Christopher Martinez had a report that could tell him what customers were talking about. The trouble was that it told him much the same thing every month. He and Kevin Yang had each run companies, collected support conversations in Zendesk, and attached tags to the cases. The resulting summaries were too vague to change what they built. An organization could diligently receive feedback, carefully file it, and still learn very little. Bureaucracy had performed its usual trick: everything was accounted for except the point.
Martinez and Yang kept talking about this frustration. They wanted to make decisions with customers in mind; their information refused to cooperate. Idiomatic grew out of that practical failure. Martinez’s account of the beginning is refreshingly unromantic. There was no sudden revelation about the future of computing. There was a pile of words, an unsatisfactory report, and two founders wondering whether those words could become useful data.
A founder watches television and sees a prospect
In June 2016, Martinez watched an episode of HBO’s Silicon Valley and recognized a problem he wanted to work on. Pied Piper, the show’s fictional startup, had attracted downloads without persuading enough people to keep using its product. Its founder consulted people who already understood technology, received vague reactions from a focus group, and struggled to explain the software to ordinary users.
Martinez knew he fitted the show’s casting brief rather well: Stanford graduate, Palo Alto resident, working on his second company. He enjoyed the joke. He also found himself wanting to pitch the fictional business. Support conversations, he argued, already contained explanations of what users found confusing. A company needed a way to read those explanations together. His demonstration used an example dataset of San Francisco Muni complaints to show how messages could become categories that informed priorities.
It is a revealing little scene from his own telling. Some viewers watch a startup comedy for escape. Martinez watched it and began mentally qualifying a sales lead. The potential customer’s nonexistence presented a modest obstacle.
Three Stanford degrees, then a different kind of homework
Martinez’s education spans mathematics and computer science, law, and business. His public professional biography lists a bachelor’s degree in Math and Computer Science, a JD, and an MBA from Stanford. Before Idiomatic, he co-founded Glow. The sequence gave him experience building a company before he began selling other companies a way to understand their customers.
The interesting part of that background is the breadth of the questions it puts beside one another. Computation concerns how to process information. Business concerns where to put scarce resources. Legal training belongs to a world where wording and interpretation matter. Idiomatic’s work sits at their intersection: interpreting what people say closely enough that someone else can decide what to do. That connection is a way to read his career, rather than a claim about what any particular class taught him.
By the time he was working on Idiomatic, the assignment was concrete. Feedback needed to survive the journey from an individual message to a decision about a product.
The first announcement came without a funding round
In September 2016, Martinez introduced Idiomatic through a blog post. The team had no plans to raise money at that point, so a conventional funding announcement would have lacked one essential ingredient. He named Ariel Kleiner, Kevin Yang, and himself, and described a business already processing hundreds of thousands of pieces of feedback across several industries.
His argument began with product friction. People could identify where an experience went wrong, even when they could not specify the feature that would repair it. Companies needed to organize those accounts at scale and use them to decide where to spend time. The ambition was to give businesses a practical route to better experiences.
The launch also contained an admission that makes the promise easier to understand: founders made judgment calls, and those judgments could be wrong. Martinez’s explanation located the need for the company in mistakes its founders had experienced themselves. He was offering a tool for an uncertainty he recognized.
“We grew tired of flying blind”Christopher Martinez, 2016
He left people inside the machine
By April 2017, Martinez could describe an approach that involved analysts working alongside automation. Idiomatic sought categories specific enough for a person to understand and act on. A label that merely sounded technical would be of limited use to a support leader trying to explain a recurring problem to an engineer.
He also described the ordinary constraints of a small, bootstrapped business. Immediate sales work competed with longer-term marketing. Early introductions came through professional networks and friends of friends. The team participated in support communities, including Support Driven. These are recognizable founder chores: finding the next conversation while remembering that tomorrow will also need customers.
There was an unexpected development, too. The founders had initially emphasized product feedback, but customers found uses within support operations itself. Martinez said the company used Idiomatic to analyze its own customer interactions, calling the practice “Idiomatic squared.” A feedback company listening to feedback about its feedback product has at least earned the right to enjoy the arithmetic.
