01 / URL analysis
From a link to a full ad brief.
Paste a product or landing page. We fetch it, read it the way a media buyer reads a page they are about to spend money against, and hand back everything worth knowing before the first pound goes out.
Ad readiness
78/ 100
Pixel present and firing, price and offer both visible above the fold. Held back by thin social proof and a CTA that repeats the brand name rather than the action.
Tracking status
OpportunityHigh
Landing page issues · 3
No Conversions API alongside the browser pixel.
Signal loss on iOS will understate ROAS and starve optimisation. Add CAPI before scaling spend.
Reviews sit below three screens of copy.
Paid traffic lands cold and leaves before reaching them. Move one review block above the fold.
Delivery terms are only on a linked page.
State free UK delivery in the buy box; it is a purchase objection, not a policy detail.
Illustrative report · every field is one the analyser returns
02 / What you get
Nine outputs from a single URL.
Zero to a hundred, with a written reason covering tracking setup, landing page quality for paid traffic, offer and CTA clarity, price visibility and social proof.
Meta Pixel, Google Analytics and Google Tag Manager, detected from the page’s own script tags rather than asked about, plus what to do where one is missing.
Three, each rated critical, warning or minor, and each with a fix written for paid traffic. Not generic SEO advice — the requirements are different.
High, medium or low, with the reasoning: likely ROAS range, audience size and how competitive the category looks from what the page shows.
The single strongest message to lead every ad with, why it will earn, the audience profile behind it, and which framework fits — AIDA, PAS, JTBD or direct response.
Meta Feed or Story, each with a hook, headline, body, CTA and a confidence score based on message-market fit, tracking readiness and offer clarity.
Primary and secondary channels with the rationale. Meta is what we publish to today; where Search or Shopping would suit the product better, the report says so.
Three themes carrying commercial intent, for the search side of the plan and for naming what the audience is actually looking for.
The red flags that waste budget rather than merely underperform: a missing pixel, no price anchor, no trust signals, a page that cannot convert paid traffic.
03 / Under the hood
What the model is actually given.
The scraper fetches the page and pulls out the parts that carry a claim: the title and description, every heading, the opening body copy, the words on the buttons, the alt text on the images, and any structured product data the page publishes about itself.
That goes to Claude with a system prompt written as a brief for a media buyer rather than a content auditor, and the model returns structured JSON — not chat. Every field is parsed and shown as itself, so there is no free text pretending to be a finding.
The important constraint is what is not sent. Price is taken only from structured product data, never from a figure matched somewhere in the HTML, because a number lifted off a page is as likely to be a crossed-out anchor or an unrelated product as it is to be the price. If the page does not publish one, the ads do not state one.
Read from every URL
- →Page title and meta description
- →Headings, H1 through H3
- →Body copy, first 500 words
- →CTA button text, up to 20
- →Image alt text, up to 20
- →Open Graph metadata and images
- →Schema.org structured data, up to 3 blocks
- →Meta Pixel, from connect.facebook.net or fbq()
- →Google Analytics and Tag Manager containers
Assembled context capped at 3,000 characters
Put the next campaign in motion.
Start with the product page. Review the campaign before anything goes live.
Start with a URL