URLs and spend, one file. Know what your media budget is funding, at the document level, with the reason attached to every low score. Before the next flight.
14% of spend on pages scoring below 40. Two named advertisers concentrated at the low end.
The pilot runs on the smallest export your ad server already produces: a URL column, a spend column, one flight. No tags. No pixels. No integration. No discovery call. If you can ship a CSV, you can run a pilot.
Every dollar of the flight sorted A through D by measured document quality, not domain reputation. Reveals what share of the buy landed on inventory that would not pass an editorial standard.
Specific URLs, ranked in dollars, each with the passage that scored low and a plain-language reason. The list you take into the DSP the next morning.
Every flag anchored to the passage that earned it. Reproducible on demand. Defensible under finance, brand safety, or client review.
Exposure clustered by supply partner, category, and placement so the change list writes itself. Renegotiate, exclude, or reallocate before the next planning meeting.
Send the sender and the flight size. Reply comes back same day.
CSV with URLs and spend. Secure link, your storage, our storage, your choice.
Every page read, every score carries its evidence, every flag points to the passage.
PDF and a working session if you want one. Read time, twenty minutes. Debate time, longer.
The campaign log is held to run the scoring and produce the report. No sharing, no aggregation into third-party datasets, no training material.
Held for the pilot, plus thirty days for follow-on questions on the report. Deleted after that, or immediately on request.
Serving runs on in-house models on Lens infrastructure. No other model sees your URLs or your spend.
The report is yours. Bring it into your next planning meeting, forward it up the chain, cite it in an audit. It is a Lens document, but you own the copy.
Every score attaches to the passage that earned it and the plain-English reason it fired. Rescoring the same file returns the same numbers. The report is challengeable at the sentence level, not the black-box level — that is the property most vendors will not commit to.
Priced flat by flight size and agreed in the reply to your request. No meter runs while you review the report.
No. Brand safety is a category and blocklist question. Lens is a document quality question. They answer different questions and sit alongside each other.
Yes. Display, video, native, or programmatic guaranteed. If the impression landed on a URL with text, the audit can score the page.
Only if you own the flight. The audit reports on the pages your money touched, not on other buyers' media plans.
The Lens applied research team. Every report reviewed before it ships.
Fill this in and it goes straight to the Lens team. A reply comes back the same day with an ETA and where to send the file.
Two other lanes are open. Get early access to the source-scoring API for an AI product, or talk about verification inside a regulated stack.