Why I May Ditch Grammarly

Forty Toggles for What You Say. Zero for How You Say It.

I’ve used Grammarly for years. I pay for Premium and have recommended it to people who ask. And I’m about to cancel, because the product promotes the exact thought-terminating language I spend my professional life dismantling.

That sentence should concern Grammarly’s product team, because I’m not a casual user who drifted away. I’m an active subscriber who writes daily — articles, books, a podcast — and every time I open a document, Grammarly’s rewrite suggestions read like a masterclass in stripping accountability from meaning.

Here’s the core problem: Grammarly’s rewrite engine systematically converts active verbs into nominalizations. I write “the committee decided.” Grammarly suggests “the committee’s decision.” I write “we analyzed the data.” Grammarly offers “our analysis of the data.” Every suggestion strips the actor out of the sentence and hands agency to an abstraction.

Grammarly’s AI thinks this is an improvement.

Grammarly’s AI is wrong.

What Nominalization Actually Does

When a writer converts a verb into a noun — “decide” into “decision,” “analyze” into “analysis,” “implement” into “implementation” — the sentence loses two things simultaneously. It loses the actor (who decided?) and it loses the action (deciding is alive; a decision sits on a shelf). The resulting sentence sounds more formal. It also says less.

“The implementation of the strategy by the organization” contains zero actors doing anything. Five people sat in a room and chose to execute a plan — and the sentence erased all of them. Grammarly calls this “clarity.” Professional editors call it bureaucratic camouflage.

The “-ing” Problem

Grammarly flags present participles — “-ing” words — as though they represent a stylistic weakness. “Analyzing,” “implementing,” “deciding” — these are verbs doing their jobs. They carry action. They anchor agents. They keep sentences alive.

When Grammarly suggests I replace “the team is analyzing” with “the team’s analysis,” it asks me to freeze a living process into a static object. The verb was working. Grammarly killed it and filed the body under “improved readability.”

I write with a specific set of editorial principles. People act; abstractions don’t. Verbs carry the weight; nouns describe things that exist. When someone reads my work, they encounter human beings doing things — not “processes” that “occur” and “implementations” that “proceed.” Grammarly pushes me in the opposite direction every time I open a draft.

Forty Toggles for What You Say. Zero for How You Say It.

Grammarly does offer a preferences panel. I spent time in it, hoping to solve the problem at the settings level. Here’s what I found.

The panel lists over forty toggles. A significant cluster addresses inclusive language: age bias, family bias, gender bias, LGBTQA+ bias, race and ethnicity bias, human rights language, language related to human slavery, disability-related metaphors, support of Ukraine, and politically incorrect language. Each subcategory gets its own toggle with its own explanation.

I have no objection to any of these. They reflect real editorial considerations, and writers should have granular control over how they handle them.

But notice what Grammarly’s engineers chose to build granular controls for — and what they didn’t. The product team carved out nine separate toggles for subcategories of inclusive language. They gave writers fine-grained authority over the social dimensions of word choice. Meanwhile, the structural linguistics of clarity — nominalizations, agent deletion, verb-to-noun conversions, stop-words — received exactly zero toggles.

The preferences panel does include “Avoid passive voice” (which I keep on) and “Rewrite sentences for clarity” (which generates the nominalization suggestions I’m complaining about). But “Rewrite sentences for clarity” is a single on/off switch that controls the entire rewrite engine. Turning it off kills the good suggestions along with the bad. There’s no way to say “rewrite for clarity, but never suggest converting an active verb into a noun form.” There’s no stop-word field. There’s no “prefer active constructions” checkbox.

Grammarly’s product team invested surgical precision in controlling what writers say and left how writers say it at the sledgehammer level. That priority tells you something about where the engineering attention went. And it tells professional writers that Grammarly doesn’t consider structural prose quality a first-class concern.

What I Actually Need (and Can’t Get)

Here’s what I want from Grammarly that the product doesn’t offer:

A stop-word list. Let me specify words I never want suggested in a rewrite. “Emerge.” “Emergence.” “Evolution.” “Evolutionary.” I have fifteen of these. Most professional writers have their own set. A text field would solve this for millions of users who know their own style.

A nominalization filter. A single toggle: “Do not suggest converting active verbs into noun forms.” One checkbox. Grammarly’s engineers could ship it in a sprint. It would immediately improve the product for anyone who writes professionally.

Directional style controls. Right now, Grammarly’s tone settings operate at the altitude of “formal vs. informal” and “friendly vs. authoritative.” These are billboard-sized categories. Working writers need street-level controls. “Prefer active verbs over nominalizations.” “Prefer concrete subjects over abstract subjects.” “Flag passive constructions that delete a named agent.” These distinctions matter, and Grammarly ignores them entirely.

A learning mechanism that actually learns. I’ve dismissed the same suggestion types hundreds of times. Grammarly doesn’t track dismissals as a pattern. It doesn’t notice that I reject nominalization rewrites every single time. It doesn’t adapt. Every session starts fresh, and I fight the same battle with the same tool that supposedly knows my writing style.

The Competitive Landscape Has Changed

I now use Claude — Anthropic’s AI — as my primary editorial tool. I gave it nine specific editorial principles, and it follows them. It knows my stop-words. It knows I want actors in the subject position. It knows I reject nominalizations. It never suggests “emergence” or “evolution.” It catches the violations I care about and ignores the ones I don’t.

I built in an afternoon what Grammarly hasn’t built in a decade of product development: a style-aware editorial layer that respects the writer’s own rules.

Claude has limitations. It doesn’t integrate with my browser the way Grammarly does. It doesn’t run passively in the background catching typos and punctuation errors. Grammarly still does those things well, and I don’t discount them. But the rewrite engine — the feature Grammarly markets most aggressively — pushes my prose in a direction I spend hours correcting.

When a premium feature creates more work than it saves, the subscription math stops working.

What This Means for Grammarly

The writers most likely to cancel Grammarly Premium are the writers who care most about their prose. Casual users accept suggestions without evaluating them. Professional writers — the subscribers who scrutinize every recommendation — hit the same wall I hit: Grammarly’s model of “good writing” is institutional, nominalized, and agent-free. It’s the voice of corporate memos and academic abstracts.

Grammarly’s team trained the rewrite engine on a corpus that rewards formality over clarity, abstraction over specificity, and institutional voice over individual voice. That training baked a specific ideology of language into the product — one that equates complexity with quality and formality with professionalism.

That ideology is costing Grammarly subscribers.

The fix isn’t complicated. Give professional writers granular style controls. Give us stop-word lists. Give us a nominalization toggle. Let us teach Grammarly what we actually want instead of forcing us to fight what Grammarly thinks we need.

Or don’t — and watch the writers who care most walk away first.