Every booking-abandonment stat we checked had no sample size, no date, no method

Most customers who give up on your online booking never tell you. They hit something small — a form that rejects their email address, a confirmation code that never shows up, a date picker too fiddly to tap one-handed — and they go back to the search results and call the next business on the list. There is no trustworthy public figure for how often this happens to a local service business. The percentages you will find online come from travel and retail studies, or from booking-software companies that never publish their method. The only number that matters is yours, and you can find it this afternoon by booking yourself as a stranger.

How many bookings am I actually losing?

Nobody knows, and you should be careful with anyone who tells you they do. Search this question and page one hands you confident numbers: that most customers now demand online booking, that two-thirds of bookings happen on a phone, that asking for too much information raises abandonment by half. Follow each number back and it was published by a company that sells booking software. None of them states a sample size, a date, or how the data was collected.

The one well-documented body of research on abandonment measures retail checkout — people putting products in a cart and leaving. That is a different act than a homeowner trying to get a plumber to their house on a Tuesday. Travel-booking research has the same problem. Neither transfers.

Here is what you can measure without buying anything:

• Form starts versus form completions. Most booking tools and analytics setups will show you both. The gap is your number.
• Bookings versus page views on the booking page. Crude, but it moves when you fix something, which is the point.
• Calls that begin with a complaint about the website. Count them for one month. Each one represents an unknown number of people who did not call.

Why don't I ever hear about this?

Because the customer who phones to tell you your website is broken is the rare one who wanted you badly enough to try twice. Everybody else leaves quietly, and from where you sit that looks the same as nobody visiting the page at all.

Your own testing hides the problem too. When you check your booking page you are logged in, on your own wifi, on a phone that has visited the page a hundred times, with your browser filling in half the fields, and you already know what every question means. A stranger has none of that. They arrive from a Google result on a cellular connection with one hand free, and they are giving you about ninety seconds before the next business gets their call.

What actually breaks a booking, in order of how invisible it is?

These are the failures that do not look like failures. The page loads, nothing shows an error to you, and the booking simply does not arrive.

The email address one is more common than owners think

The rules for what is a legal email address are wider than most homemade form validators allow. Under RFC 5322, the internet standard for message format, the part before the @ may contain letters, digits, and a list of punctuation that includes +, -, _, ', . and more. So maria+plumber@gmail.com is valid. So is o'brien@example.com. Plenty of booking forms reject both, along with long addresses and newer domain endings. The customer sees "invalid email", knows their email is fine, tries twice, and leaves.

The email that never lands is a settings problem, not a customer problem

If your booking system sends confirmations from your own domain, mailbox providers decide whether those messages reach the inbox, land in spam, or get refused before delivery. That decision is based partly on authentication records set up in your domain's DNS. Google publishes its requirements for senders; whoever set up your website or email can tell you in ten minutes whether you meet them. Until then, assume a share of your confirmations are landing in spam, because that is what happens to unauthenticated mail.

The text that never arrives is usually a registration problem

Business texting from a normal 10-digit US number runs through a system called 10DLC. The Campaign Registry, which operates it, states that brands and their campaigns are verified before they are allowed to send. If your booking tool texts confirmations and that registration was never completed or was rejected, the messages can fail without anything on your screen saying so. Ask your provider for the status in writing.

How small is too small to tap?

There is a published floor. WCAG 2.2, success criterion 2.5.8 (Level AA) says the target for pointer input should be at least 24 by 24 CSS pixels. That is the minimum, not the goal — Apple's guidance for iPhone is 44 points and Google's Material guidance is 48. Calendar grids are the usual offender, because a month of dates squeezed onto a phone screen puts thirty-one targets into a space built for about seven.

How fast does the page have to be?

Google publishes thresholds for this too, measured at the 75th percentile of real page loads, split between phone and desktop. A page is in the "good" band when the largest thing on screen finishes drawing within 2.5 seconds, when it responds to a tap within 200 milliseconds, and when its layout shift score stays at or below 0.1. That last one is the sneaky one: a page that finishes loading after the customer has started tapping moves the button out from under their thumb.

Does "booking" even mean the same thing in my business?

Most advice about booking forms is written as if every business fills fixed slots on a calendar. Half of them don't, and applying salon advice to a plumbing company is how you end up with a booking page nobody can use.

If you dispatch, the honest booking form is short: what's wrong, where, and how to reach you. Everything else is a conversation. If you run on appointments, the calendar is the product, and every hour of stale availability on it costs you twice — once when someone books a slot you can't honor, and once when someone leaves because your only free slot appears to be next Thursday.

How do I test my own booking page?

Set aside twenty minutes and book yourself as a stranger. The whole value is in refusing every advantage you normally have.

1. Log out of everything. Use a private window, or better, a phone that isn't yours.
2. Turn wifi off. Cellular only. Your customers are not on your office connection.
3. Start from Google, not from your own URL. Search the way a customer would and tap through whatever they would tap — the map listing, the directory, the ad.
4. Use a real email you can open, not your business address, and put a plus sign in it: yourname+test1@gmail.com.
5. Use a real mobile number that can receive a text.
6. One hand. No zooming. If you find yourself pinching to zoom, that is a finding, not an inconvenience.
7. Time it, and write down every second where you had to stop and think.
8. Then check the whole chain: did the email arrive, is it in spam, did the text arrive, did the booking appear on the calendar you actually look at, and did anyone get notified who needed to be?
9. Do it again at 9pm and again on Sunday. A booking page that works during business hours and dies after hours is a booking page that only works when you didn't need it.
10. Hand your phone to someone who has never seen the page and say nothing while they try. Do not help. Do not explain. Write down where they stopped.

That last step is the one owners skip, and it is the one that pays. What people say about your booking page is polite. Where they stop is true.

What if a customer still can't finish the form?

Some won't. A customer in their seventies, a customer whose first language isn't English, a customer standing in a flooded garage on a cracked phone — for them, no amount of form design is going to be enough, and they are frequently the customer with the most urgent job and the least price sensitivity.

So build the other path on purpose instead of leaving it to luck. That means a phone number that gets answered, including after hours; someone able to take the details and enter the booking on the customer's behalf; and a link sent to them afterwards to confirm or pay. A booking flow that works for most people, with a human behind it for everyone else, beats a beautiful form with nothing behind it.

MyNearbyPro builds the front half of that for free — a simple site with a booking page aimed at the way service customers actually behave — and can put an AI front desk behind it that answers the calls the form loses. Either way, run the twenty-minute test first. You cannot fix a leak you have not found.