Hardly anybody reaches the bottom of your page. Some start there.

You have never been to the bottom of your own home page

You have looked at it hundreds of times. Almost always on a computer, usually while somebody was changing something on it, and never the way a customer does: on a phone, one thumb, standing up, with four other results still open behind it.

So go and do that now, and do not read a word of it. Swipe at the speed you actually move and count how many times the screen fills with something new before you run out of page. Most small business home pages come out between six and ten.

On the machine it was drawn on, the same page is about three.

That is not a criticism of whoever built it. A wide screen holds three things side by side and a phone holds one, so every row becomes a stack and a page nobody thought of as long turns into a corridor: how that happens without anybody deciding it.

A screenful is not a fixed amount

Worth knowing before you count. On a phone the address bar slides away as you go down and comes back the moment you go up, so the amount of page you can see changes while you are looking at it, by about the height of that bar, in both directions.

Not a quirk of one phone, either. It is settled enough that the people who write the rules browsers are built to have given up on there being one answer, and now publish three separate definitions of a screen: the smallest, with the bar showing, the largest, with it hidden, and a third that changes as you travel.

So where your page appears to end on your phone is not where it ends on your customer's.

Two people are shouting and both are selling something

Ask how long a home page should be and you get two answers, each delivered with total confidence, and both older than the phone in your hand.

The first is that nobody scrolls, so everything important goes at the top. That was true, once. It was observed in the mid nineteen-nineties, when people did not yet know a page could carry on below the window, and the man who wrote the guideline withdrew it in 1997 because everybody had learnt the gesture. It has been repeated ever since by anybody who would like to sell you a busier top of the page.

The second is that long pages do better, and that comes almost entirely from people testing sales pages: software subscriptions, online courses, coaching programmes. A cold visitor being asked for a card number does need every objection answered first. Real finding, different situation. Nobody with water coming through the ceiling is being talked round. They have already decided they want a plumber.

The best evidence there is

Underneath the shouting there is proper research. The same usability group ran the same kind of study twice, fifteen years apart, watching where people actually looked.

People did learn to travel and the picture got a good deal less extreme. The shape of it did not change. Attention is spent early and what is left thins out quickly.

And now the bit that gets left off

Read the last line of each of those columns again. Both studies were run in front of desktop monitors. The second one on wide screens, sitting down, with a mouse.

The figure everybody quotes at you about scrolling on a phone was never measured on a phone.

There is a second number quoted the other way, to prove people go a long way down. Two billion visits, two thirds of the attention on a media page spent below the first screen. Also real, also about something else: a news article is a column of writing, and going down it is the entire point of opening it.

Which leaves an honest answer

Between them, every study anybody can point you at covers news sites, shops, blogs and help pages. Not one plumber. Nobody has measured how far down a small local business home page anybody gets, so a percentage quoted for your trade came off a news site or out of the air, and you are allowed to ask which.

What can be said is narrower and steadier. Attention is front-loaded everywhere anybody has looked, on every kind of page and every size of screen, and it thins out fast after the second. That has held for fifteen years and through a change of device.

Nobody stops because their thumb got tired

A swipe costs nothing. It is the cheapest thing anybody does on a phone, which is why the reasons people stop are almost never about effort. They are about the page, and there are five of them.

• A block that ends exactly at the bottom of the screen

  A band of colour or a photo filling the screen with nothing showing underneath it. The most common reason somebody stops on a page with five more screens to give. In a builder this is section padding. Bring it down one size.

• Three sections that look like each other

  Pale cards, a small symbol, a paragraph, again. Nothing marks them apart, so the page reads as though it has begun repeating itself. The fix is not less content. It is making the next thing look like a different kind of thing.

• A map that swallows the swipe

  The thumb lands on the embedded map, the map slides instead of the page, and the harder they try the further from home they get. Turn off dragging and zooming in the map block, or show a picture that opens the real map when tapped.

• A row that moves sideways

  Review sliders and photo galleries read a slightly diagonal downward swipe as a sideways one, so the page does not move at all. Most people try twice. The row is fine. Sizing it to fill the screen is what makes it unavoidable.

• Something arriving after they set off

  A reviews panel with no space kept for it turns up late, the page grows, and everything jumps under a moving thumb. They have not lost interest. They have lost their place, which is worse and faster.

The fifth one has a longer story behind it than the others, because a page that moves under somebody is also one of the most reliable ways to send them away entirely: what actually sends a thumb to the back button.

Let something be cut in half

The cheapest fix in this whole piece, and it is a spacing change rather than a writing one. A page that ends cleanly at the bottom of the screen looks finished, and it looks finished whether or not it is.

