How to build a landing page for Facebook ad traffic, in the order that matters

Most small businesses do not start with a landing page. They start with a website, and the website has a menu, an about page, a gallery, a phone number in the header and a contact form somewhere near the bottom. That page was built for somebody who came looking for you. Ad traffic did not come looking for you: it was scrolling a feed, saw something, and tapped. Sending an interrupted stranger to a page designed for a determined visitor is the most expensive ordinary mistake in a small ad account, and it never announces itself as a page problem. It shows up as expensive leads. Everything below about how Meta's own systems treat destinations was read on Meta's own pages on September 10, 2026, and where nobody publishes a number, this article says so rather than borrowing one.

What actually makes a landing page different

One action. That is the entire definition, and everything else on this page follows from it. A website is a building with rooms, and letting people wander is the point. A landing page is a corridor with a door at the end. Every element on it either moves somebody toward that door or competes with it, and there is no third category. The menu competes. The link to your other services competes. The blog post in the footer competes. On a website those are features, because a visitor who wanders is still a visitor who found you for free. On a page you paid several dollars to put somebody on, a wanderer is a refund you are never getting.

The second difference is the traffic itself. Somebody who searched for an emergency plumber arrived with the problem already formed and the decision half made. Somebody who saw your ad between a friend's holiday photos and a video about dogs did not. There is no question in their head that your page is answering, so the page has to supply the question and then the answer, in that order, in the first screen. It has to do it while they are standing up, holding a phone in one hand, with the feed one swipe away and nothing at stake if they leave.

Before any of this, a landing page is not automatically the right destination. Meta's instant forms open inside the feed, prefill from the profile and cost nothing to build, and for a simple offer answered within minutes they usually collect submissions more cheaply than any page can. Which destination suits which situation is a separate question with a separate answer, and it is written up at aevin.dev/blog/facebook-lead-ads-vs-landing-pages. This article assumes that choice has already been made and it came out on the side of the page.

Four decisions that come before the design

The order matters more than any single tip here, because three of these four are cheap to decide now and expensive to change once a page exists. None of them needs a builder, a designer or a developer. They need a sheet of paper and an hour, and skipping them is what turns page building into three weeks of moving buttons around.

1. 1.The one action, stated as an action rather than a goal. Not more customers. A booked slot on a calendar, a phone call, a form with a name and a number, a quote request with a photo attached. Write the sentence a visitor would say out loud when they are finished: I picked Thursday at ten. If you cannot write that sentence, the page has no ending and what you are about to build is a brochure.
2. 2.Who the page is for, singular. One audience, one situation, one price band. A page that serves homeowners and property managers at the same time has to hedge every sentence it writes, and hedged sentences are the ones people scroll past. If you are running two audiences, the honest answer is two pages, which is much less work than it sounds once the first one exists.
3. 3.The promise the ad is going to make. Write the ad headline before the page rather than after it. The first line of the page is then a repeat of a sentence that already exists, and the whole matching problem below solves itself instead of being patched at the end by whoever is still awake.
4. 4.What counts as a result, and where it gets recorded. Which event fires, where the lead lands, and what tells you afterwards which ad it came from. Deciding this now costs one sentence. Deciding it after launch costs a month of traffic you cannot attribute to anything, and that month is not recoverable.

The fourth is the one that gets skipped, and it is the one that decides whether the next three months are measurable or a matter of opinion. A page that converts and cannot prove which ad fed it is only slightly more useful than a page that does not convert.

The page, part by part, in the order a visitor meets them

A landing page is not a list of best practices, it is a sequence. Each part answers one question, and it only gets to ask its question if the part above it did its job. The table reads top to bottom in the order the page is scrolled, which is also the order to build it in and the order to repair it in.

Nothing in that last column is a design problem. Every one of them is a decision nobody made, which then got made by default at four in the afternoon by whoever happened to be assembling the page.

Matching the ad is a requirement, not a nicety

This is the part most guides file under advice and Meta files under policy. Meta's Advertising Standards state that the products and services promoted in an ad must match those promoted on the landing page, and that the review process may include the specific components of an ad, such as images, video, text and targeting information, as well as an ad's associated landing page or other destinations. Meta's advertising policy basics checklist is more specific about the destination itself: if your ad links to a website, it has to be functioning properly, free of sexually suggestive, shocking, misleading or disruptive content such as pop-up ads, carrying a volume of ads that is not disproportionate relative to the site's content, and clearly accessible and relevant to your ad. All of that sat on Meta's own pages on September 10, 2026.

