Grammarly: Writing the Future of NLP

Unlike most chatbots and AI assistants, Grammarly’s AI does not seek to converse naturally with its user or otherwise mimic human behavior. Yet its ever-increasing “fluency” in the intricacies of human communications surpasses many mass-adopted AI offerings.

Smarter than a spell-checker

Grammarly, a popular AI-powered writing assistant tool, boasts more than 30M daily active users and 50K professional teams that use its services to communicate better. Valued at $13B in 2021, the company has raised $400M in funding from well-respected tech investors like General Catalyst and Spark Capital to advance its natural language processing (NLP) and machine learning technologies.

What started as a simple spelling and grammar checker has evolved into a highly-intelligent, data-rich platform for communication enhancement. Grammarly’s algorithm continually learns from its mass of users, in turn providing more helpful and fine-tuned editing suggestions to individual subscribers. Grammarly’s offering has expanded beyond objective spelling and grammar, providing users with feedback on nuanced aspects of language like clarity and tone (e.g., confidence, formality).

A mission powered by NLP

When Grammarly initially set out with the lofty mission to “improve lives by improving communication,” it ironically relied first on its users to improve its algorithm. Grammarly’s machine learning engineers trained its algorithm to detect errors in written communication, which it could then prompt users to accept or reject while writing text. Grammarly received direct and voluminous feedback on which suggestions were appropriate (accepted) or not appropriate (rejected), so imperfections in the algorithm’s grammar suggestions could teach the program in real-time.

Since Grammarly attracted a wide consumer user base with a simple value proposition of free and seamless text editing, Grammarly was able to (i) quickly scale and start to build brand recognition, and (ii) rapidly upgrade its suggestion power/accuracy based on the numerous feedback points from these numerous users. This substantiates Grammarly’s strategic choice to focus on the consumer market first and worry about monetization (which many SaaS companies find through expansion to B2B markets) later, once the tool had reached a higher quality. Now, with an established and validated product in the consumer market, they are expanding into additional verticals like business and education.

Grammarly boasts impact for users on multiple fronts: from empowering non-native English speakers to communicate clearly and confidently; to enabling business users to communicate succinctly, minimizing costly workplace inefficiencies; to equipping students with intelligent tools to succeed academically.

Commitment to quality: mission alignment or smart business decision?

While Grammarly’s tool has provided an immense benefit to English-writers around the globe, the company has also been able to capture significant value from prioritizing investment in its data asset. A few areas where this is illustrated include:

Quality & user trust: Investing in its data asset allows Grammarly to instill trust in its users, given that users will (a) for the most part, experience a high degree of quality from Grammarly’s offering, and (b) when they do experience an error, assume Grammarly is going to fix errors quickly and accurately (especially in response to a user-submitted error flag). This keeps Grammarly’s brand reputation and word of mouth referrals high, such that it can successfully keep happy users on its platform at scale and win future business.

Talent: Grammarly is able to attract top talent, rivaling that of big tech companies and other tech startups, given its commitment to cutting edge technology and innovation. Rather than branding itself as simply a language or grammar app, Grammarly has successfully positioned itself as an AI-enabled tech company. By investing in a strong data team with machine learning and NLP specialist roles, engineers perceive Grammarly as a great place to continue their career, and Grammarly in turn benefits from top talent to help build an unrivaled product. This allows for more VC funding, more press/accolades and generates a virtuous cycle for company success.

Representation of Grammarly’s tech org operating model:

Challenges ahead

As NLP becomes more ubiquitous, the competitive moat that Grammarly has guarded itself with may start to erode. There are many free and paid offerings popping up today that rival Grammarly’s algorithm, including Wordtune (in the consumer market) and Sapling (in the B2B market). However, quality does matter when it comes to the most discerning users, who most likely take advantage of Grammarly’s more advanced, paid offerings. If Grammarly focuses on fine-tuning its nuanced language features and beating competition on data sophistication (rather than just prioritizing mass adoption), it is likely to continue to win in this market.




Asian Paints: India’s Biggest Data Science Company that Sells Paint


Mercedes Benz: A Race of Data and Machine Learning

Student comments on Grammarly: Writing the Future of NLP

  1. Thanks for your post Steph! I really enjoyed reading it. I would imagine a variety of technologies and expertise goes into building the Grammarly AI, and it needs a large corpus to begin with! It was also interesting to learn that Grammarly has now branded itself as an AI-enabled tech company focusing on grammar checks – I think that’s a clever strategy, and sets it apart from other AI-writing tools such as GPT-3. I’m curious to know whether Grammarly’s current pricing strategy works well and whether a sufficient number of users subscribe to non-free versions of Grammarly.

  2. Really enjoyed reading this post and about how Grammarly works on the backend. I’ve been a user for several years and it’s interesting to think that while users have been using Grammarly to improve their writing that Grammarly has been using users’ errors and corrections to improve their machine learning and recommendations. I also had a similar question as Yifei though around their freemium pricing strategy. I wonder if they’ll stick with that as they continue to grow or if they’ll think of other ways of monetizing their machine learning tools (maybe partnering with Microsoft Office?)

  3. Thank you for sharing! As a user of Grammarly, I cannot agree more with what you articulate in this post, especially to the point that when it comes to grammar, the quality matters the most. I also wonder if Grammarly might look into expanding to other languages in the future!

Leave a comment