Mehdi Daoudi and the trouble with green lights

The people in the operations room looked comfortable. Feet were up. The monitoring said everything was fine. Somewhere beyond that reassuring picture, a major service failure was under way. Mehdi Daoudi has returned to this scene when explaining the idea behind Catchpoint: a company could watch its own machinery closely and still miss the experience of the person trying to use it.

The scene contains a small embarrassment for a whole profession. The instruments were working; the question they were answering was too narrow. A green light can tell you something useful about a server. It cannot, by itself, tell you whether a customer has managed to get through. Daoudi’s career has been an extended investigation of the distance between those two statements.

A very small file, a very large bill

He has an unusually personal reason to care about outages. In 1999, while working at DoubleClick, he accidentally deleted a pixel file during a cleanup. In his later account, the mistake disrupted 5,000 ad servers for three hours and produced a $1.5 million refund to one client. He remembered being fortunate that his boss kept him on. A file small enough to remove casually had consequences large enough to follow him into another company.

It is an awkward origin story for someone selling reliability, which makes it a useful one. Daoudi had occupied the chair of the person who caused the problem. He knew what happened after the reassuring assumptions stopped being reassuring. Catchpoint would give him an opportunity to build tools he wished he had possessed earlier.

His work at DoubleClick involved buying, building and deploying monitoring for the DART infrastructure, which handled billions of transactions a day. Later came Google, following its acquisition of DoubleClick. This gave him a particular education in scale: monitoring was part of operating an advertising system whose customers expected it to keep serving, rather than an interesting exercise conducted at a safe distance from the result.

Another lesson concerned the promises surrounding the machinery. While setting up DoubleClick’s quality-of-service team, he found himself dealing with service-level agreements after the company paid more than a million dollars in penalties to a single customer. A contract had promised 100% uptime. The sentence was admirably clear. Delivering it proved considerably less straightforward.

His response reached beyond engineering. The team collected contracts, recorded obligations and penalties, and brought legal, finance, sales and operations into the same discussion. The work required a shared account of what the company had promised and how it would measure delivery. Reliability acquired a commercial vocabulary. A missed target could now be connected to the commitment that made it matter.

The garage had a short lease on the story

Catchpoint began in 2008 with four people who had already learned to work together: Daoudi, Dritan Suljoti, J. Scotte Barkan and Veronica Ellis. They had collaborated in DoubleClick’s research and development team. The shared history matters. They were carrying working relationships into a new company, along with a fairly developed understanding of the shortcomings they wanted to address.

The garage episode lasted about a week, in Los Angeles that July. Then came apartments in New York and work around a dining table. Remembering the timing later, Daoudi offered a compact review of founding during a financial crisis: “A great year to start the company… 2008, highly recommend.” The humor has the economy of someone who has already paid for the lesson.

The new business also required them to choose where to spend their expertise. Their background made advertising technology an obvious possibility. They stayed with monitoring. It was a decision to keep working on a problem they had experienced directly, even when a different application of the same team might have been easier to explain.

There is something revealing about the technical choices that followed. Catchpoint built its own database, Orchestra, to handle granular performance information. Its first monitoring node was in Las Vegas. The founders were trying to construct a system around the kind of evidence they wanted, including statistics and percentiles that could preserve detail instead of smoothing it away.

A customer has no use for your average

By 2013, Catchpoint had reached its first 100 customers. Signing Priceline marked a further change: it was the first business-to-consumer brand the company signed, after an early focus on advertising and Internet infrastructure companies. The problem was spreading across the customer list. A business did not need to be an Internet supplier to depend on Internet delivery.

The network grew with that proposition. In September 2023, Catchpoint reported 2,459 monitoring vantage points. The number describes places from which to look, rather than a score for how well everything was performing. That distinction is central to Daoudi’s approach. A service can be reachable along one route and troublesome along another. Where a measurement begins changes what it can reveal.

In presentations, he has explained synthetic monitoring through the idea of mystery shoppers. Tests approach a service from different locations and providers to discover what happens along the way. The ordinary retail metaphor is helpful. Inspecting a shop’s stockroom tells you something; sending someone through the entrance, down the aisles and to the till tells you something else about the same business.

At Cloud Field Day in 2025, the discussion extended from detecting symptoms to locating their causes. A slow page is the beginning of the inquiry. The practical task is to identify which dependency needs attention, get that information to the right people and check whether the repair worked. For Daoudi, collecting evidence earns its keep when it helps an operator decide what to do next.

Learning to leave the broom within reach

The interest in computers came before the executive responsibilities. His first machine was a Commodore 64. He encountered the Internet while studying in Paris in 1992, and remembered seeing the possibility of a change in how people communicated and shared knowledge. His degree was in international trade, marketing and business. The career that followed put those subjects in sustained conversation with technical operations.

He has also said he wishes he had taken an MBA earlier. That admission sits comfortably beside his warning that founders need to learn how to bring a product to market. Making something work creates one set of demands. Persuading a customer, building a sales organization and supporting the buyer create others. The willingness to learn an unfamiliar discipline becomes part of running the business.

“Well, I think you can say that we are consistent.”

Mehdi Daoudi, during the founders’ 2021 conversation

His preferred account of company culture makes room for practical work regardless of rank. At Catchpoint, that culture emphasizes grit, persistence, humility and a willingness to sweep floors. It is a useful image for a chief executive to keep nearby. A customer problem rarely arrives neatly divided according to the organization chart, and someone still has to pick up the broom.

Trust has been another recurring subject in his management thinking. He has described an interest in Patrick Lencioni’s approach to teams and in making objectives visible. Away from the office, he has named family, cooking and travel as ways to unwind. The public personality is a mixture of technical urgency and an ability to admit that some of the management lessons arrived later than he would have liked.

The next user may skip the login screen

On December 2, 2025, LogicMonitor announced that it had completed its acquisition of Catchpoint. Daoudi was Catchpoint’s CEO and co-founder at the announcement. By early 2026, his published title was general manager of Catchpoint at LogicMonitor. The combination brought Internet performance monitoring together with LogicMonitor’s infrastructure observability and AI capabilities.

His personal note about the deal spent time assigning credit. Suljoti had supplied the Catchpoint name at the eleventh hour. Ellis had established engineering practices and built the front end. Barkan had constructed the database. Daoudi also thanked mentors, colleagues and families, including the people whose birthdays, holidays and dinners had absorbed the cost of customer emergencies. The letter made a company’s operational history visible through other people’s calendars.

The ambition he described with LogicMonitor was to connect more of the evidence and move towards anticipating trouble. In a May 2026 conversation, he stressed that automation still depends on having sensors in the right places. AI can help join scattered clues, but the usefulness of the result depends on what was collected. Humans would remain involved in setting guardrails and refining the response.

He was also asking harder financial questions. In January 2026, he argued that observability needed a business case: spending and data collection should connect to outcomes such as customer retention, support costs and losses from disruption. The old concern about misleading reassurance had reached the boardroom. An impressive collection of tools still had to justify the decisions it helped people make.

By July, his writing was considering another change in the person, or thing, at the other end of the connection. An AI agent using a service may never visit a login screen. Familiar signals of human activity can disappear. For a founder who built around the user’s vantage point, this poses a fresh version of a familiar question: where should you stand to see what the service is actually doing?

The green light remains a useful piece of information. Daoudi’s instinct is to look past it, towards whoever is waiting for something to work.