The practical version is narrower than the policy and worth doing regardless of it. Use the same nouns. If the ad says eighty-nine dollar drain cleaning, the page says eighty-nine dollar drain cleaning, not affordable plumbing solutions. If the ad shows a photograph, that photograph is the first thing on the page. A visitor spends about a second deciding whether the tap took them where they meant to go, and they decide it on words and pictures rather than on meaning. Rewriting the promise into better English somewhere between the ad and the page is one of the few genuinely self-inflicted injuries available in this whole business.

One action, and nothing on the page competing with it

Strip the navigation. Not shrink it, remove it. The header menu exists to send people to other pages, which is the precise opposite of what this page is for, and every link in it is an exit you paid for. The same goes for the footer sitemap, the social icons, the link to your other locations, and the chat widget that opens itself over the form on a small screen. What stays is the logo without a link, and a phone number only if calling is one of the ways this page is allowed to end.

Repeat the action instead of adding a second one. A long page can carry the same button four times, after the headline, after the proof, after the price and at the end, and that is one action offered four times rather than four things to do. What it must not carry is a genuine second option: a newsletter box, a downloadable guide, a call us or fill this in choice presented as equals. Every added option hands the decision back to the visitor, and the visitor is standing in a queue holding a phone.

The form: every field is a toll

Ask for what you will use tomorrow morning and nothing you would merely like to have eventually. A name and one way to reach somebody covers most service businesses, because everything else can be asked by the person who calls, and a question asked in a conversation costs nothing while the same question in a form costs you some share of the people who would have answered it on the phone. The exception is a field that disqualifies rather than describes: a budget band, a timeline, whether they own the property. Those pay for themselves in calls you do not have to make, and they are the only fields worth defending.

Here is where a number would help and where this article is going to disappoint you. The figures that circulate on this subject, that cutting a form from four fields to three lifts conversion by some exact percentage, that the average landing page converts at 2.35 percent, that one second of load time costs seven percent of conversions, all trace back to vendor studies of somebody else's traffic, republished until the original context fell off. On September 10, 2026 we could not find a first-party source for any of them solid enough to print next to your business. What is safe to say is the direction and the reason: every required field is a moment where a person in a queue decides whether your callback is worth the typing. Your own number is two ad sets and a fortnight away, and it will be about your offer rather than about landing pages in general.

Speed, measured against a bar somebody actually publishes

Speed is the one part of this with an objective published standard, and the standard is not Meta's. Google's Core Web Vitals put a good page at a Largest Contentful Paint occurring within 2.5 seconds, an Interaction to Next Paint of 200 milliseconds or less, and a Cumulative Layout Shift of 0.1 or less, assessed at the 75th percentile of page loads and split between mobile and desktop. Those are Google's thresholds for Google's purposes, published on web.dev and read on September 10, 2026. Meta publishes no speed threshold of its own, and none of this decides whether your ad wins an auction. Borrow the numbers anyway, because a stated bar beats a feeling, and measure on a phone over mobile data rather than on the desktop the page was built on.

Meta does hand you one measurement, and it beats any lab test because it is your own traffic. Landing page views count the number of times a click on an ad resulted in the successful loading of a destination, registered either through the Meta Pixel on your site or through the in-app browser on Meta technologies, and Meta says outright that you can compare landing page views to link clicks to understand how many times people may have clicked on your ad but left before your website loaded. Meta names the causes of a gap too: the web page is not fully loading after the click, or the ad has more than one link destination and some clicks went elsewhere. Some gap is normal. A wide one is plumbing, and plumbing gets fixed before copy. How to read those two columns beside each other is at aevin.dev/blog/how-to-read-facebook-ads-manager.

It opens in Meta's browser, not in the one you tested

Here is a detail almost nobody checks, and it changes what testing even means. Meta's own documentation states that when a customer sees a business ad on the Facebook app and taps on it, web links in the ad open in the in-app browser by default, on iOS and on Android. The page most of your paid visitors see is therefore rendering inside Facebook or Instagram rather than inside Chrome or Safari. Meta describes that browser autofilling information from the person's Facebook profile, and it also says people can use the menu inside it to open the link in their system browser instead. So part of your traffic is in one browser and part in another, and a page that behaves in one and misbehaves in the other looks like random noise in your numbers rather than like a bug. Test the way it will be seen: put the link in a post only you can see, tap it from inside the app on a real phone, and go through the form with your thumbs.

One more thing worth knowing while you are in there. Meta states that it may collect and use website content and customers' activity on websites viewed in the in-app browser to improve advertising and other business-related products and services, and it publishes the way out: a disallow for the meta-externalads user agent in your robots.txt file, which Meta says applies to any way it collects that information and not only to that crawler. Read that before copying it. Most small advertisers will want to leave it alone, since the same signal feeds the delivery they are paying for. It is still a decision, and it is easier to make on purpose than to find out about later.

