04-01-19
On its 10th anniversary, Grammarly looks way beyond grammar
The service that began by checking prose for glaring mistakes increasingly wants to help its nearly 20 million daily users do more than simply avoid errors
Plenty of technology companies give away stickers of the sort their fans can slap on the back of a laptop. But the ones available for the taking in the reception area at Grammarly’s San Francisco office are distinctly its own—willfully low-key and thoughtful rather than brash and boastful. (If you know of another tech company that has promoted itself by quoting Anaïs Nin, please let me know.)
Being low-key and thoughtful is a logical tone for a company that is in the business of helping people fine-tune their written words, whether they’re meant for a business document, school paper, or social media post. That is what Grammarly has been doing for a decade, since its founding on April 1, 2009, under its soon-abandoned original name of Sentenceworks. But it’s also how it wants to run its business, which—rather than moving fast and breaking things—waited six years before offering a free version and another two before taking on outside funding.
Like every tech company, Grammarly sees itself as having a higher calling. “It’s important to write well; it’s important to write accurately,” says Brad Hoover, the company’s CEO. “But fundamentally, the problem we’re solving is so much bigger. Communication is something that most of us do from an early age, and it both binds us together as people and tears us apart. And a tool set that helps people bridge those gaps, and connect those dots, and communicate what they mean, really does have the promise to fundamentally change how the world works.”
Grammarly’s aspirations may be lofty, but it’s also adept at the mechanics of building a product and turning a profit. In 2017, General Catalyst managing director Hemant Taneja led the $110 million series A investment round that marked Grammarly’s first intake of outside capital. In recent years, says Taneja, many a startup has claimed to be an AI company simply because it’s trendy. Grammarly, by contrast, “is a business that from the beginning was truly about natural language processing and machine learning. And [it] was working, and it was at scale.”
That
scale continues to grow. Grammarly is celebrating its 10th birthday by
announcing that it’s on the cusp of reaching 20 million daily active users,
including both users of the free version and those who pay $30 a month (or $140
a year) for Grammarly Premium or $15 per user per month for Grammarly Business.
That’s up from 15 million last October and just 1 million at the end of 2015,
the year it introduced its free version. The company is an uncommonly effective
direct marketer; even if you’ve never tried its service yourself, there’s a
pretty good chance you’ve been exposed to it on YouTube. And even if you hit the
“Skip Ad” button as fast as you could, enough viewers have paid attention that
YouTube rated Grammarly’s spot as the most
effective “TrueView for Action” ad of 2018, based on reach, clicks,
and engagement.
Over
the last year or so I’ve tumbled through the Grammarly marketing funnel myself,
going from seeing its ads to trying the freebie version to paying for the
Premium edition. Now I run some of my Fast Company writings—including
this article—through the service. About 40% of its suggestions, I consider but
ultimately reject. Another 40% pinpoint weaknesses I already know my writing
has, but have failed to eliminate from a first draft. And 20% identify issues I
might never have noticed without Grammarly’s assistance.
Now,
perhaps you think that I should feel embarrassed admitting I’ve started using
Grammarly, given that I’m in my third decade of writing for money—as if I were
a pro bicycle racer waxing enthusiastic about my new set of training wheels.
But it’s not that it’s transformed my prose. I just find it helpful as a sort
of linguistic Jiminy Cricket that watches me at work, steers me in the right
direction, and points out when I slip up. Grammarly is only capable of doing
that because it’s already gone well beyond its original mission of flagging
grammatical and punctuation glitches. And there’s more to come.
IN THE BEGINNING
The
Grammarly story begins not in San Francisco but Kiev, the capital of Ukraine,
where Max Lytvyn and Alex Shevchenko founded a company called MyDropbox. No
relation to Dropbox—which didn’t come along until
later—their startup detected plagiarism in academic papers. MyDropbox ended up
being acquired by Blackboard, an educational services company; a version is
still available under the name SafeAssign.
Lytvyn
and Shevchenko’s next idea for a company was inspired by a question left over
from their MyDropbox experience: Why do students plagiarize?
On a superficial level, at least, the answer was obvious: because effective
written communication is hard work. Which led the entrepreneurs to ask
themselves another question: Instead of detecting purloined words, why not help
people who want to write better?
The
notion of using a computer to check grammar was not new. (Old-time wordsmiths
will recall Grammatik, a utility that debuted in the 1980s
and was folded into WordPerfect in the 1990s.) But Lytvyn, Shevchenko, and
Dmytro Lider, another MyDropbox veteran, began their new project in an
opportune era. They could check grammar in the cloud—where available computing
power is nearly limitless and rules can be updated on the fly—rather than on
somebody’s PC. And advances in AI meant that they could analyze text with a new
level of sophistication.
