct traffic has a ready-made alibi in 2026. Cookie blockers, consent rejections, privacy-conscious browsers. Everyone in marketing has learned to expect a slice of traffic that simply refuses to identify itself. The path of least resistance was to nod, blame the general state of the internet, and move on.
Wise declined the alibi. The portion was big enough that she wanted to know whether something was actually broken or whether this was just the new normal of users declining consent. That distinction matters, because one of those answers requires a fix and the other requires acceptance, and confusing them wastes everyone’s time.
“We need to dive into, is something going wrong here or is this just that now that we have all of this in place, a lot of our users aren’t even giving us the consent to be tracking them.”
A Bug Living Between the Subdomains
The investigation found the culprit, and it was structural. The company ran a sprawl of domains and subdomains: a .com, a .es, a .fr, plus separate subdomains for the app itself and for the signup form that creates an app account. Cross-domain tracking is what stitches a single visitor’s journey together as they move across that estate. It was broken.
The consequence was quietly destructive. A visitor who started on the .com and crossed to the .nl was being logged as a brand-new session with a direct source, rather than the continuing journey they actually were. Every one of those crossings erased the original campaign that brought the person in. The marketing team was not looking at privacy noise. It was looking at its own campaign attribution leaking out through a seam in the website.
This is where the story stops being a tidy technical anecdote and becomes an organizational one. A marketing team owns its website content and can fix it directly. It does not own the underlying website code. Product and engineering do, and that team works in tight sprints toward its own goals. Wise could not simply patch the problem. She had to get it booked into someone else’s roadmap, and signups, a top-of-funnel metric leadership watches closely, were being mismeasured the entire time. That combination is what turned a bug into a scramble.
How the Scramble Became a Process
Two things resolved the situation, and only one of them was code. The technical half was straightforward enough: work with product and engineering to pass visitor information correctly between every subdomain. The more instructive half was about communication.
Rather than handing leadership a single frightening figure, the team did the analytical work to break the direct-traffic number into three honest categories. Some of it really was direct. Some of it was the cross-domain bug, identified and under active repair. And some of it was traffic that consent rules made impossible to track, a portion no fix would ever recover.
“It’s 25%, not a full, you know, 100% of what we thought it was, that also helps bring the trust back, which I think you can lose in situations like this.”
The breakdown did the heavy lifting. When leaders can see precisely what went wrong, what is being fixed, and what was never actually broken, a crisis shrinks to its real dimensions. Wise also reframed the episode as evidence of progress rather than failure. Before the HubSpot rollout, the team could not have detected this problem at all. Now it could spot it and quantify it within two days instead of two months, which is the entire point of building a serious data infrastructure. The scramble was uncomfortable, but it left the team measurably more honest about its own numbers than it had been before.
For the full interview breakdown, see our complete Expert Insight with Mira Wise.
Tools Mentioned in the Interview
The following tools and platforms were referenced during this conversation.


