The problem this solves
Ad review and landing-page review live in different rooms. The ads team optimizes creative and watches click costs; whoever owns the website ships pages on a separate cadence; and the handoff between them, the three seconds after the click, belongs to nobody. Yet that handoff is where most paid budgets actually leak: the ad promises a specific outcome, the page opens on something generic, and the visitor who was promised one thing and shown another leaves without becoming data anyone reviews.
The mismatch is invisible in every dashboard you own. Analytics shows a bounce, the ad platform shows a click it happily charged for, and neither shows what the buyer actually saw: the hero that loads after a beat, the layout on a 390-pixel phone screen where the CTA sits below three scrolls of filler, the testimonial images that quietly stopped loading last month. Reviewing this properly means experiencing each page the way the paid visitor does, and nobody has time to do that across every high-spend ad weekly.
This mission does exactly that experience, literally: it pulls your highest-spend ads, opens each destination in a real browser, records the scroll a visitor would make, captures the phone view most clicks actually land on, and collects every asset failure and script error along the way. Then it renders the verdict ad by ad.
How the mission runs
- Pull the highest-spend ads and their promises. From your connected ad accounts the mission identifies the top ads by current spend, and extracts what each one promises: the headline claim, the offer, the call to action. This promise becomes the standard its landing page is scored against.
- Record the scroll-through. Each destination page is loaded in a rendered browser and captured as a smooth top-to-bottom scroll recording, the closest artifact possible to sitting behind the visitor's shoulder. The recording shows load behavior, content order, and how far down the page the ad's promise is finally honored.
- Capture the phone view. The same page is captured in mobile emulation, where the majority of paid social clicks actually land. Desktop-designed pages routinely bury their CTA or break their layout at phone width, and this capture is where those failures become visible.
- Check for broken assets and script errors. While each page loads, the browser harvests console errors and failed requests: images that 404, tracking scripts that fail, embeds that error out. A page can look acceptable in a screenshot while quietly failing to measure the very traffic you are paying for.
- Score each page against its ad. Each landing page gets a verdict: does the first screen honor the ad's promise, how many scrolls until the offer appears, does the phone experience survive, and is anything broken. Pages are ranked so the worst promise-to-page gap on the biggest spend surfaces first.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A review of every high-spend ad destination containing the scroll recording, the phone-view capture, the asset and script health check, and a promise-match score per page, ordered by spend at risk. The recordings do the arguing for you: when a page buries its offer, the video shows the burial.
Make it yours
- Run it before scaling any campaign: the review of five pages costs minutes and has veto power over thousands in planned spend.
- Include competitor landing pages found through the ad libraries, so the review shows how your click experience compares with the one your buyer sees an hour later.
- Schedule it weekly during heavy campaign periods and route only regressions to Slack, so the team hears about pages that got worse rather than re-reading stable reviews.
Frequently asked questions
Why recordings instead of screenshots?
A screenshot shows composition; a recording shows experience. Load stutter, content order, the distance between the headline and the CTA, and the moment a promise finally appears are all temporal facts, and the scroll recording is the only artifact that carries them into a review.
What does the promise-match score actually check?
It compares the specific claim in each ad with what the first screen and first scroll of its destination deliver: the offer named, the language mirrored, the action available. A page can be beautiful and still score poorly because it opens on the company story instead of the thing the ad sold.
Does the mission change anything on the pages or in the ad accounts?
Nothing. The browser is read-only, and the ad accounts are only read for spend and creative. The output is a review with evidence, and every decision about pausing ads or editing pages stays with your team.