Company · AI / Customer Intelligence

The startup that read every complaint so your support team didn't have to

Idiomatic built a machine that reads every complaint so support teams don't have to - one custom-tuned category at a time.

Every company likes to say it listens to its customers. Almost none of them actually read the mail. A mid-sized consumer brand can field tens of thousands of support tickets, app reviews, survey comments and forum posts in a single month. Somewhere in that pile is the answer to what to fix next. It is also, for a human being, unreadable at that volume. Idiomatic was built for exactly that gap - the distance between "we care about feedback" and "we counted it."

Founded in 2015 in Palo Alto by Kevin Yang and Christopher Martinez, Idiomatic described itself, plainly, as "Voice of Customer in a box." The product ingests unstructured feedback from wherever customers leave it - help desks, surveys, product and app reviews, social media, communities - and then does the tedious part: it reads, categorizes and quantifies all of it, close to real time.

2015Founded in Palo Alto
$4MSeed round, 2022
~10Employees at acquisition
2024Acquired by Front

The core ideaCategories in your customers' words, not a template

Plenty of tools promise to analyze feedback. Most hand you a fixed list of buckets - "billing," "shipping," "bug" - and hope your business fits inside them. Idiomatic's argument was that this is where accuracy quietly dies. A streaming service, a grocery-delivery app and a wireless carrier do not complain in the same language. So instead of a generic model, Idiomatic used supervised machine learning to build a new taxonomy for each client, tuned to that company's own products, jargon and edge cases.

The payoff is granularity you can trust. In the company's own demo, a telecom's tickets don't just land under "technical issue." They split into "Poor Wifi / Internet Connection," "Area Service Outage," "Poor Television Signal," "Poor Speed" and "Lag Time" - each with a share of volume attached. That specificity is the difference between knowing something is wrong and knowing what to fix on Monday.

Idiomatic dashboard showing a donut chart of most common ticket categories and a bar chart of signal category breakdown
The pile, sorted. Idiomatic's dashboard turns a month of angry emails into a ranked donut - here "Signal" alone is 46% of tickets, 4,846 of them - with a sub-breakdown showing exactly which flavor of "the internet is broken" is winning.
"Those years taught me respect for customer service professionals and sparked an obsession with unlocking valuable customer insights buried inside support conversations."

- Kevin Yang, Founder

How it worksFrom inbox noise to a ranked list

01 / Ingest
Connect the channels
Pull feedback from help desks, surveys, app and product reviews, social and forums into one place.
02 / Categorize
Build the taxonomy
Supervised ML tags each message against a category structure unique to the client.
03 / Quantify
Add sentiment & volume
Score tone, count occurrences, and flag anomalies as they emerge.
04 / Act
Route & prioritize
Rank issues by volume and impact so teams fix what matters first.

Who used itNames you'd recognize, a team you wouldn't expect

For a company of roughly ten people, Idiomatic's customer list read like a much larger operation. Publicly, it counted HubSpot, Instacart, Pinterest, Slack, SmugMug and Upwork among the brands running voice-of-customer programs on its platform. These are companies with real feedback firehoses - the kind of volume where manual tagging stops being a chore and becomes impossible.

Illustrative: how a support inbox breaks down by category
Poor Wifi / Internet
87%
Area Service Outage
76%
Poor TV Signal
65%
Poor Speed
46%
Lag Time
31%
Figures mirror Idiomatic's own product demo and are illustrative of the output, not a specific customer.

The businessBoring on purpose, and better for it

Idiomatic sold the way most useful B2B software does: subscription access, priced against feedback volume and the number of sources connected, aimed at support, product and CX teams. GetLatka pegged its revenue at roughly $1.1M - a reminder that the company stayed lean rather than chasing scale for its own sake. The pitch to buyers was simple arithmetic. Replace the human hours spent reading and tagging feedback, and surface the issues that quietly cost the most.

Investors bought the thesis. In May 2022, Idiomatic raised a $4 million seed round led by Freestyle Capital, with participation from Adobe's Scott Belsky, Amazon's Gert Lanckriet, Gaingels, Hyphen Capital and Xoogler Ventures - a roster that skews toward people who have shipped products at enormous scale and know what unread feedback costs.

Where it sits in the market
  • Category: voice-of-customer (VoC) and customer feedback analytics.
  • Rivals include Thematic, Keatext, Chattermill, Enterpret, Medallia and Qualtrics XM.
  • Differentiator: per-client custom taxonomies built with supervised ML, not off-the-shelf keyword buckets.
  • Buyer: support, CX and product teams drowning in unstructured feedback.

The turnAcquired, folded in, renamed

In November 2024, Front acquired Idiomatic. Front, a customer service platform serving more than 9,000 businesses, wanted to pull automated insight - inferred satisfaction, product feedback - straight out of every conversation flowing through its inbox. Rather than build that from scratch, it bought the team that had spent a decade on the problem. Founder Kevin Yang joined to lead Front's AI strategy and product roadmap, and the R&D group came with him.

"Idiomatic's expertise in AI and voice-of-customer analytics will redefine how Front's 9,000+ customers engage with and learn from their customers."

- Dan O'Connell, Front CEO

The product itself didn't disappear. Type idiomatic.com today and you land on Siena Insights - the same feedback engine, now paired with Siena's AI customer-service agents. In two years the technology has carried three names, which is its own small lesson in how B2B software actually changes hands: the logo on the door matters less than whether the thing still counts your complaints correctly.

The takeawayOne narrow job, done well enough to buy

There's a whole quiet class of AI companies that never trend. They pick a task humans genuinely hate, automate it with more rigor than a demo requires, and sell it to the teams buried in that work. Idiomatic ran that playbook cleanly - narrow scope, real accuracy, marquee logos, a strategic exit. It didn't try to be a platform for everything. It tried to read the feedback, count it honestly, and hand back a list you could act on. That was enough.

#voice-of-customer#feedback-analytics#sentiment-analysis #customer-intelligence#ai#machine-learning #saas#cx#palo-alto