Checking landing pages before a paid campaign

Campaign pages are often built outside the main site's release process — a page builder, a new template, a separate domain — and skip the checks the rest of the site gets.

The traffic is mostly mobile and arrives cold, so loading speed and layout stability decide whether the visit survives the first seconds.

Measurement is part of the product being bought: a tag that does not fire makes the campaign unmeasurable, and one that fires twice makes it look better than it is.

Each of these is observable from outside, before a single click is paid for.

Every landing URL used in the ads should answer 200 directly. Tracking parameters and shortened links often add redirects, and each hop costs time on a mobile connection.

A redirect chain of several hops is worth collapsing; a loop is a campaign that sends every click to an error.

Pages built on a separate domain or subdomain need their own certificate and HTTPS redirection checked, not assumed from the main site.

A campaign page left with a staging `noindex` is fine for search — most landing pages do not need indexing — but a staging password is not.

Scan the exact URLs the ads will use, parameters included, since a rule that matches the clean URL can treat a parameterised one differently.

Measure the page under a mobile profile: largest contentful paint, layout shift and total blocking time are the laboratory signals that predict a bounced visit.

Page weight and the number of requests are the usual causes on a builder-made page, with oversized hero images first.

Third-party scripts added for the campaign — pixels, chat widgets, heatmaps — each add requests and main-thread work.

Form fields should declare their type and autocomplete purpose, so a phone offers the right keyboard and can fill them.

The largest image on a landing page is usually its hero, and loading it lazily or late delays the largest contentful paint the visitor waits for.

The tracker inventory lists which analytics and advertising tags the page loads, which is how a missing campaign tag is found.

The same tag loaded twice — directly and through a tag manager — doubles recorded conversions and is reported as a duplicate.

Where consent is required, tags that set cookies or send data before any choice is made are reported as loading before consent.

On shops, purchase event names observable in the markup are reported; a tag's presence alone never proves its events are configured correctly.

Fix what breaks the visit first: errors, loops, missing HTTPS, a page that renders nothing on a phone.

Then fix what distorts measurement: duplicate tags and tags firing before consent, since both make the campaign's numbers unreliable in opposite directions.

Speed findings come next, starting with the largest element and the heaviest resource.

Rescan once the fixes are in and keep the report: when campaign numbers look strange later, it shows what the pages were like on day one.

Where a consent mechanism is present, the audit reports which consent management product is in use and whether accept, reject and preferences controls are present in the served HTML.

A reject or manage-preferences control that is not observable in the markup is reported as such, since a banner without a real refusal option changes what the consent is worth.

Tags firing before any interaction are evaluated against that mechanism, which is what makes the before-consent finding meaningful rather than a blanket warning.

Campaign pages built on a separate tool often carry their own banner, or none; checking each landing page separately is the only way to know.

Whether a given banner satisfies the law where the campaign runs remains a legal question the audit does not answer.