Landing Page Optimization: What Actually Moves Conversion

Landing page advice is full of tactics, button colours, urgency timers, social proof widgets, that produce small effects on pages whose fundamentals are wrong. A page converting at one percent does not have a button problem.

This is ordered by what actually moves the number, which means the interesting tactics come last and the boring diagnosis comes first.

Diagnose before optimising

Three failures look identical in the conversion rate and need opposite fixes.

Traffic mismatch. The visitors are wrong. No page change fixes this. Symptom: high bounce, very short time on page, traffic from a source unrelated to the offer.

Message mismatch. The visitors are right but the page does not address what brought them. Symptom: they read, they scroll, they leave. Time on page is reasonable and conversion is not.

Offer problem. The page is clear, they understood it, and they do not want it. Symptom: they reach the pricing or the form and stop.

Watch ten session recordings before changing anything. Free tools do this and it takes twenty minutes. It will tell you which of the three you have, and that single piece of information is worth more than any list of best practices.

The fundamentals, in order of impact

1. Message match with the source

The single largest factor and the most commonly broken.

Someone who clicked an ad or a link about a specific thing should land on a page whose headline addresses that specific thing, in similar language. Every step of vocabulary drift between the source and the page costs conversions.

If the ad says "invoicing for freelancers" and the page says "streamline your financial operations", the visitor has to work out that these are the same thing, and a meaningful share will not.

2. A headline about the outcome

Not what the product is. What the visitor gets.

The test: could a competitor put their name on your headline and have it still make sense? If yes, it says nothing.

3. The problem, described precisely

Before any features. The visitor should feel recognised, because that is what establishes that you understand their situation and therefore might have a solution to it.

Specificity is the whole mechanism. A precise description of a narrow situation is far more persuasive than a general one, even though it applies to fewer people.

4. One action

Multiple competing calls to action reduce all of them. A page asking someone to buy, book a call, download a guide and follow you on social converts worse than one asking for a single thing.

Repeat the same action several times down the page. Do not add different ones.

5. Objection handling

Every visitor has two or three reasons not to act. If the page does not address them, they leave and take the objection with them.

You find these by asking. Non-buyers who clicked and left, in a short genuine email, will tell you exactly what stopped them, and it is usually not what you assumed.

6. Proof

In descending order of effect: specific results with numbers, named testimonials with detail, case studies, logos, counts.

Vague testimonials ("great product, highly recommend") are close to worthless and may reduce credibility by signalling that better ones were unavailable.

7. Saying who it is not for

Counterintuitive and consistently effective. It signals you are not trying to sell to everyone, which makes the rest of the page more believable, and it filters out people who would have refunded.

Friction reduction

Every field, step and decision costs conversions.

• Remove form fields. Ask for what you need to deliver, nothing more. Each additional field measurably reduces completion.


• Do not require account creation before a purchase where it can be avoided.


• Show the price. Hidden pricing on self-serve products kills conversion. People assume expensive and leave.


• Show total cost early, including shipping or tax. Surprise costs at the final step are a major abandonment cause.


• Enable wallet payments. Apple Pay and Google Pay remove card entry, which is where mobile checkouts lose people. The Apple Pay guide covers implementation.

The technical floor

• Speed on mobile. Slow pages lose visitors before they read anything, and this is invisible in your analytics because those visitors barely register.


• Test on a real phone, on mobile data, not on a desktop browser resized.


• Test the whole flow yourself, with a real card, on a phone. The number of live checkouts that are quietly broken is higher than anyone expects.

Testing with low traffic

Split testing needs volume most small sites do not have. Running an underpowered test and acting on the result is worse than not testing, because you will confidently make the wrong change.

What works instead:

• Sequential changes. Change one thing, wait, observe. Slower, and it does not require statistical significance.


• Session recordings. Ten recordings tell you more about a broken page than a thousand sessions of aggregate data.


• Ask non-buyers. Qualitative, small sample, and consistently the most actionable input available.


• Test big changes, not small ones. A different headline or a different offer produces effects large enough to see with modest traffic. Button colour does not.

What the tactics are worth

Ranked honestly:

• Message match with the source: very large effect.


• Headline about the outcome: large.


• Showing the price: large, on self-serve products.


• Reducing form fields: moderate to large.


• Objection handling: moderate to large.


• Specific proof: moderate.


• Page speed: moderate, larger on mobile.


• Urgency and scarcity: small, and negative if it is obviously manufactured.


• Button colour, exit popups, timers: small, and easily net-negative for trust.

Most optimisation effort goes into the bottom three.

The uncomfortable conclusion

If a page converts badly after the fundamentals are right, the problem is usually the offer or the traffic rather than the page. Optimising a page for an offer nobody wants produces a well-optimised page nobody buys from.

Ask the non-buyers. They will tell you which it is.