What the Insight Injector does
The Insight Injector reads a piece of copy and marks the sentences that sound persuasive without saying anything specific: a line like "we deliver outstanding quality" is asserted, and the tool flags that gap. It is built to catch the pattern, because the fix depends on a fact only you have: the actual number, the actual mechanism, the actual thing your product does that a competitor's does not.
This matters because vague praise is also the kind of claim that draws scrutiny under the rules governing advertising in the UK, known as the CAP Code. The Code does not object to confidence. It objects to a claim made without something behind it, so a sentence that reads well on a first pass can still be the weakest sentence on the page.
How the tool spots a weak claim
The Insight Injector works sentence by sentence, checking each claim against a set of patterns associated with generic marketing language: unqualified superlatives ("the best", "the most trusted"), benefit words with no supporting detail ("powerful", "seamless"), and comparisons that name no comparison ("better results", "faster than ever"). Where it finds one, it does not delete the sentence. It marks the claim and prompts you for the fact that would support it.
A worked example
Take the sentence: "Our onboarding process is quick and easy." The tool would flag "quick" and "easy" as unsupported and ask what makes it quick, compared with what. A writer working through that prompt might land on: "Onboarding takes fifteen minutes and needs no account setup." The second version says the same thing the first one was trying to say, but it gives the reader a fact to check rather than a mood to accept.
Reading the output
Each flagged sentence comes with a short note explaining why it was picked up. That distinction matters: a tool that only tells you something is weak leaves you guessing at the fix, while one that names the missing piece (a number, a named comparison, a mechanism) gives you somewhere to start. Not every flag needs acting on. Some copy is deliberately broad because the specific detail belongs elsewhere on the page, and the tool has no way of knowing that context.
Treat the output as a list of sentences worth a second look. A page can pass every flag and still be dull, and a page with three flagged sentences can still convert, if the missing detail genuinely was not needed there.
What the tool cannot tell you
The Insight Injector cannot check whether a claim is true. It can only tell you that a sentence makes a claim without evidence attached to it in the text itself; whether the evidence exists at all, and whether it is accurate, is something only you can confirm. It also cannot judge your specific product, because it has no knowledge of what you sell beyond the words on the page in front of it.
Like any automated check, it can misread a sentence: flagging a fair, well-supported claim because it uses a word the pattern-matching treats as vague, or missing a genuinely empty claim phrased in a way it does not recognise. This is a practical craft check. For a definitive read on what a specific claim is allowed to say, consult the CAP Code and the ASA's own resources directly.
Before you run your own copy through it
The tool works best on copy you have already tried to tighten by eye, because it is built to catch what a first edit misses. First, read our guide on identifying weak copy, which walks through what a vague claim looks like before it has been flagged for you. Then run your copy through the Insight Injector to catch what a memory of the CAP Code will not.