The people who hear customers need a hearing
Martinez’s January 2018 advice to support teams began with an act of perspective-taking: understand the product manager’s predicament. Product teams receive competing requests from sales, executives, support, and customers. A support leader’s certainty about an issue does not automatically settle the competition for engineering time.
He proposed bringing evidence with three qualities: quantified, contextualized, and curated. Count the problem. Explain the circumstances. Make the presentation relevant to the people who have to use it. His examples included Upwork, Slack, and Intercom, businesses using customer conversation analysis in different ways. The argument treated support as a department capable of contributing to decisions across a company.
There is a practical courtesy in that advice. Understanding customers requires understanding colleagues as well. A pile of complaints dropped on someone’s desk leaves that person with another pile of work. A clearly described problem gives them something to assess. Martinez’s focus was on preparing the information for that second encounter.
Even the sympathetic listener sees a slice
In an earlier essay, published in November 2017, Martinez had questioned familiar rituals of customer listening. Having engineers answer support tickets could build empathy, but their encounters represented a limited sample. An engineer might fix the problem directly in front of them while a more widespread issue remained elsewhere in the queue.
He extended the criticism to support teams and product analytics. An agent sees particular conversations. Usage data records behavior without necessarily explaining the intention behind it. Predefined tags can constrain what a team notices because someone must anticipate the categories in advance.
His concern was the gap between contact and understanding. The company that can recount a memorable customer conversation may still struggle to describe the experience of its wider audience. This helps explain why his work repeatedly returns to the same question: how does a business hear beyond the people who happened to catch its attention?
Watch the 2½-minute talk ↗The courteous answer can still be no
In March 2019, Martinez addressed the objections support teams hear when they bring feedback to product managers. Too few users may be affected. The business impact may be limited. Other priorities may already occupy the roadmap. He treated those objections as matters to investigate, rather than evidence of an unfeeling product department.
His advice was to clarify who needed the change, what it would accomplish, and what it would cost. He recommended looking beneath a requested feature to the underlying job a customer needed to perform. Asking repeatedly why someone wanted a change could uncover a problem that admitted several solutions.
This is an important qualification to his case for listening. A company can take feedback seriously while declining the proposed fix. Listening well leaves room for judgment. In Martinez’s writing, the support team’s responsibility includes making the case in terms its colleagues can evaluate. Customer concern and product discipline have to share the same meeting.
Basketball, cooking, and a line from Dune
Outside his professional biography’s degrees and job titles are some less corporate details. Martinez cooks, plays basketball and other sports, and travels with his wife and four children. His favorite quotation comes from Dune: “fear is the mind-killer.” It is a compact reading preference for someone whose working subject involves extracting meaning from far less compact collections of language.
He has also spoken positively about leadership development through Leaders in Tech. In a public testimonial, he described learning to see himself differently and interact with others more effectively through its T-group format. The detail belongs beside his repeated advice to support leaders: consider how the person across the table experiences the situation. His public work gives that advice a professional setting; his testimonial shows him describing learning about interaction himself.
Led by Freestyle VC
November 19
Almost nine years, then Front
Front announced its acquisition of Idiomatic on November 19, 2024. It named HubSpot, Instacart, and Pinterest among the brands that had used Idiomatic’s technology. Kevin Yang and the Idiomatic team would help develop Front’s AI product roadmap, carrying customer conversation analysis into a business serving more than 9,000 customers at the time.
Martinez marked the acquisition by looking backward. Almost nine years had passed since the start, he wrote. He thanked colleagues, customers, and investors for committing their careers, success, and money to the business. His description of the experience held reward and difficulty together, without attempting to tidy away either.
The ending returns to the report that started the trouble. Martinez had spent years working on the distance between receiving a customer’s words and making a decision because of them. An acquisition changed the setting for that work. The original question remained wonderfully plain: when someone tells a company what is wrong, does anything useful happen next?