Eight reached our form, two started it, none sent it

A question you probably cannot answer about your own property

You know last month's occupancy to one decimal, and whether it was up or down on the same month last year. Now a harder one: in the last seven days, how many people opened your direct booking page and left without finishing? Not how many booked. How many started, stopped, and at which step.

Most independent operators cannot answer that, and it is the number sitting directly between traffic and money. We could not answer it for our own site either, until we instrumented it. What follows is our own measurement, with the sample size stated plainly, and a method that costs nothing to copy.

Seven days on our own site

The Nightfill site is a single page with a contact form at the bottom. Read from our own event logging this morning, for the trailing seven days:

• 29 sessions from 25 visitors, 37 page views.
• 8 sessions reached the form.
• 2 of those sessions started typing in it.
• 0 sent it.
• Of the two starts, the single abandonment our logging could place landed on the free-text message box, which is the last field on the form.

Two more readings from the same week. The contact section held 102.2 seconds of attention per view; the pricing section held 4.4 seconds per view. And of the sessions that scrolled at all, 48.3% passed the halfway mark of the page and 41.4% reached the bottom. Nobody was bouncing off the top. They went the whole way down, stood in front of the form for close to two minutes, and left.

Read the caveats before you borrow any of that

29 sessions is not a sample, it is an anecdote with a timestamp. Seven were our own automated probes. All three call-to-action clicks in the window arrived with no pointer position recorded, which is what a keyboard user, a script or an owner testing his own site looks like, not a visitor with a mouse. Our device split that week, in Google Analytics, was 15 desktop to 7 mobile, because we sell to operators; a guest-facing booking page rarely looks like that.

One more thing worth knowing before you ever compare two dashboards: our own logging counted 29 sessions that week and Google Analytics counted 22, for the same site and the same days. Neither is lying. They define a session differently and lose different traffic to blockers. Pick one, stay with it, and read the movement rather than the absolute level. We wrote about that discipline in measuring from the day you changed the price.

What is worth taking from our week is not a number. It is that we can name a field. Most properties, asked where their booking path loses people, can only say "the website".

Why the field matters more than the rate

A conversion rate is one number laid over eight or ten separate decisions. It tells you people are not finishing. It does not tell you where they stop, and where they stop determines the entire fix. Someone leaving at the date picker is a different problem from someone leaving at payment, which is different again from someone leaving at a box that asks them to write a paragraph about themselves before you will quote a price.

This is the same reason our pricing engine does not keep one price. At Maka Hostel in Oaxaca, our own property, it maintains five rate bands by how far ahead the booking falls: same day, 1 to 3 days, 4 to 7, 8 to 21, and 22 or more. One average price across the five would hide which lead time is actually empty. Splitting it is not sophistication. It is the difference between a number you can act on and a number you can only recite.

What splitting a number taught us about discounting

Across 32 logged runs covering 22 calendar days between 14 August and 5 September 2026, the engine pushed Maka's same-day discount from 0% to -8%, in three-point steps, chasing weak near-term occupancy. Then it compared the two regimes: the 23 logged days at -8% averaged 33.7% occupancy, the 5 days at 0% averaged 42.0%. Deeper discounting was not filling more beds. The engine wrote that in its log and held the cap rather than cutting further.

Handle that finding carefully, because we do. It is low season only, two regimes with very unequal sample sizes, no control for day of week, and it is the engine reading its own log rather than a clean experiment. It is directional, not proof. But it was readable at all only because the two price regimes were recorded separately. Had the log kept one line - average occupancy, roughly 35% - there would have been nothing to learn, and the discount would have kept walking down.

Your booking form has the same structure. A single completion rate is "average occupancy, roughly 35%". The step counts are the bands.

Four numbers and a notebook

You do not need software. At the volume a 20 to 80 bed property runs, a spreadsheet is enough. Record four numbers a week:

• Sessions that reached your booking page or enquiry form.
• Sessions that picked dates, or typed into the first field.
• Sessions that reached payment, or the send button.
• Bookings or enquiries completed.

Then do three things with them. Count the fields on your direct path, front to back, including the ones added after payment, and account creation if there is one. Every required field is a checkpoint where a person can stop. Name the last one. A lot of enquiry forms end with an open text box, and ours is exactly where our only located abandonment happened. Then change one thing for two weeks - make a field optional, delete it, or move it to after the booking is confirmed - and compare enquiries per 100 sessions, never total enquiries, because your traffic moves for reasons that have nothing to do with your form.

If your booking engine will not show step-level data, ask your provider for it by name. It exists on their side.

What this does not prove

Nothing here says that removing a field will raise bookings at your property. Our sample is 29 sessions and two form starts, on a business-to-business site, in one week, and we have told you which parts are probably us. The Maka pricing result is low-season, directional, and the engine's own read.

What it does say is narrower and more useful. The question "where does my booking path lose people?" has an answer. That answer is cheap. Most properties are currently choosing not to have it. We ran one week of our own traffic and we can now name a field; the week before, we could not. That is the entire gap, and closing it costs a spreadsheet filled in on the same day each week.

If you would rather have the pricing side running while you go fix your form, see the three Nightfill tiers and start a free 14-day pilot on your own data: view pricing. Nightfill is independently operated; results vary by property, market and season.