Why Nobody Knows Where Bookings Drop Off

There are three questions every company with online appointment booking should be able to answer, and that I almost never get answered:

How many people started a booking last month? How many completed it? And at which point did the others drop off?

What is available instead is a single metric: the number of booked appointments. It lives in the booking system, it's accurate, and it tells you nothing about how many people actually tried. That makes any optimization pure guesswork—you can't tell whether the offer lacks visibility, the flow is too long, or the meeting types aren't right.

Without a funnel, there is no diagnosis. There is only a result you like or don't like.

The Six Stages That Need to Be Measured

The path from interest to revenue has six transitions. Every single one can leak, and each has a different root cause and a different fix. That's why simply knowing the start and the end isn't enough.

And beyond that, off the website: the closed deal. Only when Stage 6 is connected to the outcome of the call can you tell which channel generates not just bookings, but actual business.

The most critical metric in this table lives in Stage 3 and is overlooked almost everywhere: How often does someone abandon because no suitable time slot was available? This isn't a conversion problem; it's a capacity problem—and it gets misdiagnosed as a marketing weakness as long as no one measures this stage.

Why the Numbers Are Missing: The Technical Side

1 · The Booking Flow Belongs to Someone Else

This is the primary cause, and it is structural. In most cases, the booking process doesn't run on the company's own website, but inside an embedded iframe or on the vendor's subdomain. To the browser, this is a different origin—and that's where tracking stops dead at the border.

Specifically, this means three things. First, the cross-domain tracking mechanism that analytics tools use to stitch sessions together doesn't work for embedded frames: it triggers on a link click, not when an iframe loads. The session on the page and the session inside the frame appear as two completely separate visits.

Second, your tag manager cannot see inside the embedded iframe. This isn't a misconfiguration; it's a browser security rule—scripts from one domain are not allowed to read content from another.

Third, and this is the costliest issue: The booking event fires in the vendor's analytics account, not yours. The booking happens, and it gets counted—just somewhere else. In your own report, it looks like a visit that simply drops off at the booking window. It reads like a tracking error, but in reality, it's an architectural flaw.

2 · Consent limits what is even visible

In German-speaking markets, a factor comes into play that is often overlooked in international guides. Analytics tools with user recognition require consent, and far from everyone grants it. The numbers vary considerably depending on the banner design and industry: An analysis of over 1.2 million visitors showed that roughly a quarter accepts everything on first contact, while about a third simply ignores the prompt. Other surveys report drops in measurable visitor numbers on the order of ten to thirty percent.

This doesn't mean tracking is pointless. It means two things. Absolute numbers from the analytics tool are systematically too low—anyone comparing them to booking figures from the core system will always find a gap, and that gap is not a bug. And conversion rates are more reliable than absolute volumes, as long as the consent rate remains reasonably stable: If a certain percentage of tracked users converts, that percentage also applies approximately to the untracked users.

Ignore this, and you'll end up in endless debates about which system is right. Name it, document it once, and you can actually work with the numbers.

3 · Source tracking gets lost along the way

Campaign parameters—the tags appended to a URL that identify channel and campaign—are the only way to attribute a booking to its source. They get lost more often than you'd think: during redirects, when entering the booking flow, or when switching from an app to the browser.

But the more frequent damage isn't loss—it's inconsistency. Analytics tools treat every variation in spelling as a distinct value. "LinkedIn", "linkedin", and "LI" become three separate channels. A newsletter named differently by three team members appears as three distinct campaigns. The result is a report no one trusts, because everyone misses their own way of counting.

Why it still doesn't happen: the real root cause

Everything mentioned so far is solvable. The processes are documented, the tools exist, and the effort is manageable. Yet the majority of companies I work with lack end-to-end tracking. The reason isn't technical.

The chain spans four distinct areas of responsibility. Marketing sets the campaign parameters. IT or the web agency manages the tag manager. The booking flow belongs to the software vendor. Blending this with closed deal data falls to the Data Warehouse or Business Intelligence team. Four stakeholders—and therefore nobody responsible for the chain as a whole.

Every single party can deliver their part correctly, and the result still doesn't work. That's precisely what makes this so frustrating: there is no single party at fault and no obvious error. There is only an outcome no one can produce because it relies on four handoffs that none of the four feel accountable for.

On top of that, there is a knowledge gap that is rarely acknowledged. The marketer who needs the report usually can't evaluate why the booking window breaks the session. The IT engineer who could fix it doesn't know what business question the report needs to answer. Both are competent, but there is no one between them to translate across the divide.

And finally: Tracking is nobody's goal. It is in no department plan, it has no budget, it produces nothing showable. It is the prerequisite for others to prove their goals — and tasks like that lose all prioritization. That's the exact same pattern as the booking button that sat in the footer for two years.

The smallest setup that works

The most common mistake at this stage is setting up a massive initiative: data model, warehouse, attribution model. Projects like that take quarters and, in the end, often still don't answer where the booking process drops off. The path I recommend is the opposite — start small, but complete.

After that — and truly only after that — linking to the deal data from the sales system is worth it. It is the actual purpose of the exercise, because it turns booking statistics into a revenue calculation. But it requires the four steps before it to be in place.

What you can answer with it

The end result isn't a data landscape, but the ability to answer a few questions that currently remain unanswered:

• Where is the drop-off? If half of the people leave the booking flow at the form stage, that is not a marketing question, but a question of field count.
• How often was there simply no availability? The capacity question disguised as a conversion problem.
• Which channel brings meetings, which only brings clicks? Only then can budget be reallocated instead of increased.
• What does a booked meeting cost per channel? The metric that belongs in place of cost-per-click in the consulting business.
• Which meeting type leads to a closed deal? The most insightful analysis of all — and impossible without end-to-end tracking.

None of these questions require an attribution model or a warehouse. They require six cleanly defined events, a closed domain boundary, and one person who takes ownership.

Takeaway

The number of booked meetings is a result, not a diagnosis. If you want to know where it's bleeding, you need six stages: impression, entry, selection, form, confirmation, attendance. The most common technical reason for the gap is the domain boundary to the booking flow — where the event often fires in the provider's account instead of your own. But the more common reason is organizational: the chain runs through four areas of responsibility, and none of them owns the whole thing. That's why the solution doesn't start with a tool, but with a name.