I’ve been tasked, yet again, with evaluating the ROI of our various SaaS subscriptions, and Grammarly’s enterprise tier is on the chopping block. The sales team swears by it for crafting "clear, compelling, and conversion-ready" copy. The vendor’s case studies are, predictably, full of glowing testimonials and vague metrics about "improved communication." So, I decided to put the core feature—the vaunted "clarity" score—to a very simple, admittedly unscientific test.
I took the raw text from our five highest-converting landing pages from the last quarter. These are pages that demonstrably *work* in the wild, driving sign-ups. I pasted each one, stripped of HTML, into the Grammarly editor. The results were... illuminating, in the sense that they illuminated nothing useful whatsoever. The clarity scores ranged from a "poor" 72 on our best performer to an "excellent" 96 on our third-place page. The second-best converter scored an 82. There was no correlation—positive or negative—between a high Grammarly clarity score and actual, measurable conversion success.
This leads me to a rather uncomfortable question: what, precisely, is Grammarly optimizing *for*? Its own proprietary, black-box definition of "clarity," which seems completely divorced from the practical goal of the text. The tool happily suggested I change active, imperative verbs on our high-scoring, low-converting page to more passive constructions on the lower-scoring, high-converting one. It flagged our most effective call-to-action phrase as "wordy" while praising more verbose, softer alternatives that we’ve A/B tested into the ground. It’s as if the algorithm is tuned for a sterile, academic, or corporate memo standard, not for persuasive, action-oriented marketing copy.
I’m left wondering if we’re paying a premium for a universal style guide that actively works against context-specific best practices. In security and IAM, we talk about "least privilege"; in marketing copy, maybe we need "least genericization." Grammarly seems to want to sand down all our edges until we sound like everyone else. For internal compliance documentation? Maybe that’s fine. For the sharp end of the spear where we compete for attention and clicks? I’m deeply skeptical.
Has anyone else done a similar gut-check against their own performance data? Or are we all just collectively nodding along, accepting the green squiggles as gospel because the vendor told us it represents "quality"? I’d be particularly interested in hearing from folks who've negotiated their enterprise contracts—what service-level metrics did you tie to the renewal that actually reflect business value, not just algorithmic satisfaction?
Just my 2 cents
Just my 2 cents