Following its $135 million Series D last week, Monte Carlo became the latest unicorn in a fast-rising category: data observability, which the startup defines as “an end-to-end approach to enable teams to deliver more reliable and trustworthy data.”
If you are wondering how serious data quality issues are, Monte Carlo CEO Barr Moses has an answer: “Data quality issues still plague even the most data-driven companies. Just a few weeks ago, Unity, the popular gaming software company, cited ‘bad data’ for a $110 million impact on their ads business.”
Moses’ startup isn’t the only one to go after the data observability market opportunity. On the same day that Monte Carlo disclosed its newly minted $1.6 billion valuation, competitor Cribl confirmed its unicorn status with a new round of funding.
“While smaller than Cribl’s Series C, which came close to eclipsing $200 million, the Series D values the company at $2.5 billion post-money, according to a source. That’s up from $1.5 billion as of August 2021,” TechCrunch’s Kyle Wiggers noted.
Any three-digit deal would be noteworthy in isolation. Two of them in the same space, even more so. But what really caught our attention is that Monte Carlo’s and Cribl’s deals were announced now, right in the middle of a broad startup downturn.
We know that large rounds can take time to get both closed and disclosed, meaning that Monte Carlo’s and Cribl’s Series D rounds might reflect the state of the market a few weeks ago. But there’s a more recent data point to take into account: hiring, which is still happening.
On one side of the table, companies are still filling the kind of positions that create demand for data quality solutions. “Despite the volatility, data engineers and analytics jobs are increasing and companies are continuing to hire at record numbers for these roles,” Moses told TechCrunch. On the other, data observability startups themselves are hiring. Not just unicorns like Cribl and Monte Carlo, but also competitors like seed-funded startup Sifflet.
Could data observability be recession-proof? To find out, we talked to Moses, as well as Sifflet CEO Salma Bakouk. To complete their firsthand knowledge, we collected notes from two investors familiar with the space: FirstMark partner Matt Turck and Data Community Fund general partner Pete Soderling.
The picture that emerged from our conversations is that tailwinds for the data observability category as a whole might not translate into wins for each and every startup in the space. Why? Let’s explore.
Rising with the data tide
When we mention tailwinds for data observability, it’s because demand is driven by a broader trend. TL;DR: More and more companies are becoming data-driven, and therefore facing the kind of data quality issues that data observability startups are made to address.
Sizing a growing opportunity is never easy, but in our conversations, we heard that data obs could soon be a universal problem for large companies.
“I’m a big believer that every company, both tech and non-tech, is going to need to become not just a software company, but a data company,” Turck said. “That’s why people are excited about the opportunity — it’s a very large market and a huge trend.”
That the addressable market for data observability is large is one thing. But it would be meaningless if target companies themselves weren’t seeing reliable data as a need. According to Moses, that’s increasingly the case in all kinds of sectors.
“As companies ingest more and more data to power critical parts of the business, the opportunity for data downtime — in other words, periods of time when data is missing, unreliable, or otherwise erroneous — only grows,” Moses said. “These data-quality issues can cause organizations to lose money, waste resources and erode valuable trust with customers. In fact, Gartner estimates data downtime and poor-data quality costs the average organization $12.9 million per year, a number that’s bound to increase as businesses get more data dependent.”
In practical terms, being data dependent refers to the fact that companies are increasingly relying on sophisticated analytics and use cases driven by machine learning not just for R&D purposes, but “to power their businesses,” Soderling said. In this context, Turck noted, “being able to trust your data is not really optional anymore.”
Sailing the recession?
It’s still early to tell how demand for data, let alone data observability, will fare if the downturn were to turn into a full-blown recession. Data observability, Soderling said, “is clearly already a priority for data-driven companies.” But, he added “I think we’ll see in this current downturn which companies are really prepared to put their money where their mouth is when it comes to being truly data driven versus data curious.”
In other words, the downturn could temporarily reduce the total addressable market of data observability startups. However, it may also make it easier to preach to the converted. Sifflet, for instance, reports seeing “more customer demand than ever before in the past couple of months.”
As to why demand is increasing, Bakouk ventured that companies want the kind of difficult decisions they need to make right now to be backed by reliable data. As exemplified by Unity, the market is also less forgetting during times of uncertainty, she noted. “Data incidents, or what I call data catastrophes, are simply not tolerated.”
But even if demand for data observability holds up or increases, startups in the space might still be affected by the downturn. More precisely, they could pay the price of past euphoria.
“As the opportunity is both very large and fairly obvious, a number of startups were launched in a matter of a couple of years to go after it,” Turck said. “And that coincided with a very hot funding market, both in general but also in the data infrastructure space, as many VCs got really excited after the amazing success of Snowflake. As a result, you end up in this situation where you have a lot of smart founders running startups that are promising and well funded, but everyone is pretty early in their journey and the category is already very crowded.
