• AI visibility

Guide · AI search

How can an AI writing tool win users when ChatGPT also writes?

By owning a specific writing job that general assistants handle poorly, and making sure the independent lists, reviews and comparisons that AI assistants read say so. Writing is one of the main things people already do inside ChatGPT, Gemini and Copilot, so the assistant a user asks for a recommendation is also your largest competitor. The tools that grow from AI search will, we expect, be the ones with a reason to exist that a third party can verify.

By Underneath Team Guide Updated

The short version

  1. Writing is the most common work use of ChatGPT: 40% of work-related messages in June 2025, and about two-thirds of writing messages ask it to edit the user’s own text, per a study by OpenAI and Harvard economists.
  2. General suites now include writing help: Google (opens in a new tab) cut the price of Workspace Business Standard with Gemini from $32 to $14 per user per month by folding AI into the plan.
  3. Business demand is broad: 95% of B2B marketers told the Content Marketing Institute (opens in a new tab) their organizations use AI applications, and 89% of those use AI content creation tools.
  4. A writing subscriber can be worth more than a typical app user: in RevenueCat (opens in a new tab)’s subscription benchmarks, most AI apps passed $0.63 per install by day 60, twice the $0.31 median across all apps.
  5. The lists AI cites are often written by vendors: in our study, 24.2% of AI-cited numbered “best X” lists with an identifiable publisher ranked that publisher first.

Who uses AI writing tools, and what is a user worth?

Students, professionals and marketing teams, and most pay a modest subscription after starting free.

The audience is wide and young. Pew Research Center (opens in a new tab) found 34% of US adults have used ChatGPT, and 28% of employed adults now use it for work. Among teens, 26% use ChatGPT for schoolwork, up from 13% in 2023, per a separate Pew survey (opens in a new tab). These are the same people who look for paraphrasers, grammar checkers, essay helpers and email assistants. Students are also a core audience for AI research assistants, covered in how AI research tools get found by students.

Businesses buy too. In the Content Marketing Institute’s research, content creation tools for marketing copy were by far the most used kind of AI tool, at 89% of marketers whose organizations use AI, ahead of creative asset tools at 53%. Marketing teams are the natural buyers of brand-voice, long-form and SEO writing tools, and they buy seats, not single subscriptions.

A single user is worth little; a habit is worth a lot. Grammarly’s plans page (opens in a new tab) shows a free tier and a Pro plan at $12 a month, a typical structure for the category. RevenueCat’s data on subscription apps shows AI apps earning more than $0.63 per install after 60 days, matching only Health and Fitness. The same report finds “not enough usage” is the top reason people cancel, at 32% to 47% of cancellations across categories. For a writing tool, the value of an AI recommendation depends on whether the user makes it part of daily work.

How do general AI assistants change the market for writing tools?

They are both the biggest competitor and the place users ask which tool to use.

The OpenAI and Harvard study of ChatGPT messages found writing covers drafting emails and documents but also “editing, critiquing, summarizing, and translating text provided by the user.” Writing is especially common for management and business users, at 52% of their work-related messages. The same study notes writing fell from 36% of all ChatGPT usage in July 2024 to 24% a year later as other uses grew, but it remains the leading work task.

The suites are absorbing basic writing help. Google’s Workspace announcement says Gemini now helps users “summarize, draft, and find information” across Gmail and Docs, for $2 more than Workspace without Gemini. A standalone tool that only rewrites a paragraph is competing with a feature the buyer already pays for. Productivity apps face the same pressure; see how productivity tools compete with built-in AI.

Some independent tools still win at scale. In Andreessen Horowitz’s fifth Top 100 list (opens in a new tab) of consumer AI products, QuillBot and Gamma are among the fourteen “All Stars” that have appeared in all five editions of its web list. Grammarly, which says it works across more than 1 million apps and websites, competes on being present wherever people already write. We infer the durable positions are specialization (academic, legal, technical, brand voice), workflow presence and trust, not raw text generation. Image generators face a similar test; see how specialist image tools hold their ground.

