Conversion focused web design: one decision per page

Conversion focused web design has a bad reputation, and it earned it. The phrase has been attached to green buttons, countdown timers and pop-ups that ask for an email before the visitor knows what the business does. None of that is design for conversion. It is decoration applied to a page that has not decided what it is for.

What follows is the method we actually use. It is mostly about deciding things in the right order, and the order is the part that gets skipped.

Start with the decision, not the layout

Every page exists to move one decision forward. Not to communicate, not to showcase, not to build brand. To move one decision, made by one kind of person, at one point in their thinking. Writing that sentence down before anyone opens a design tool is the single highest return thing on this list, and it is free.

The test is subtractive. Once the decision is written down, every block on the page has to justify itself against it. A section that does not help someone make that decision is not neutral, because it costs attention and pushes the thing that matters further down. Most pages get better by deletion long before they get better by addition.

A page is an argument, and arguments have parts

A page that converts is making a case, in the order a sceptical person needs to hear it. We think there are four parts, and the missing one is usually the third.

1. The claim: what this is and who it is for, said plainly enough that someone can repeat it to a colleague
2. The evidence: how it works, what is included, what it costs, and anything checkable that supports the claim
3. The objection: the reason this person is about to leave, answered before they have to ask for it
4. The next step: one action, described in terms of what happens next rather than what the button does

The objection is the interesting part, because it is specific to your business and nobody can write it for you from a template. It is usually one of four: it costs too much, it will take too long, we tried something like this and it failed, or I do not believe you can do this for a business like ours. The page that names the objection out loud converts the reader who was about to leave silently.

What the first screen is actually for

The first screen does not have to sell anything. It has to survive a glance and earn the scroll. That means it answers three questions fast: what is this, is it for me, and what happens if I keep reading. A headline describing a feeling rather than a thing fails all three.

We build hero sections in the browser on a phone first, because the desktop version flatters everything. On a small viewport, the headline, one line of qualifying detail and one action are often all that fits above the fold, and that limit is a feature. It forces the decision about what matters.

Friction you can see, and friction you cannot

Visible friction is the form with eleven fields. Invisible friction is subtler and more expensive: a price the page will not name, a service that could mean three different things, a company that will not say where it is based, a required account before anyone can see what they are buying.

Our rule is to answer the questions in the order a buyer asks them, not in the order a business likes to present them. That usually means moving price, scope and timeline much earlier than the marketing team is comfortable with. A visitor who cannot find out roughly what something costs assumes the worst and leaves, and nothing in your analytics will record why.

Forms: ask for less, and say what happens next

The form is where intent is converted into a record, and it is where most of it is lost. A few things reliably help.

• Ask only for what is needed to have the next conversation, and collect the rest in that conversation
• Label every field visibly, because a placeholder disappears the moment someone starts typing and takes the instruction with it
• Use the right input types and autocomplete attributes, so a phone offers the right keyboard and the browser can fill what it already knows
• Validate on submit rather than punishing someone mid-word, and put the error message next to the field it belongs to
• Say what happens after submission: who replies, from where, and roughly when
• Keep consent to marketing a separate, unticked choice, which the Digital Personal Data Protection Act, 2023 requires of anyone collecting personal data in India

The last point is worth reading properly at meity.gov.in/data-protection-framework rather than trusting a plugin's defaults. A pre-ticked consent box is not consent, and bundling marketing permission into a contact form is a compliance problem before it is a conversion problem.

Speed is a conversion feature, not a technical one

A page that is still moving when a thumb arrives converts worse than one that is not, and no amount of persuasive copy compensates. The Core Web Vitals thresholds at web.dev/articles/vitals are a reasonable definition of good enough: how quickly the main content appears, how quickly the page responds to the first tap, and whether anything jumps while being read.

The one that quietly destroys conversion is layout shift. A button that moves as an image or a banner loads produces a mis-tap, and a mis-tap on a phone reads as the site being broken. Reserving space for images with width and height attributes or an aspect ratio costs nothing and removes most of it.

Performance is a budget set at the start and held during the build, not a clean-up at the end. In our experience a site that is slow at launch is slow because of decisions made in the second week, and those are expensive to unpick later.

