Atomic Reach has launched a new product that aims to help marketers produce better content by using artificial intelligence to edit it into material audiences will actually read.

Using machine learning, Toronto-based Atomic AI measures the “readibility” level of any digital, text-based content it produces, which could range from posts on social media and email newsletters to whitepapers or reports. The platform then makes suggestions for ways the content could be changed to match the readibility level of the desired audience.

That could entail making the language less sophisticated, using more emotive phrases or simply adjusting the length of the content. It also analyzes schedules to determine which time of day the content is likely to see the highest engagement.

According to the company’s own research done over the three years that the platform has been developed, matching content’s readibility to the level of the desired audience drives three times as many conversions on Facebook, three times the level of engagement and double the amount of pageviews.

“Readibility is often a blind spot for marketers,” says Kerri Henneberry, director of marketing at Atomic Reach. “Many companies feel like they’re creating content that matches what their audience is expecting and wants to engage with. But we will run a linguistic analysis and show them the readibility across all the different types of content they are producing, that’s when you understand and see how much of a blind spot it is.”

In addition to increasing engagement with existing core audiences, Henneberry says one unnamed client is using the same platform to identify new markets and tailor content to it. Data can be pulled from Atomic’s own database of over two million pieces of content to provide blanket insights for different consumer segments or industry sectors. But it also uses clients’ engagement data from existing content to tailor suggestions to its brand and target audiences.

In terms of competition, Atomic AI appears to be the first offering that combines writing analysis and consumer insight in this way. Aside from standard grammar and spelling checkers, IBM has developed a “tone analyzer” for its Watson AI platform that analyzes the tone in a piece of writing. However, it has yet to be paired with any kind of consumer data that could guide users to which style of writing will resonate best with a certain audience. Though in its infancy and its data set is still growing, Atomic AI also has a “context overlay” that can help creators ensure their content is conveying the tone and emotion that fits with their brand.

In a press release announcing the launch, Atomic Reach founder Bradley Silver cited a report from marketing research firm Gleanster published last year that suggested 25 cents of every dollar spent on content marketing at B2B companies went towards programs that under-performed. A broader study in 2015 by Accenture that covered all categories found one in five marketers had difficulty identifying objectives when it came to creating content. In the same survey, 88% said they had detailed processes and procedures to measure success, but that those processes were inefficient and taking up most of time spent on content marketing.

Atomic AI is available to any enterprise client that produces content, be it client- or agency-side. The platform can be used through a stand-alone mobile app, through CMS, Google Docs and browser plug-ins or extensions for Facebook, Twitter and LinkedIn, with other options currently in development.