Which questions do writers ask AI assistants about tools?

Use-case, comparison, “free” and trust questions, often naming the general assistant as the alternative.

These example prompts are ours, written to show how students, professionals and marketers phrase writing-tool questions; they are not observed data:

  • Use case: “Best AI tool to rewrite academic paragraphs without changing citations.”
  • Free tier: “Free AI paraphrasing tool with no daily word limit.”
  • Comparison with assistants: “Grammarly vs QuillBot vs just using ChatGPT for editing a thesis.”
  • Team fit: “AI writing tool that keeps our brand voice for a 10-person marketing team.”
  • Privacy: “Which AI writing assistants do not train on my documents?”
  • Integrity: “Is it allowed to use an AI grammar checker on a university essay?”

Integrity questions matter in this category. Pew found only 18% of teens say it is acceptable to use ChatGPT to write essays, while 42% say it is not. A tool that is clear about what it does, editing rather than ghostwriting, has an answer ready when a student or teacher asks. Comparison questions that include “just use ChatGPT” are the most direct test of a tool’s reason to exist.

How does an AI recommendation turn into paying users?

Through a free signup that becomes a daily habit, then a subscription or a team plan.

We infer the path for most writing tools runs like this:

  1. Named. A user asks an assistant for a tool for a specific writing job and gets three to five names, sometimes alongside advice to use the assistant itself.
  2. Free use. The user tries one, usually on a free tier or trial.
  3. Habit. The tool earns a place in the browser, word processor or email client. If usage stays low, the user leaves; RevenueCat’s cancellation data shows how often that happens.
  4. Paid. A paywall on volume, quality or features converts the habitual users; for business tools, one user brings in a team.

RevenueCat reports that across categories, most trial-to-paid conversions happen immediately and that lifetime value rises nearly 60% from month 1 to year 1. Those figures cover mobile subscription apps of all kinds, not writing tools alone, but they show why a recommendation that reaches the right user at the right moment is valuable: the first session decides a lot. The AI answer itself rarely appears in analytics; it shows up as direct visits, branded searches and signups. For more on that measurement gap, see why analytics miss AI visibility.

What decides which writing tools an AI assistant names?

Platforms document little; studies point to “best of” lists, independent coverage and clear, checkable claims.

Documented by the platforms. Per OpenAI (opens in a new tab), a request such as “best paraphrasing tool” is typically turned into “one or more targeted queries” sent to search providers, and Google (opens in a new tab) describes AI Mode running “multiple related searches” across subtopics. Neither says how writing tools are ordered within the answer that comes back.

Observed in studies. Ranked lists carry outsized weight. In Kumar’s (opens in a new tab) data from an AI visibility platform, the “best-of” listicle was the most-cited content format, at about 21% of all citations. Writing tools are a category full of vendor-written “best AI writing tools” lists. In our study of numbered lists cited by AI, 24.2% ranked their own publisher first, and publishers that included themselves put themselves first 92.9% of the time. For software questions, Chen and colleagues (opens in a new tab) found AI search drew 72.7% of its sources from independent “earned” sites, and in our brand entity study, each tenfold rise in independent sites naming a brand went with 4.7 times the odds of a recommendation.

Trust factors specific to writing tools. We infer users and assistants weigh: whether the tool trains on user text, how it handles confidential documents, quality on specific kinds of writing, academic integrity positioning, language coverage and integrations. Quality is not a given: in the Content Marketing Institute’s survey, 12% of marketers using AI for content said its quality had decreased. Specific evidence, such as published samples, independent reviews and clear data policies, gives an assistant something concrete to repeat. For how self-ranking lists fit with fair competition, see can competitors game AI recommendations.

What does a writing tool lose when the assistant names a rival, or itself?

The user often stays inside the assistant, so the lost signup is invisible.

