Where your vet website quietly loses the patients your ads paid for.

There's a specific kind of failure that almost never shows up as a failure. Your ads run. Your dashboard fills with numbers. Clicks come in, the budget spends down, and on paper things look like they're working. Meanwhile, somewhere between the click you paid for and the patient who actually booked, the connection quietly snapped — and you have no way of knowing it did, because a broken connection doesn't throw an error. It just shows you a smaller number than the truth, or attributes the booking to the wrong place, and lets you carry on.

This matters more than it sounds like it should, and here's why. Google's bidding doesn't just report your conversions — it learns from them. Every booking it can see, it uses to find more people like that booker. Every booking it can't see is a lesson it never gets. So a tracking leak isn't only a reporting problem you could shrug off. It actively makes your ads dumber, because you're paying Google to optimize toward a picture of success that's missing pieces. The campaign chases what it can measure, and quietly starves the parts it can't.

I've written before about why most vet practices can't prove their Google Ads work — that conversion tracking is the steering wheel, not the rear-view mirror. This is the companion piece: the specific, physical places where the steering linkage comes loose. Think of it as a walk through your own site, following a worried pet owner from the moment they click your ad to the moment they book — and noting every spot where, if you're not careful, they vanish from your data.

The click is the only part you can see for free

Start with what's solid. When someone clicks your ad, Google knows it. That part is reliable and free. The trouble is that a click is not a patient. A click is a person who has arrived at your front door — nothing more. Everything that turns that arrival into a booked appointment happens after the click, on your turf, and that's exactly the stretch where the measurement is yours to keep or lose.

So the question that actually decides whether your ad budget is working isn't "did they click." It's "did they do the thing" — call, book, submit — and did anything record that they did. Almost every practice I've looked at can answer the first question and not the second. Here's where the second question goes to die.

Leak one: the booking button that sends them somewhere else

This is the most common and most expensive leak, and it's almost universal in veterinary practice, because of how vet booking software works.

You've got a "Book Appointment" button on your site. It's the single most important thing on the page — the whole point of the ad was to produce that click. But when the visitor presses it, where do they go? For most practices using a scheduling tool, the answer is: a different website. The booking happens on the booking provider's domain, not yours. The visitor leaves your site entirely to do the one thing you most needed them to do.

Here's why that's a problem, stated precisely, because the nuance is where people get it wrong. When a visitor crosses from your domain to the booking provider's domain, your analytics loses sight of them — they've walked off your property. Two things break at once. First, attribution: when the booking provider sends the visitor back (or when the visitor returns later), your analytics often counts the booking domain as the traffic source, so the booking gets credited to the booking tool instead of to the Google ad that actually drove it. Second, and worse, the conversion itself: the booking happened on a site you don't control, so unless you've done something specific about it, nothing on your side ever recorded that a booking occurred at all.

There's a half-fix people reach for that I want to warn you about, because it's the source of a lot of false confidence. You can add the booking domain to an "unwanted referrals" list, which stops it from stealing the traffic-source credit. That's worth doing — but it only fixes the attribution half. It does nothing for the harder half: actually seeing that the booking happened. To capture the conversion itself, you need cross-domain measurement — and that's only possible if you can install your own tracking on the booking provider's domain. With most vet scheduling tools, you can't. So the booking, the actual revenue event, stays invisible no matter how clean your referral list is.

The practical upshot: if your booking lives on someone else's domain and you've never deliberately solved for it, you are almost certainly under-counting your real bookings — which means Google is optimizing toward a fraction of your true conversions, and your best-performing campaigns might look mediocre simply because their payoff is happening where your data can't watch.

Leak two: the phone call

For a veterinary practice, this is the leak that hurts most, because for a huge share of pet owners — especially the anxious ones, the after-hours ones, the my-dog-just-ate-something ones — the conversion is a phone call. They don't fill out a form. They find your number and they call. That call is the patient. And by default, Google has no idea it happened.

A phone number sitting on your website is just text. When someone clicks it on their phone, or reads it and dials, there is no signal that travels back to Google Ads saying "the ad worked." The call connects, you book the patient, everyone's happy — and in your ad data, that click looks like it went nowhere. You're effectively telling Google "this search term produced nothing," when in fact it produced your best lead of the day.