The check that happens before you spend anything

Fifteen minutes, on a phone, before the campaign is switched on. This is not a review of the ad account, which is a different pass on a different schedule and lives at aevin.dev/blog/facebook-ads-audit-checklist. This one is about the page and only the page, and every item on it is something you can do yourself in the next quarter of an hour.

1. 1.Open the page on a phone over mobile data rather than office wifi, and count out loud until the headline is readable. That count is the number that matters, not what a desktop testing tool reports from a data centre.
2. 2.Put the ad and the first screen side by side and read them together. Same words for the same thing, same picture, same price. If you find yourself explaining why they differ, they differ.
3. 3.Try to leave. Tap everything that is not the button. Anything that takes you off the page gets deleted or unlinked, including the logo.
4. 4.Fill in your own form on the phone with your thumbs, in full, without autofill. Every field that irritates you will irritate a stranger more, because they are not the one being paid at the end of it.
5. 5.Submit it. Then check three things: the lead arrived where you said it would, the confirmation screen tells a human what happens next and when, and the event you are optimizing for actually fired.
6. 6.Read the confirmation screen as a stranger. If it says thank you and stops, it is a wasted screen, and it is the one moment when you have somebody's full attention and their permission.
7. 7.Load the page with the images not yet arrived. If the button is invisible until a photograph loads, it is invisible to exactly the people on the worst connections, who are not a rounding error.
8. 8.Do the whole thing again from inside the Facebook app, by tapping a link rather than typing the address, so you see what the in-app browser makes of it.

On tracking, one thing changed under people without their noticing. Meta's help page states that as of January 1, 2024 your ad traffic may have parameters added automatically: a campaign source of Facebook or Instagram, a campaign medium of paid, and the ad, campaign, ad set and campaign tracking group ids. Parameters you set yourself in the URL parameters field override duplicates in the website URL field. Which means the ad ids are probably already arriving at your page, and the job is not to invent a naming scheme but to store what arrives on the lead record.

No leads. What to fix first, and in what order

The rule is dull and it saves weeks: repair the first step where the numbers fall off, and change one thing at a time. A page rebuilt in five places at once teaches you nothing next week, because five changes and one result is not an experiment, it is a story you tell yourself afterwards.

A landing page campaign has five steps you can actually see, and each one has a different owner. The table names the step, the number that shows it, and what a bad answer there means. Read it downward and stop at the first bad answer, because everything below a broken step is being measured through it.

Meta gives you one more instrument for telling the ad apart from the page, it is free, and it is mostly ignored. Ad relevance diagnostics report three rankings against the ads competing for the same audience: quality, engagement rate and conversion rate. The combination is the diagnosis rather than any single ranking. For an ad with average or above quality and engagement but a below average conversion rate ranking, Meta's own reading is that the ad is not producing conversions, and its recommendation is to improve the call-to-action of your ad or post-click experience, or target a higher-intent audience. That is Meta naming your landing page as a suspect in its own interface. The other combination worth memorising is below average quality with average or above engagement and below average conversion, which Meta labels click-baity or controversial and answers with: adjust your ad to more clearly represent the product or service you are advertising. That is the same mismatch read from the other end. Two limits before you lean on any of it. Meta states the diagnostics are not available for ads with fewer than 500 impressions, and that they are not inputs into the ad auction. Read on September 10, 2026.

The second version is where this gets cheap

Your first landing page will not be very good, and that is the correct outcome, because its actual job is to produce the information the second one is built from. What it buys you is a set of numbers with your name on them: your click-through rate, your gap between link clicks and landing page views, your form completion rate, your cost per booked job at your prices. Once those exist, every later decision is arithmetic rather than argument, and the second page usually beats the first by more than any article's advice could.

The building itself is the part that stalls people, and it is also the part that is mostly mechanical once the four decisions above are made. Aevin, our own product at $97, $197 and $497 a month, builds landing pages in the same account it runs the ads from: a published page turns up as a destination you can pick when you set the campaign up rather than a web address you retype from memory, the leads it collects land in the same pipeline with their ad ids already attached, and the images for the creative are generated in the same place, which is the other half of the matching problem. None of this article needs our software, and everything in it works on a page built in any tool or by hand. It is also worth saying plainly that the software proposes changes to a campaign and waits for you to approve them rather than making them on your behalf. What no tool decides for you is the one action, and that decision is the whole page.