When an assistant is asked which writing tool to use and names others, or suggests it can do the job itself, the user may never search for you at all. New tools start furthest behind: in a test of 112 Product Hunt startups, Sharma (opens in a new tab) found ChatGPT surfaced them in only 3.32% of discovery questions. The reasons a young paraphraser or grammar app gets overlooked are set out in why ChatGPT misses new products. Assistants that search do favor recent pages: in our freshness study, 17.4% to 22.6% of each assistant’s dated citations were under 90 days old, against 6.9% of Google’s top 10. That gives a tool with fresh, specific coverage a real chance, even against older names.

What does GEO involve for a writing assistant or editing tool?

It makes your specific strength findable, consistent and verified by others; it cannot promise a mention.

For a writing product, generative engine optimization (GEO) tends to span six pieces of work:

  1. A sharp entity. Say plainly what writing job you do best, for whom and where (browser, Docs, Word, email), and say it the same way on your site, app store listings, review profiles and social accounts.
  2. Honest comparison pages. Publish fair comparisons with named competitors and with general assistants, including where they are better. Our comparison-page research summary explains what tends to help.
  3. Independent lists and reviews. Get tested by independent reviewers, educators, writing communities and marketing publications rather than relying on your own “best of” page. Our note on how best-of lists shape AI recommendations explains why outside lists count for more.
  4. Public trust pages. State your data and training policy, privacy, security and academic integrity position in plain language, so an assistant asked a trust question can find your answer.
  5. Proof of quality. Publish before-and-after samples, published benchmarks with methods and real customer examples by use case.
  6. Measurement. Put your use-case, comparison, free-tier and trust questions to ChatGPT, Gemini, Perplexity, Claude, Copilot and Google on a regular schedule, and add a “where did you hear about us?” question to signup.

What don’t we know yet about AI assistants sending people to writing tools?

Nobody has published how often AI answers point people to a writing tool, or how many of them pay.

What we know about writing inside ChatGPT comes from a study co-written by OpenAI’s own researchers. The revenue benchmarks come from RevenueCat’s data on mobile subscription apps of all kinds, which may not match web-first writing tools. The Content Marketing Institute heard only from B2B marketers, so students and everyday writers are not represented. Studies of AI citations cover many categories, not writing tools specifically, and the role of self-ranking lists in writing tools is our inference from those studies. We also cannot yet say how often assistants recommend themselves instead of a tool; we have seen no study that measures it.

How should an AI writing tool check whether assistants are sending it subscribers?

Ask assistants about the writing jobs you do best, and note which tools, or which assistant, they suggest.

Run your use-case, comparison, free-tier and trust questions through the main assistants, then line the answers up against signup and free-to-paid conversion data to see which missing mentions cost you subscribers. To plan the independent reviews and published proof that would close those gaps, ask us for a review of your writing tool’s AI visibility. How we help a writing product state its one clear strength and get it confirmed by outside reviewers is set out on the generative engine optimization service page.

Frequently asked questions

Will ChatGPT recommend a competing writing tool?

It can. Asked for the best paraphraser or grammar checker, assistants do list third-party products, and no platform publishes how that list is chosen. Whether an assistant suggests itself instead has not, to our knowledge, been measured.

Do “best AI writing tools” lists on our own site help?

They are cited, but in our study about a quarter of AI-cited numbered lists ranked their own publisher first, and readers and assistants can see who wrote them. Independent lists and reviews carry more weight with buyers, and are harder to dismiss.

Should a writing tool compare itself with ChatGPT on its website?

Yes, honestly. Users ask that comparison, and a fair page that explains where a specialized tool is better, and where it is not, gives assistants a clear answer to quote.

How do we know if AI search is sending us users?

Combine three signals: whether assistants name you for your core questions, signups from AI referrals and direct traffic, and a “how did you hear about us” question at signup. Then follow those users through to paid plans.

Sources

Free strategy call

Some questions are easier to answer about your own business.

Bring the one that matters most. On a free 30-minute call we’ll take a first look at it and send you a short written read afterward.

Ask about my businessHow we work