Closing this one means call tracking: Google can swap the number on your site for a forwarding number that records which ad drove the call, and counts a call as a conversion if it lasts past a length you set (so the thirty-second wrong numbers don't pollute the data). It's genuinely worth doing for a phone-driven business. One current caveat worth knowing, because the landscape just shifted: Google is retiring the old "call-only ads" format — as of early 2026 you can't create new ones, and the existing ones stop running in 2027 — so the modern way to capture call intent is phone numbers added as call assets on your regular search ads, not the standalone call ad. If a guide tells you to build a call-only ad, it's working from an old playbook.

Leak three: the form that quietly posts somewhere else

The third leak is subtler because the form looks like it's on your site. The visitor fills it in, hits submit, sees a thank-you — everything feels native. But where the form actually sends the data determines whether you ever see the conversion.

Two common versions. First, the embedded third-party form: a contact or intake form from some other platform, dropped into your page in a frame, that submits on the other platform's domain. To your tracking, the submit is happening somewhere it can't watch — same cross-domain problem as the booking button, wearing a costume. Second, the "contact us" that's really just an email link: tap it and the visitor's mail app opens. There's no submission event at all, nothing fires, and that lead — a real person who wanted to reach you — leaves no trace in your data.

The fix here is the most controllable of the bunch, because the form is on your page: a conversion event should fire on the actual successful submission, and the cleanest setup routes the thank-you confirmation through a page or signal your tracking can see. The point is simply that "the form works" and "the form is measured" are two different claims, and most practices have only verified the first.

Leak four: the one that opened this month

This last one is new — new enough that most material written even a few months ago doesn't mention it, which is exactly why it's worth flagging.

As of mid-June 2026, Google changed how a visitor's privacy choices govern whether a conversion can be linked back to an ad. In plain terms: if a visitor declines advertising-data consent in your cookie banner and then goes on to book or call, Google Ads can no longer connect that conversion to the click that produced it. The booking still happens; the link back to the ad is simply gone. Before this change, a second signal could sometimes keep that connection alive; now a single consent control decides it, and a "no" closes the door.

I'm not going to pretend to hand you a tidy fix in a blog post, because this one genuinely depends on how your consent banner is configured and is the kind of thing that deserves being set up correctly rather than guessed at. But you should at least know it exists, because here's how it shows up if you don't: starting around now, some of your real conversions will quietly stop being attributed, your numbers will dip for no reason you can see in your campaigns, and the instinct will be to blame the ads. The ads may be fine. The trail just got cut by a consent setting. Knowing the leak exists is the difference between diagnosing it and chasing a ghost.

A note on how fast the ground moves

One more thing, less a leak than a sign of the weather. If you ran your site through Google's PageSpeed Insights recently, you may have noticed a new section appear next to the familiar Performance and SEO scores: Agentic Browsing. Google added it in mid-2026, and it grades something none of the older checks do — how readable your site is to an AI agent, the kind that will increasingly visit a site on a person's behalf to find a phone number or book an appointment. It's early and explicitly experimental, so I wouldn't lose sleep over the score. But notice what it tells you: the definition of "a good website" just grew another dimension, quietly, without anyone asking your permission. One of the things that audit checks is layout stability — whether elements jump around as the page loads — which is the very same thing that makes a human thumb miss your booking button. The bar for what a site has to do keeps rising, and it rises in places you won't see unless you're looking. Which is the whole point of this piece.

The shape under all of them

Step back and every one of these has the same anatomy. A real thing happens — a booking, a call, a form, a patient. Your data doesn't see it. And Google, learning only from what your data reports, optimizes toward an incomplete picture and away from the parts it was never shown. The dashboard stays full of numbers the whole time. That's what makes these leaks so durable: nothing looks broken. You only find them by walking the path deliberately and asking, at each handoff, does anything actually record what just happened here.

This is why I keep coming back to the same unglamorous discipline: measure to the booked patient, not to the click. The click is the cheap, visible, seductive number. The booked patient is the one that pays your rent, and it's the one that lives at the far end of exactly the handoffs that tend to break. An ad program you can't measure to the booking isn't a cheaper version of one you can — it's a different thing entirely, optimizing toward a target it can't actually see.

If you want to know which of these are open on your own site, that's a concrete thing to check rather than worry about — it's part of what the free audit looks at: not just whether your site loads and ranks, but whether anything is actually counting the patients who arrive.