Accessibility is conversion work

This gets filed under compliance, which is how it ends up last in the schedule. Most of WCAG 2.2, published at w3.org/TR/WCAG22, describes things that help everybody buying on a phone in poor light with one hand.

• Target size: controls big enough to hit without precision, which is the difference between a tap and a mis-tap on a moving train
• Focus appearance and focus not obscured: the person tabbing through your form can see where they are, and a sticky header does not hide the field they are filling
• Dragging movements: anything you can drag can also be done with a simple tap, because dragging is hard for a lot of people and unreliable on touch
• Accessible authentication: no cognitive puzzle in the way of a login, and paste allowed in password fields
• Contrast that survives sunlight, which is where a good deal of mobile browsing happens

None of that is charity. It is the same list you would write if you only cared about revenue and had watched enough people use a checkout on a phone.

Write the button, do not decorate it

Button copy is a small thing that repays attention, because it sets the expectation for what happens next. A label describing the outcome beats one describing the mechanism. The visitor is not deciding whether to submit a form, they are deciding whether to start a conversation with a stranger about money.

The same applies to what sits next to it. A single line saying what a call involves, how long it takes and whether it ends in a quote removes more hesitation than any change to the button itself.

One page, one primary action, repeated as often as the length of the page requires. Two competing calls to action of equal weight is not giving the visitor a choice. It is asking them to do your prioritisation for you, and the common result is that they do neither.

Testing honestly when your traffic is modest

Split testing is sound in principle and frequently misapplied on business sites, because a page receiving a modest number of enquiries a month cannot resolve a small difference in any sensible timeframe. Calling a winner early, on a handful of conversions, produces confident decisions built on noise.

When traffic is thin we prefer a different loop. Watch session recordings and look for the point where people stop. Read the enquiries and the sales calls for the question that keeps coming up, then answer it on the page. Change one substantial thing at a time and compare over a long enough period, accepting it as a judgement rather than a measurement. That is less scientific than a test and more honest than a badly powered one.

The things a sceptical buyer checks

Before anyone fills in a form they run a quiet background check on you. The items are unglamorous and they are nearly always the same.

• A real address and a phone number that is answered, on a page that is easy to find
• Names and faces of the people who will do the work
• Prices, or at least the shape of a price and what moves it
• What is not included, said before they have to ask
• Evidence in a form they can check, rather than logos and adjectives
• Some sign the business is currently operating, which for most sites means something published recently

A page that handles those calmly outperforms a beautifully art directed one that avoids them. Most conversion problems we are asked to fix turn out to be trust problems wearing a design costume.

How we would sequence a rebuild

If you are doing this in order, and the order is most of the benefit.

1. Write the decision for each page, in one sentence, and delete what does not serve it
2. Write the objection for each page, and answer it in the copy rather than the layout
3. Design in the browser on a phone, and review it on a mid-range device rather than a resized window
4. Set a performance budget and hold it at build, not at launch
5. Instrument the form so a lost submission is visible, then watch what people do before they leave
6. Change one substantial thing at a time, and write down what you expected to happen before you change it

Questions people ask about this

What is conversion focused web design?

It is designing each page around one decision a specific visitor needs to make, then building the argument that moves that decision: the claim, the evidence, the objection answered, and one clear next step. The layout, copy, speed and form design all follow from that decision rather than from a template or a style preference.

Does page speed really affect conversion?

Yes, and the mechanism is more ordinary than the usual framing suggests. A slow page gets abandoned before the argument is read, and a page still shifting as it loads produces mis-taps that read as a broken site. The Core Web Vitals thresholds describe loading, responsiveness and visual stability, and they are a reasonable definition of good enough.

Should I A/B test my website?

Only if you have enough conversions for a result to mean anything within a sensible period. Most business sites do not, and calling a winner on a handful of enquiries produces confident decisions built on noise. With modest traffic, watch recordings, read the sales calls, change one substantial thing at a time, and treat the outcome as a judgement rather than a measurement.

How many calls to action should a page have?

One primary action, repeated as often as the page length needs. Secondary links can exist, but they should look secondary. Two actions of equal visual weight ask the visitor to do your prioritisation for you, and in our experience the common outcome is that they do neither and leave to think about it.