Logz.io Ranked #1 on Google for Open Source It Didn’t Write — Then Sold $40K Contracts

Halfway through a June 2017 taping, Nathan asked Tomer Levy who Logz.io actually competed with. The first answer out of Levy’s mouth wasn’t a company. “We actually compete mostly with engineers setting up their own open source,” he said — and only after that did he name Elastic and AWS. The awkward part is that the same free software was also his cheapest customer acquisition channel. Logz.io’s entire top of funnel ran on ELK, a project his team had not written and did not own.

Logz.io sells hosted log analytics: it takes the machine data that servers, databases and applications throw off, and cuts the time engineers spend chasing problems through it. Levy named British Airways among the customers on tape. Buyers were mostly DevOps and IT operations, with about a quarter of customers coming from security teams correlating log events into threats. The company started in October 2014 and shipped the product in October 2015, which made it, in Levy’s words, “about a year and a half in the market” at the taping — consistent with GetLatka’s June 2017 capture date, with the October 2016 round he calls his most recent, and with the 2017 event budget he discusses in the present tense.

Renting distribution from a project you don’t own

Asked to explain ELK to a non-technical listener, Levy kept it concrete: “Think about it like a set of libraries. You install them on a server and they basically take all the data, any data you have — it can be an application, it could be data about a credit card swiping in your e-commerce, it doesn’t matter. You put it into this one bucket, and what ELK allows you to do is basically visualize the data and understand what’s important in it.” Nathan compressed it to “Google search for all your log data.” Levy: “Perfect, yes.” Earlier in the same exchange he had put ELK at roughly half a million downloads a month.

Then came the admission that makes this business unusual. “Unfortunately, we’re not the one who actually developed the open source from the beginning,” Levy said. “It was actually developed by different folks.” What Logz.io built was the layer above: a managed log-management application in the cloud, so customers never had to install or maintain the stack themselves, with machine learning and other capabilities piled on afterwards. Levy had a label for the category — open SaaS, open source as a service — and put GitHub, Pantheon and Hortonworks in the same bucket.

Nathan went straight at the hole in it: if you didn’t create the software, how do you make sure you’re the one collecting leads when people go and use it for free? The answer was content, started before the product existed. “We’ve been writing content before, just as we launched the company,” Levy said. “Today we would be the number one contributor of content to the ELK community. So if you search Google for ELK, or AWS ELK in production, we’re number one in Google in all of these searches.”

The open source has to be good enough to get started — good enough, easy to get started and use it, so it will have mass distribution. But it has to get to a point where it’s hard to scale it, hard to make it production grade. This is where commercial companies have a lot of leverage, because otherwise people just keep making open source stuff.

Tomer Levy, co-founder and CEO, Logz.io

What that did to acquisition cost

Levy would not give an absolute customer acquisition cost, but he was emphatic about where it sat relative to everyone else. Bringing in a customer was “currently extremely low — well beyond the standard of the industry, because we bring customers through content.” On payback he was more specific: the norm he expected for this kind of business was a 12-to-24-month return period, and Logz.io was “well below these numbers.” Nathan restated it as recovering the cost inside the first year; Levy agreed it depended on the customer and the cohort, “but definitely by the second.” That claim excluded the self-serve tier, which he treated as its own animal.

The go-to-market org carrying that was small. Asked how many of the 70 employees were in sales and marketing, Levy put marketing at “probably eight or nine, team is about ten.” Nathan later read the split back as eight salespeople and seven marketing people, and Levy agreed — the two answers don’t quite reconcile, and neither was corrected. Either way the balance of the 70 sat with R&D and product, which is how Levy described the company. Against the CAC benchmarks founders disclose on the show, a content-fed funnel paying itself back inside a year is the cheap end of B2B software.

The price of the position

Cheap acquisition bought Logz.io a hard competitive spot. Its biggest rival was free and self-installed. Its commercial rivals were Elastic, which owns the open source Logz.io resells as a service, and AWS, which Levy described as offering “a simplified version of the open source” that resolves some of the pain. His answer to both was to go up-market: “We see ourself as more of a premium product, which much more expensive, working with larger companies to help them get a more strategic solution.”