“It’s unclear how it all shakes out,” Turck added.
We tend to agree, but let’s still look at what we know so far.
Don’t crown winners yet
One of the things we know is that Monte Carlo just raised a massive Series D round (and so did Cribl.) When asked how to interpret this, sources shared mixed impressions.
Bakouk described Monte Carlo’s announcement as “a great signal for the category as a whole” and a “great proof of the category’s legitimacy.” It could help further educate customers who, “although well aware of the consequences of bad data, are still unsure of what to expect from a tool let alone the budget that should be allocated to it.”
However, Monte Carlo’s new unicorn status doesn’t necessarily reflect a new wave of enthusiasm for data observability startups. “The Monte Carlo funding is more a trailing indicator than a leading indicator of excitement about the space, in my opinion,” Turck said.
Turck has another hypothesis on Monte Carlo’s fundraising record. “The fact that Monte Carlo has basically raised four back-to-back rounds in less than two years,” he told TechCrunch, “is also a bit of a ‘shock and awe’ strategy from both the founders and the VCs to will into existence the category-dominating platform that covers all parts of the [data-quality] problem.”
Mega-rounds like Monte Carlo can act as a deterrent rather than an incentive. “That kind of strategy can have the side effect of discouraging others from entering, or further funding, the space,” Turck said. Either way, he doesn’t expect as many new players as a couple of years ago: “As the financing market cools down, I don’t anticipate a lot more brand new entrants, unless they offer something truly differentiated.”
How about existing players? Again, feelings are mixed. “Funding will certainly become trickier, but great ideas with solid metrics will continue to get funded,” Bakouk said. Even if they have funding, they should still be wary of spending cash as if market conditions hadn’t changed. “I think all startups with high valuations compared to their actual revenues (like Monte Carlo) should be cautious heading into the current market downtrend,” Soderling warned.
More than its spending plans, Monte Carlo’s fundraising strategy seems to reflect the company’s vision for a space that its co-founders have compared again and again to another one: software observability. “Just like every engineering team leverages a reliability solution like Datadog or New Relic, every data team needs an observability solution to ensure their data products are trusted and accurate,” Moses argued.
As Alex Wilhelm noted, “the analogy [between data observability and software observability] is reasonable as both software niches deal with flagging issues with software systems in motion, if somewhat rosy for Monte Carlo.” Indeed, both New Relic and Datadog, well-known software observability tools, are public companies.
If data observability is anything like application performance monitoring, then it may not be a winner-take-all market. The APM market, Soderling noted, has “several large players: Datadog, New Relic, AppDynamics, etc.”
Oligopoly then, not monopoly. But that still leaves room for consolidation, something that is due to data observability having a lot in common with adjacent categories. “Right now, you have a bunch of startups tackling different parts of the problem: lineage, data quality, observability per se, orchestration, etc. But ultimately customers just want their business problems fixed: Is my data reliable? If not, what’s the problem? And once I know the problem, how do I fix it?”
Turck is well placed to talk about consolidation. He participated in the most recent funding round of data orchestration startup Astronomer, which recently acquired data lineage Datakin, in which he was the lead investor. As for Soderling, he invested in Anomalo, Soda and Superconductive, which strictly speaking aren’t data observability startups but could be seen as competitors to Monte Carlo and its closest peers. These include Sifflet, but also Acceldata, Datafold and Metaplane.
Bakouk too wouldn’t be surprised to see more concentration. “I think there will be consolidation with data lineage, data catalogs, which will be useful for the data management category.” She hopes her startup is one step ahead; Sifflet invested in building its own data lineage solution and describes itself as a full data stack observability platform. “We monitor the data across the whole enterprise data pipeline to ensure its reliability from source to destination,” she said.
Moses also describes Monte Carlo’s vision as encompassing. “We believe that the winning approach will apply an end-to-end lens across the entire data stack, providing not just a point solution for the data warehouse or BI layer, but a comprehensive platform that gives data teams the tools to detect, resolve and prevent data issues at each stage of the pipeline — no matter what stack they’re using.”
Extrapolating from Turck’s comments, going broad seems to be a good bet. “Some sophisticated customers will want to stitch best-of-breed solutions together, but many (most?) will want a broad platform that takes care of all parts of the problem in a cohesive, integrated manner,” he said.
The APM analogy once again came in handy. “If you look at Datadog as a comparison,” Turck said, “they’re a broad horizontal platform that takes care of everything. I think that’s what people are going to want in the data space, too.”
Therefore, “it’s not hard to imagine how a very big public company could be built in the data infrastructure space, as that market continues to grow exponentially over the next few years.”
Who could that be, if anyone? Only time will tell.
Disclosure: I am a former contractor of Pete Soderling’s Data Council. I don’t have ties of any kind to his fund or portfolio companies.
Comment