A source behind
every borrowed word.

A LaunchWords Ad X-ray compares a sample of a brand’s active Meta ads with its most recent customer reviews, to find what customers repeat that those ads may be missing.

It suggests messages to test. It cannot tell you which ad will sell more.

01 / Collect a defined sample

We request 10 active ads from Meta’s Ad Library and up to 40 most recent Trustpilot reviews through Monid. We verify the brand match and discard inactive ads, duplicate IDs and error rows. A keyword search can miss ads or find similarly named businesses; uncertain matches are not published as verified brand research. Catalog and carousel card text is included; template placeholders are discarded.

The result shows the collection date, the review window and the actual number returned. A cached result is reused for seven days. Gallery examples remain dated snapshots after that window.

02 / Count mentions, not votes

We count unique reviews mentioning each theme. One review can belong to more than one theme, so counts can overlap. Reviews are self-selected and may be invited; these counts are not estimates of all customers. Short windows, small samples and unfulfilled orders affect what you can infer. The overall Trustpilot rating and total review count appear beside the sample. Profiles below 3.5 carry a note about complaint skew. A second review source is planned; version 1 uses Trustpilot only.

03 / Check the words

Every displayed customer quote is limited to 20 words and checked against the exact review it links to. Normalisation handles whitespace, Unicode punctuation and formatting. Words are never stitched together from different passages. Invalid quotes are dropped. Reviewer names and usernames are excluded.

04 / Keep disagreement in view

Each theme can include counter-voices. The model proposes messages using that evidence, and the page labels those suggestions as proposed copy. A missing counter-quote does not establish agreement.

05 / Keep the claim small

A gap means a customer concern may be underrepresented in the collected ad text. We do not analyse every creative, private audience setting, impression, click or conversion. “Running since” is only an ad’s start date. Ad age is not evidence of performance.

06 / Show ads small, and only next to the analysis

Research examples and X-ray pages show small copies (at most 480 pixels) of a brand’s own ads, captured with the sample, so you can see what the analysis refers to. Each is attributed and linked to its original in Meta’s Ad Library. Videos appear as still previews. We never use another brand’s creatives in LaunchWords’ own advertising or social posts, and we don’t embed or hotlink Meta’s pages.

Corrections and removal

Brands and platforms are not affiliated with LaunchWords. We aim to respond to correction and takedown requests, including requests to remove an ad thumbnail, within 48 hours. Request a correction or takedown here.