Video ad landing pages: routing signals, creative delivery and playback

What makes a video ad click different

A click on a search ad usually opens in the browser the person already uses. A click on a video ad usually does not: it opens inside the host app's own in-app browser — YouTube, TikTok, Snapchat, Facebook and Pinterest each ship their own — and that single difference removes or weakens several of the signals a routing rule set normally leans on.

Three things change at the same time:

• Referrer. Depending on the platform and the operating system, the destination sees the platform's own domain, a generic value, or nothing at all. A rule that expects to read the originating page simply never fires.
• User agent. The string describes the host app's webview, not the browser the person chose. Version and engine details are usually narrower than a standalone browser reports, and different app versions produce different strings over the same underlying engine.
• Storage continuity. The webview container starts clean. A returning customer looks like a first-time visitor, because nothing from the earlier session is readable here.

What does survive is the click id the platform appends to the destination URL, plus everything below the application layer: the exit IP and its network type, the device class, the language settings, the time of day. Those are the signals worth weighting on video placements. The mechanics of the webview hop itself are covered in more depth in mobile traffic, webviews and referrers.

Routing signals that still work on video placements

Since the application-layer signals thin out, the practical move is to shift weight onto what the hop does not destroy.

• Whether a click id is present, and whether its shape is right. Each platform appends its own parameter, and each format is recognisable. A click claiming to come from a video campaign but carrying no click id — or one whose format does not match the platform it claims — is a different proposition from one carrying a well-formed id.
• Network origin. Residential and mobile-carrier exits behave differently from hosting and datacenter ranges. This signal is untouched by the webview, which makes it one of the steadiest available here.
• Device and environment coherence. Not the individual attributes, but whether they agree with one another — whether the declared device class, screen characteristics and language settings form a combination that occurs in the real world.
• Geography against campaign targeting. A click landing well outside the campaign's targeted regions deserves a different verdict from one inside it, and on video campaigns this gap is often the single clearest indicator.

The failure mode to avoid is importing thresholds from search traffic unchanged. Search clicks carry richer referrer and storage signals, so a rule set tuned there implicitly assumes they are present. Point that same rule set at video traffic and the missing signals read as suspicion — the outcome is rarely more machines caught, it is more real viewers kept on the safe page. Where to put that boundary is its own subject, covered in tuning a pass rate that is neither too strict nor too loose.

Platform-side fetches around the click

Video platforms fetch the destination themselves, at several moments that have nothing to do with a viewer:

• when the ad or its destination is first submitted, for policy and safety checks;
• when a preview card or link unfurl is generated;
• when media on the page is prefetched so that playback starts faster.

These arrive with their own identifiers and network origins, and they are exactly the traffic a safe page exists for. The practical cost lands on the measurement side: once they are counted as page views, the ratio between platform-reported clicks and recorded sessions stops being interpretable — and that ratio is usually the first thing worth reconciling when a video campaign starts behaving strangely. Reading routing analytics walks through the reconciliation.

Routing the video creative, not just the page

Routing usually gets discussed at page level: which of two pages a visitor sees. On video campaigns there is a second layer that is easy to miss — the video on the page is a module like any other, and modules can be selected per visitor inside the same URL.

That makes three things possible without any redirect:

• Language. Viewers in one market get the cut with their own voiceover or subtitles, another market gets theirs, and neither leaves the link they tapped.
• Continuity with the ad creative. When several video creatives run in one campaign, the page can open with the same footage the viewer just watched instead of a generic brand cut — the click id already identifies which creative the click came from.
• Device fit. A vertical cut for phones, a wider one for tablets and desktop, chosen at load time rather than letterboxed after the fact.

The URL never changes, the address bar stays the same, and every visitor is on one link — the same in-page mechanism used to swap a headline or a hero image, applied to a heavier asset. And because video is the most expensive creative most advertisers produce, the same delivery path carries the anti-scraping and anti-mirroring protection that applies to hero images and copy.

When the landing page is the video

Some campaigns go further and make playback the page: vertical, full-screen, in the rhythm the viewer was already in when they tapped. It turns the click-through into a continuation rather than an interruption — and it brings constraints a text-first page never has to think about.

• Autoplay is conditional. Mobile browsers and in-app webviews permit autoplay only when the video is muted, and some environments decline it regardless. Sound-off has to be the design assumption, not the fallback.
• The poster frame does the real work. Whatever shows before playback begins is, for a meaningful share of visitors, the entire page. It has to read as a complete message on its own.
• The message cannot live in the audio. Captions or on-screen text carry it; sound is an enhancement for the viewers who choose to turn it on.
• Weight is a routing input. Connection quality varies widely across the markets video campaigns reach, and a cut that starts instantly in one market may not start at all in another — which is the reason to treat asset choice as part of the routing decision rather than a fixed property of the page.

What to watch after a video campaign goes live

Four numbers, in this order:

• Platform clicks against recorded sessions. A persistent one-directional gap usually means either platform-side fetches are being counted, or real viewers are being held back. The two look alike in a summary and separate cleanly once split by country and network type.
• Safe-page share, split by placement. If one video placement sits far above the others, the rule set is probably reading that platform's webview as unfamiliar rather than as machine traffic.
• Playback start rate. The share of sessions where the video actually begins. Low values point at the poster frame, the file weight, or an autoplay assumption that does not hold in that environment.
• Per-rule hit counts. Which specific gate is filtering video traffic — rather than an aggregate pass rate that hides which one is doing the work.

Split each of these by placement instead of reading a campaign-level average. Video placements differ enough from one another — and from search and display — that averaging across them mostly hides the thing worth seeing. The wider comparison across source types is in how routing signals differ across search, social and native traffic.

Frequently asked questions