Though
based in Kiev, the service was about English grammar from the start; to this
day, the company has no plans to launch in additional languages. Lytvyn says
that the fact its creators learned English as a second language was an asset.
“When you communicate in your native language and your native culture, it is
easy to assume that everybody around you understands you perfectly,” he
explains. “But when you have to do it with a culture and language that aren’t
your own, you pay more attention to the challenge.”
At
first, the new company’s founders called their service Sentenceworks and aimed
it at the education market, which they knew well from their plagiarism checker.
(Their new service could scan for uncited use of other people’s words, too,
and still does.) Then they emphasized the student
audience even more with a new name, EssayRater. It wasn’t until May 2010 that
they landed on a moniker that made it clear what the service was about:
Grammarly.
Another
landmark moment happened later that year when Lytvyn received a call from
Hoover, who had spent six years as a venture capitalist for Bay Area-based
General Catalyst. “I don’t even know how Brad found my personal phone number,
but he called me on my cell phone, completely unexpectedly,” Lytvyn remembers.
Hoover, who had been looking for ways to improve his writing and had considered
hiring someone to provide advice, had stumbled on Grammarly: “I was blown away
by the quality and just instinctively reached out to the founders,” he says.
Hoover
“started talking about the product and got so excited about it on the call,”
says Lytvyn. “I remember that he said, ‘Oh, I’ll come by next week.'” That was
quite a trek—Lytvyn and Shevchenko were based in Toronto at the time.
The
three bonded, and Hoover came on board as a consultant. Within a few months,
Lytvyn says, it was clear Hoover should be Grammarly’s CEO. Rather than nudging
Lytvyn and Shevchenko out of active involvement in the company they’d started,
he convinced them that they, and Grammarly headquarters, should relocate to San
Francisco. Lider stayed in Kiev; all three founders are still
with the company, which now has 120 staffers in Kiev, 70 in San Francisco, and
a handful in New York.
THE BOOTSTRAP ROUTE
Despite
naming a VC as CEO and putting down stakes in San Francisco, Grammarly did not
attempt to morph into a classic venture-funded startup. Instead, it continued
to bootstrap itself, offering a paid service rather than pursuing user growth
first and figuring out monetization later. From 2010 on, it was cash-flow
positive.
Grammarly
started with the most basic of interfaces: You pasted your text into a
browser-based window, then pressed a button to check it. That was sufficient
for its original primary purpose of reviewing student papers, but didn’t
exactly encourage users to treat the service as a core part of their writing
workflow. Over the years, the company has brought its tools to most of the
places people sling words, with browser extensions, Microsoft Office and Google
Docs add-ins, and keyboard apps for iOS and Android.
Along
the way, Grammarly transcended its academic origins, finding a growing market
in users who applied it to personal and professional writings. Indeed,
companies were eager to adopt Grammarly before the company was quite ready for
them. “We saw somebody buying 5 or 10 Premium accounts within just minutes on
the same credit card,” says Lytvyn. “And [our] fraud detection system flagged
it as a fraudulent purchase because it didn’t fit the pattern.” It turned out
it fit a new pattern: companies rolling out Grammarly to multiple employees, a
scenario the company then formalized with a version called Grammarly Business.
As
successful as Grammarly was at finding people who were willing to pay for the
service, it was obvious that there were even more who might find it valuable
but weren’t about to plunk down any money—at least until they’d lived with it
for a while. So for several years, the company mulled over the idea of
introducing a free version. The move had its risks: Grammarly already had
plenty of paying users, some of whom might have opted to downgrade to free
accounts.
Moreover,
Grammarly didn’t want to do the things that many services do to wring revenue
out of non-paying customers. “We felt very strongly about the need to protect
our users’ privacy,” says Hoover. “And that’s one of the reasons that we chose
to keep that paid subscription model rather than moving to a model where we
would have sold or rented user data, which we do not do, or make money off
advertising.” (The company’s one brush with privacy-related controversy came
last year, when a Google researcher discovered a vulnerability in the company’s browser extensions—quickly
patched by Grammarly—which could have leaked text you entered into its online
editor.)
In
early 2015, Grammarly finally went freemium, introducing a free version with a
subset of the 400-plus language checks currently available in the paid
incarnation. As it hoped, giving all comers a taste of the service for free
resulted in more people choosing to pay for the fuller experience.
