Someone reads your page to the end, clicks "book a meeting" and receives a calendar link. From where you sit, scheduling is now automated. From where they sit, it has just begun: open the link, guess a duration, pick a day, find the result again in their own calendar, and then actually turn up. We will look at three things: where that journey really breaks, what solid numbers exist for the most popular remedies, and why the most effective change is almost never the reminder.
Why is a calendar link so often the end of an enquiry?
Because it hands back a task instead of taking one away. The moment in which somebody wants a meeting is short. It arrives at the end of a piece of text, after a price comparison, between two other jobs. Whatever is not completed in that moment is rarely picked up again.
One of the few independent studies with a clean method shows how poorly this goes in practice. In the first quarter of 2025, Hennessey Digital filled in the contact form of 1,333 US law firms as a mystery shopper, always between 10am and noon local time, always with the same invented story. The result: the median response time was 13 minutes, 25 per cent replied in under five minutes - and 26 per cent never replied at all. Of those who did reply, 87 per cent picked up the phone.
That last figure is the interesting one. The dominant response to an enquiry is still the callback, which is an attempt to recreate the moment of intent later on. In a quarter of cases nothing happens at all. A calendar link is an improvement on that state of affairs, but it only half solves the same problem: it is available while the moment is still alive, yet it still asks the prospect to do the work. How much that difference is worth has never been measured independently. There are vendor numbers, and they should be read as such: in 2025 Chili Piper analysed four million form submissions from its own customer base and reported a 66.7 per cent booking rate where the visitor can self-schedule immediately after submitting. The comparison figure of roughly 30 per cent comes from the same source and has not been independently verified. Useful as a direction, not as a number.
Where exactly does the journey break?
In two separate places, and that is where most setups stop thinking too early. A meeting comes about in two steps, each with its own rate: an enquiry becomes a booking, and a booking becomes a conversation that actually happened. Anyone who only tracks one of the two is optimising blind.
The arithmetic is deliberately simple, because it is meant to show an order and not to forecast anything. Both improvements are the same size, but one works on 100 enquiries and the other only on the 30 meetings that made it into the calendar at all. So the rule is: fix the booking journey first, then the no-show rate. In practice it usually happens the other way round, because reminders are easier to switch on than a rebuilt enquiry journey.
How much do reminders actually achieve?
Less than promised, and this time it can be shown properly. In marketing there are only vendor case studies on this question. In healthcare there are controlled trials, because missed appointments have been a cost problem there for decades. A systematic review with meta-analysis, published online on 6 March 2026, screened twelve randomised trials and pooled ten of them covering 8,236 participants.
Two things in that chart are uncomfortable reading. First, the overall effect is an eleven per cent relative improvement, not a halving. Second, of all things it is SMS - the tool every booking product offers first - that misses statistical significance on its own. Its point estimate is the highest of the three, but the confidence interval reaches below 1.0. Translated: the effect is probably there, the data are simply not strong enough to prove it.
How small the steps are in reality is shown by the largest published evaluation to date. In 2024 a US health system added an automated call with confirm, cancel and reschedule options to its existing SMS reminder for 59,994 patients at high risk of missing their appointment, over a four-week period.
One extra reminder stage, rolled out across nearly 60,000 cases, is worth 1.7 percentage points. That is a real, cleanly measured effect, and it is worth having because it costs almost nothing. It is simply not what tool comparisons claim reminders do. And it comes from healthcare, where people hold an appointment they actually need. For a first conversation with a service provider, which can be dropped without consequence, the effect is more likely to be smaller than larger.
What belongs in a booking journey that holds?
Five points, in this order, because the first two work on the larger number.
Bookable where the intent appears. Not "we will be in touch", not only after a form, but directly after the action somebody has just taken voluntarily. Anything that interrupts the moment costs bookings. That is the real point of speed to lead: not calling back faster, but not having to call back at all.
As few decisions as possible. An empty calendar offering four weeks of choice is not freedom, it is a task. Set the duration, propose two or three concrete slots, and hide everything else behind "another time".
Dose the qualifying questions deliberately. Every additional question before the booking costs you bookings and buys you better conversations. There is no public number for that trade-off that fits your business, which makes it the one place where you genuinely have to measure for yourself. What that looks like without friction is described in From enquiry to meeting.
A confirmation that contains something. The name of the person they will be speaking to, the duration, one line of agenda, and a cancellation link that works in a single click. A cancellation is not a loss, it is the better outcome: the slot comes back and you know in advance. That is precisely what the automated call in the trial above does, offering confirm, cancel and reschedule in one.
Keep the two rates apart. Booking rate and no-show rate belong in your CRM separately, not in a shared "meetings" tile. Otherwise you watch the meeting count fall over the summer without knowing whether fewer people are booking or more people are failing to appear. Those are two entirely different problems with two entirely different fixes.
What you do not need for any of this is an AI agent. A booking journey is a process with fixed rules, and that is exactly what an automation workflow is for, at a lower price. Where the boundary runs is set out in AI agent or automation workflow.
Three levers for the coming weeks
Book yourself. Take your phone, walk the path from the home page to the confirmation email, and count the clicks and decisions. If you have to think twice about how long the meeting should be, a stranger certainly will.
Separate the two rates. For one month, count enquiries, bookings and conversations held as three distinct figures. That is one extra row in the spreadsheet, and it answers the question of which of the two places actually has a problem.
Build the cancellation before you build the third reminder. One click to cancel buys you free slots and honest numbers. On the evidence, the third reminder buys you one to two percentage points and a mildly irritated prospect.
If you want to know where your booking journey loses people, we are happy to walk it from the outside with you. It is rarely the tool - it is usually three decisions too many. 📅