He reads people for a living. The job titles keep changing - MTV producer, HubSpot design chief, AI founder, now CEO - but the work underneath has always been the same: figure out what someone is about to do, then build the thing that meets them there.
The pitch meeting where someone first calls customer churn "the most beautiful problem in the room" probably featured Keith Frankel. As CEO of Totango, the customer-success platform, his daily obsession is a deceptively simple question: which customers are about to leave, and what would have stopped them. He leads product, design, and engineering toward turning raw customer data into growth that arrives before the renewal date does, not after it has already slipped away.
He landed in the chair the hard way - by building the thing Totango wanted. Frankel co-founded and ran Parative, an AI customer-intelligence startup that spent six years teaching software to predict who would churn. The models eventually crossed greater than 99% forecasting accuracy. In 2024 Totango acquired Parative's technology and team, folded it into a new AI engine called Unison, and handed Frankel the keys. He arrived as Chief Product Officer and rose to chief executive.
That is the polished version. The stranger truth is that long before any of it, Frankel produced reality television at MTV. It is an odd line on a software resume until you notice what the two jobs have in common: both are about watching human behavior closely enough to predict the next move. Reality TV taught him to read people. Customer-success AI is the same instinct, run at scale and pointed at revenue.
Designers are not artists, and the fact that you view them as such is hurting not only your working relationship, but the quality of the work you receive from them.
At HubSpot, where he ran creative and design from 2011 to 2014 and helped steer the company through its IPO, Frankel described his actual job as translation. Every day he stood between the designers and everyone else in the building - executives who wanted things prettier, marketers who wanted them faster, salespeople who wanted them yesterday. His core argument, delivered loudly and in print, was that treating designers as artists was a category error that quietly degraded the work.
Design, in his telling, is a job of solving problems and making someone's life easier. Aesthetics serve function, not the other way around. It was a contrarian thing to say from inside a creative department, and it became something of a signature. He carried the same idea onto conference stages with a talk called "Escape the Cave," arguing that the experiences a creative collects outside the studio - reality TV included - are exactly what make their work original.
When Frankel gave a CreativeMornings Boston talk, he titled it "Committing to Not Committing." Read his resume top to bottom and the title stops sounding like a paradox and starts sounding like a method. MTV to language-learning UX at Transparent Language, to design leadership at HubSpot, to running product at an adaptive-education company, to founding an AI startup, to the corner office at Totango. Each jump looks unrelated. Each one quietly compounds the last.
Parative was the bet that customer churn is not a mystery but a math problem you have not finished yet. Frankel and his team spent more than six years on it, building models that surfaced data issues quietly breaking go-to-market motions and forecasting future churn with accuracy that climbed past 99%. Bain Capital Ventures and others backed the company. The thesis was proactive, not reactive: catch the customer drifting toward the door while there is still time to walk them back.
When Totango acquired the technology and the team in October 2024, it did not bury the work - it rebranded and amplified it. Parative's IP became the spine of Unison, Totango's AI growth-intelligence engine that predicts churn, flags risk, and points to expansion opportunities. It ships in two flavors: standard models that deploy in days, and custom enterprise models trained on a company's own proprietary data. Frankel framed the move as a way to put enterprise-grade AI in front of revenue teams everywhere, whether or not they were Totango customers to begin with.
Moving from reactive to proactive churn handling helps teams protect revenue before it's too late.
Design exists to solve problems and make a life easier. Pretty is a side effect of getting the function right, never the goal.
By the time a customer churns, the story is already over. The whole game is reading the signal early enough to change the ending.
The experiences you collect away from the work - reality TV, odd detours, dead ends - are what make the work original.
His career began behind the camera, producing reality television at MTV.
He went on record arguing that calling designers "artists" actively damages their work.
He hosts CreativeMornings Boston, the city's chapter of the global creative breakfast series.
Two companies he shepherded were acquired - Firecracker by Wolters Kluwer, Parative by Totango.
Parative's churn forecasting topped 99% accuracy after six years of model work.
His CreativeMornings talk title, "Committing to Not Committing," doubles as a career summary.
Frankel in conversation at the Seattle Interactive Conference with AOL's David Shing, on collaboration and the craft of creative thinking.