So the useful thing to do with that bottom edge is to make sure something is crossing it. The top half of a heading. The first line of the next paragraph. The top of a photograph. Anything sliced by the edge of the screen is taken as a promise the page continues, and it is taken without anybody deciding to take it.

You cannot control where the edge falls. What you can control is how many clean stopping places the page offers, and a page with fewer full-width blocks sized to fill a screen has fewer places to be mistaken for finished.

Some people arrive at the bottom on purpose

There is a habit worth knowing about, because it turns everything above on its head. A certain kind of person opens a website on a phone and goes straight to the bottom. Not by scrolling. Three or four hard flicks, looking at nothing on the way down.

They are not reading and they are not lost. They are checking, and they have learnt that the bottom of a website is where the true things are kept, because the top is where the marketing is. What they want is always the same four things. Is this a real business. Is it near me. Is it allowed to do this work. How do I ring it.

Nobody has counted these people and anybody quoting you a figure is guessing. What can be counted is the cost of being ready for them, which is nothing at all.

The part that is not optional

Some of what belongs down there is not a trust signal at all. It is a condition of advertising, and a website is advertising.

• New South Wales

  Fair Trading wants the licence holder's name, the licence number and a business phone number in all advertising, and it says websites in as many words. Residential building work over $5,000, and electrical, plumbing, draining, gasfitting and air conditioning at any value at all.

• Queensland

  The QBCC wants the licensee's name and the licence number, easily legible and reasonably prominent. If you are not licensed the advertisement has to say what you cannot do, and for plumbing and gas fitting the figure it prints is zero.

• Victoria

  Energy Safe Victoria wants a registered electrical contractor's number on anything at all that says they will do electrical work. The building regulator asks much the same of registered practitioners.

Licensing is state by state and that is three of them. Everywhere else the rule is the same shape: find your own regulator's page about advertising and read it once. It is usually about a screen long, and it is the only page mentioned in this piece that can fine you.

And the part that is only sensible

• The hours, and what happens outside them

  Somebody looking at this at nine at night is deciding whether ringing now wastes their evening. Say it, including when the answer is no.

• Where you actually go

  The area, or the handful of places you genuinely work. Not forty suburbs in small grey type, which reads as a trick because it is one.

• Your ABN

  Nothing requires this on a website. The law wants it on your invoices. Down here it answers somebody wondering whether you are a real business.

• The name, address and number, exactly as Google has them

  Same words, same order, down to whether it says Street or St. Search engines cross-check it. So, in four seconds, does a customer.

The button that goes back to the top

It comes up every time and it is worth one paragraph. There is no evidence either way, in any direction, for any kind of page. It costs nothing, and on a nine-screen page the alternative is a long swipe with nowhere to rest, so put one in if your builder offers it and think no more about it. Be clear what it is, though. It is a handrail on a staircase somebody built too long.

Six things to try on your own page

About five minutes, on your phone, standing up, at the speed you actually move. Sitting at a desk being thorough, every website ever built reaches the bottom. Nothing here is scored and there is nothing to send anybody.

1. 01The count. Open your home page and swipe to the bottom at your normal speed, counting the screens.Write the number down before you do anything else. It is the only measurement in this piece you can take yourself, and most owners are out by about half.
2. 02The flick. Do it again without reading, three or four hard flicks straight to the bottom.Look at where you land. That screen is the first thing a certain kind of customer sees, and quite often the only one.
3. 03The edge test. Go down slowly and stop at every join between one section and the next.You are looking for a join that lands flush with the bottom of the screen. Each one is a place your page looks finished when it is not.
4. 04The trap test. Put your thumb on the map, then the reviews, then the gallery, and try to scroll past each one.If the page stops moving, or the thing slides sideways instead, your customer met the same wall. Usually a switch in the block settings, not a rebuild.
5. 05The screenshot test. Screenshot every screenful on the way down, then look at the shots together.Two that look the same are the ones to worry about. On a phone they are eight seconds apart, close enough to feel like going backwards.
6. 06The Google check. Read your last screen next to your Google listing, word for word.Business name, address, phone number. Not roughly the same. The same. Cheapest thing here and the one most often wrong.

If you only do one of them

Do the count, and do it before you change a thing. Almost every argument you are going to have about this page needs a number nobody currently has, and it takes about eleven seconds to go and get it.

There is no right number of screens

Worth saying plainly, because the obvious way to act on all this is to go and cut the page in half, and somebody will tell you within the week that a longer one does better. Both are confident answers to a question nobody can answer, and that includes us.

Here is the one thing worth committing to. However long your page turns out to be, it has an end, and the end is not where you ran out of things to say. It is a screen with a name, a number, an area and a licence on it, and for a certain sort of customer it is the first screen they will look at.

Build that screen on purpose, and the argument about the eight above it matters a good deal less.