“Word-of-mouth growth really, really took off,” says Hoover. “That was an
inflection point for us.”
Two
years later, Grammarly had more than 10 million users and was finally ready
to take on outside investment, which came from
Hoover’s former firm, General Catalyst as well as IVP and Spark Capital. When
Hoover approached General Catalyst to make a pitch for funding, “I had no idea
what the founders and Brad had done with the business,” says Taneja. “And
that’s when we started talking about what their overall vision was. It was an
unusual series A investment for me—I’m usually very focused on early-stage
investing. Given all the dialogue we had, I bought into the mission of the
company.”
According
to Hoover, the funding round was less about the cash—which he says Grammarly
didn’t need—than the advice, connections, and raised profile that investors can
offer. “A big piece of how we operate here is looking to learn from others and
never assuming that we have the answer,” he says. “Being able to talk to others
who have been through similar challenges and learn from them—so that we can
make new mistakes rather than old mistakes—is really a critical aspect of our
success and how we build the organization.”
That
organization has been expanding rapidly: The current headcount of 200 is up 40%
in a year, with job listings posted for another 37 positions. That’s still far
from gigantic—Hoover says that Grammarly could theoretically be a 1,000-person
organization if unfettered growth were the goal—but the company will need the
additional brainpower to keep its momentum growing. Even if it’s the clear
leader in its category, it increasingly competes with two of AI’s dominant
forces: Microsoft and Google.
In
2017, Microsoft encroached on Grammarly’s turf by giving Word an AI-powered “Editor” feature that
underlines weak or unnecessary words (begone, “maybe” and “really”). Google
introduced its own grammar checker for Docs in February. A
Word or Docs user who found value in such built-in tools might never get around
to investigating Grammarly, though Hoover says that the service is most popular
for personal communications—such as email and social media posts—rather than
word processing.
Investor
Taneja acknowledges the risk of tech giants creating features that make
Grammarly feel redundant—but also maintains that Grammarly can do things that
functionality built into a particular app cannot. “When you think about all the
places that we end up doing written communications now, it’s not just in
Microsoft, and it’s not just in Google,” he says. In Grammarly, “you have a
cloud-based engine that is getting smarter every time you write, no matter
which environment you’re writing in. It naturally lends itself to be an
independent, enduring company.”
Taneja
may be right. But if some of Grammarly’s existing basics are destined to become
commoditized, the company needs new, advanced features that a critical mass of
people will pay for. With that in mind, the company has spent the past couple
of years developing an AI-infused tone checker, its most dramatic step yet
beyond its roots in grammar.
The
new feature, Lytvyn says, was inspired by the fact that “emotion and tone are
essential parts of communication,” but tougher to master in written form than
with spoken words. “At one point, Alex, my cofounder, and I were working on a
presentation about our vision for our team,” he recalls. “And to make it more
energetic and motivating, we just overloaded it with expletives and high-energy
words. But what ended up happening was we just scared the team. Because
everybody perceived it as ‘the sky is falling,’ and it was totally not our
intention.”
Grammarly
has been training its algorithms to identify the flavor of a piece of text,
whether it comes off as curious, optimistic, urgent, or concerned. It does so
by analyzing words and phrases: A word like “alarming,” for instance, can be .
. . well, alarming, while “unexpected” is not freighted with the same tension.
Then
again, tension could be what you wish to convey. “This feature is not trying to
force a writer to use a specific tone,” says Lytvyn. “It’s more designed to
create self-awareness and to help a writer strike the right chord. Sometimes
you may want to communicate dissatisfaction and may use this tool to even make
it stronger.”
Accurately
gauging the tone of a piece of text is a far greater challenge than spotting
misplaced commas and passive phrasing. Suggesting revisions to achieve a
desired effect is even more ambitious. So Grammarly will implement its new
functionality in chunks rather than one big bang. Later this quarter, the
company plans to add the ability to assess tone. Eventually, it intends to let
Grammarly propose changes. It’s still working on the interface for all this,
but it may not look so different from the service in its current form, with
squiggly underlines calling your attention to places where it has advice.
If
Grammarly can pull off its tone checker in a form that’s useful, it will be
quite an accomplishment. Hoover, however, is careful not to sound too giddy.
Instead, he’s sticking to the humble, methodical perspective that he says is
core to the company’s identity. “Where we’re at, as exciting as it is, is just
a starting point,” he says. “We’re a percent of the way to the world’s 2
billion English speakers. We have a very, very long way to go.”
No comments:
Post a Comment