Click your own ad, right now, on a phone. Read the headline you land on. If it does not repeat the promise you just clicked, you have found the most expensive leak in your account, because every visitor arrives asking one silent question: am I in the right place? Landing pages answer it in the first second or pay for the doubt in the conversion rate.
Message match is the whole first screen
The headline should restate the ad’s promise in nearly the same words. If the ad says “fixed-price boiler installation”, the page says “fixed-price boiler installation”, not “welcome to our heating services”. Match continues through the subheading, the hero image and the button label. In practice this means one landing page per ad group theme, not one page for the whole campaign. Fewer, tighter pages beat one generic page every time it is tested.
Run the check in the other direction too. If the page leads with an offer the ads never mention, test the offer in the ad copy. The ad and the page are one argument, and they improve fastest when edited together.
Speed is a conversion setting, not an IT ticket
Aim for Largest Contentful Paint under 2.5 seconds on a mid-range phone over 4G, because that is the threshold Google treats as good and, more importantly, the point where abandonment curves steepen. The usual fixes are unglamorous: compress and properly size the hero image, remove tag manager bloat, preload the main font, and drop any script that does not directly serve the page. Test with PageSpeed Insights and trust the field data over the lab score.
One page, one job
A paid landing page should offer exactly one action. Remove the site navigation, remove the footer link farm, and keep a single call to action repeated down the page. Every extra choice is a subsidised exit. If you need to serve two intents, buy two pages.
Forms people actually finish
Every field must earn its place. Name, phone or email, and one qualifying question is enough for most service businesses; the sales call can gather the rest. Label fields above the input, use the right mobile keyboards for phone and email fields, show errors inline as they happen, and put reassurance next to the button: what happens after submission and how quickly. For longer forms, break them into steps and put the easy questions first, because a person who has answered three questions rarely abandons the fourth. Send completions to a proper thank-you page, fire the conversion there, and give that page a real job too: what happens next, who will call, and from which number.
Proof belongs next to the ask
Place your strongest evidence beside the form, not in a carousel at the bottom: a named review, an accreditation, a guarantee, a real photo. Specific beats general. “Called back within one working hour” outperforms “excellent customer service” because it makes a checkable promise.
A testing cadence you can sustain
Test one meaningful change at a time: the offer, the headline, the form length. Ignore button-colour folklore. Before starting, check you have the traffic: a page needs roughly a few hundred conversions per variant for a trustworthy read, so low-volume accounts should run bolder changes for longer rather than micro-tests forever. Review results fortnightly, keep a log of what won and why you think it won, and never test two things on the same page at once. When a test wins, roll the winner into every similar page the same week; a documented win that ships nowhere else is a wasted experiment.
Where to start this week
Take your highest-spend ad group, click the ad on a phone, and score the page honestly against this guide: match, speed, focus, form, proof. Fix the worst failure first. Paid traffic is bought one click at a time, which means every improvement here reprices your whole account.