The Waffle-o-Meter.
A free corporate jargon checker.
Paste your copy, or scan any page. You get a Waffle Score out of 100, a breakdown of what is dragging it down, and your five worst sentences quoted back with a plain-English alternative for each. Higher score means more waffle.
-
What is dragging the score
Your worst offenders
Every flagged term, and what to say instead
Can AI actually read this page?
Waffle is a readability problem. This is a plumbing problem. Both stop you being cited.
How the score is calculated.
No vibes. Seven measures, fixed weights, same maths every time. Here is exactly what each one looks for, so you can argue with it.
Buzzwords & jargon
Weighted matches against a curated lexicon of roughly 120 terms and phrases - synergy, leverage, best-in-class, seamless, holistic, move the needle. Terms are weighted 1 to 3 by how reliably they replace meaning rather than carry it. Scored per 100 words, so length does not skew it.
Sentence length
Mean words per sentence, plus the share of sentences running over 30 words. Under 20 on average reads cleanly. Long sentences are where a point goes to hide.
Claims without proof
Counts absolute claims - leading, unmatched, world-class - against the concrete numbers, percentages and figures in the same copy. Superlatives outnumbering evidence is the single clearest tell of B2B waffle.
Nominalisation & passive voice
Abstract nouns built from verbs (optimisation, functionality, deliverables) and passive constructions. Both let a sentence describe an action without naming who does it.
Paragraph length
Mean words per paragraph and the share running over 90 words. One paragraph should make one point. Dense blocks are also harder for an answer engine to lift as a self-contained passage.
Filler & hedging
Openers and padding that can be deleted with no loss of meaning: in today’s fast-paced landscape, it’s no secret that, in order to, we believe that.
Time to the point
How many words a reader gets through before the copy says what you actually do. Small weight, large real-world consequence.
The AI visibility layer
URL scans add checks the score does not touch: whether OpenAI, Anthropic, Perplexity and Google's AI crawlers are blocked in robots.txt, whether your text is actually in the served HTML or only appears after JavaScript runs, plus schema, headings and llms.txt. Reported separately, because they are a different kind of problem.
It is a rules engine, not a judge. It cannot tell that robust is the correct word in a structural engineering paper, or that a long sentence is doing real work. It reads one page, not your positioning. Treat a flagged term as a prompt to look again, not a verdict. And a score of zero is not the goal - a page can be perfectly plain and still say nothing worth reading.
Frequently asked questions