Every product team runs on a quiet lie: that it knows what its customers actually want. In practice, the knowing is scattered across a Zendesk queue, a Reddit thread, a one-star App Store review left at 2am, a survey nobody read, and a Slack message from a founder who saw a tweet. The signal exists. It just never sits in one place long enough to act on. unitQ, a San Francisco company founded in 2018, built its entire business on closing that gap.
The setup is deceptively simple. unitQ ingests customer feedback from more than 100 sources - app store reviews, support tickets, call transcripts, surveys, social posts across Instagram, Reddit and TikTok - in more than 100 languages. Its machine learning translates the text, strips out personal data, tags sentiment, and sorts everything into thousands of granular categories. What was a firehose becomes a sorted, searchable, real-time map of what is breaking and why.
OriginThe bug that started a company
unitQ's founders - Christian Wiklund, Niklas Lindstrom and David Eklov - did not arrive at customer feedback as a theory. They lived the problem at Skout, the social and dating app they built before this. When you run a consumer app with millions of users, complaints arrive constantly and in every format. The hard part is not collecting them. The hard part is answering one question on any given Monday: of everything users are unhappy about, which single thing should we fix first?
That question has an expensive wrong answer. Fix the loud bug that only affects a handful of vocal users, and you have burned a sprint. Miss the quiet one silently pushing paying customers out the door, and you have burned revenue. Wiklund, Lindstrom and Eklov built unitQ so teams would stop guessing.
The metricReducing quality to one honest number
The company's first widely known idea was the unitQ Score, introduced in 2020. It is refreshingly blunt. The score benchmarks product quality in real time, and the math is easy to hold in your head: a score of 90 means roughly 10% of the feedback unitQ is watching references a quality issue. Higher is better, and the number moves as the world does - a botched release drags it down within hours, not at the end of a quarterly review.
Read it like a report card that never sleeps. At 90, about one in ten things your users are saying is a complaint. The arc fills as quality climbs; it empties the moment a release goes sideways.
The pipelineFrom firehose to a fixable list
The interesting engineering sits between "a customer said something" and "a team knows what to do." unitQ normalizes every incoming message, translates it, removes personally identifiable information, enriches it with metadata like device type, location and payment tier, then clusters it. That last step is the payoff: instead of 40,000 loose comments, a product manager sees a ranked list of issues, each with a count, a trend line, and the segment of users it hits hardest.
Just askagentQ, and the question you type at 9am
In 2024 unitQ added agentQ, autonomous AI that lets anyone interrogate the feedback in plain language. A product manager can type "which bugs are driving the most support tickets?" or "what are the top feature requests related to search?" and get an immediate, source-cited answer, enriched with the metadata behind it. It won unitQ the 2025 Artificial Intelligence Excellence Award from the Business Intelligence Group in the Innovative AI Products category. The point is not novelty. It is that the analyst who used to build that report by hand every week now spends the week fixing things instead.
The platformSix products, one throughline
In 2026 unitQ pulled its tools into a single AI Quality Intelligence Platform. Six products, and every name ends in Q. The throughline connects a spike in complaints to the revenue, retention or risk it represents - so a team can act before the dashboard everyone actually watches starts to slide.
monitorQ
Continuous AI that surfaces emerging issues across every channel in real time.
metricQ
Ties customer signals to revenue and retention so teams prioritize by impact.
competeQ
Reads competitors' customer feedback and benchmarks quality across the market.
supportQ
Grades support quality across both human agents and AI bots, every interaction.
interviewQ
AI interviews with survey speed and qualitative depth to find root causes.
socialQ
Captures social mentions and links them back to the support tickets they cause.
supportQ is the quietly provocative one. It scores human agents and AI chatbots on the same scorecard - a comparison plenty of companies would rather not run, which is exactly why it is useful. competeQ has a similar edge: it reads your rivals' one-star reviews so their weaknesses can quietly become your roadmap.
The customersWho trusts it, and why it's different
unitQ sells to product, engineering, support and CX teams at consumer-scale companies - the ones where a single bug reaches millions of people before lunch. The customer list reads like a phone's home screen: Spotify, Pinterest, Bumble, PayPal, Adobe, Block, DraftKings, Dropbox, Udemy and DailyPay among them.
The market unitQ plays in - voice of the customer and CX analytics - is crowded, with survey-first incumbents like Medallia and Qualtrics and newer feedback tools such as Enterpret, Dovetail, Chattermill and Thematic. unitQ's difference is orientation. Survey-first tools ask customers to come to a form. unitQ starts from feedback customers already left, on their own, everywhere, and works backwards to the product. It is always-on rather than campaign-based, and it insists on connecting the complaint to a business number rather than leaving it as a sentiment chart.
The businessHow it makes money, and who backs it
The model is straightforward B2B SaaS: subscription access, priced by feedback volume, sources and modules rather than a public sticker price. Backing it are Accel, which led the $30M Series B in 2021, Creandum, and Google's AI-focused Gradient Ventures, which joined the earlier $11M Series A. Total raised sits at roughly $41M. Zendesk - first a customer, then a strategic investor via Zendesk Ventures - is the tidy proof of the pitch: a company that ran the product decided to own a piece of it.
The betWhere it fits
unitQ's wager is that the most valuable data a company owns is the pile of feedback it is already ignoring, and that reading it at scale - honestly, in real time, tied to money - is a job better done by a machine than by a Monday-morning guess. For a category that has spent a decade telling companies to "be customer-obsessed," unitQ's contribution is less a slogan and more a receipt: proof of what customers said, sorted by what it costs. Whether that becomes the default layer every product team runs on, or one strong option in a busy market, is the open question. The founders, at least, have already lived the version where nobody was counting.