The other cost showed up in gross margin. Nathan asked whether Logz.io sat in the mid-80s like most SaaS. Levy: “It’s not way lower, it’s about that, maybe a bit lower. We will be at around that number, hope, next year.” Nathan settled on roughly 80% for the taping year. The drag was hosting. “We have so much data, we have an Amazon bill that go to the millions,” Levy said — Nathan pinned that at one to five million dollars a year paid to AWS, and Levy confirmed it. “Just think about how much the constant effort of technology to make sure you’re not just moving the money to Amazon is always a challenge.”

He was also candid that he wasn’t solving it. “In a manager meeting or board meeting you say, hey, I have two engineers extra, where should I put them — on a new feature to bring a new customer to grow the business, or should I bring them to reduce the cost? And it usually goes to not reduce the cost,” he said, adding that Logz.io was investing very little in cost reduction and everything in growth and expansion. A company with almost no acquisition cost can afford to leave margin on the table; that is the same trade seen from the other side.

The revenue Levy wouldn’t give

2014 revenue was zero. The product shipped in October 2015 and Levy put that year at “six figures”. For the end of 2016 he didn’t have the number to hand and estimated “probably a few hundred K”. Then Nathan asked whether they had broken a million-dollar run rate and the shutter came down: “I would love to share, but we made a decision in the company not to share numbers, for competitive reasons.”

So Nathan built a floor out of the disclosures Levy had already made. A few hundred paying customers — call it 300 — at the $10K annual minimum Levy had quoted for the SMB tier is about $1,000 per customer per month, or $300,000 a month. That is, as Nathan put it, well over a three million dollar annual run rate — the metric he paused to define on tape as current monthly recurring revenue multiplied by twelve. Levy’s response was the only revenue confirmation in the episode: “That’s a minimum. I expect probably doing significantly higher than that.”

$3M+the annual run rate floor Levy confirmed as a minimum in June 2017 — he would not give the real figure

The pricing ladder underneath that floor had four rungs, and Levy walked all of them:

• Self-service — bought online, a few hundred dollars a month, or a few thousand dollars a year.
• SMB — about $10,000 to $15,000 a year for the product.
• SME — $50,000 a year minimum for small and medium enterprises.
• High-end enterprise — a few hundred thousand a year.

The last ten deals had landed at roughly $40,000 in annual contract value, which is the number that matters most here: the open-source funnel wasn’t only producing self-serve credit cards, it was producing mid-market contracts. Retention backed the model. Self-serve churn was the higher of the two but “still single digits yearly, still less than 10% yearly”; for anyone paying $10,000 a year or more, Levy put it “very close to zero.” That is what let him run the business as land-and-expand: “If you feel comfortable coming in with a small project for five or ten or twenty K, that’s perfect for us. And if you’re happy — and usually you are happy — you will grow.”

The money behind it

The funding ladder

Levy said Logz.io could have been a lifestyle business. “We had traction day zero. We made a decision we want to build something big, given our experience in the past — we already built companies.” On tape he put the total at twenty-four million dollars raised in two and a half years, with the most recent round, $15.6 million, closing in October 2016.

• 2015 · $6.7M the first round in the GetLatka funding ladder.
• 2016 · $16M the round Levy dates to October 2016 and sizes at $15.6M.
• 2017 · $23M the row dated to the taping year.
• 2019 · $52M the largest single round on record.
• Dec 2020 · $23M bringing total funding to $120.7M.

The two accounts don’t line up perfectly, and the gap is worth naming rather than smoothing. The dataset’s first two rounds sum to $22.7 million against Levy’s $24 million, and it carries the October 2016 round at a rounded $16 million. The third row, $23 million, is dated 2017 — effectively on top of the taping, and Levy’s own $24 million total says it had not closed when he spoke. Where the two disagree, the figure he gave on tape is his; the ladder dates are the dataset’s collection dates, not announcement dates.

What the dataset shows after the tape

The revenue record picks the story back up three and a half years later. GetLatka records $17.5 million as of November 2020, and $48 million as of June 2024 — that later figure carries the dataset’s estimate flag, so it is directional rather than confirmed. Nothing between those two points is recorded on the revenue series.

Headcount is the better-evidenced series, and it is the one that says something about the thesis. Logz.io went from 70 people at the taping to 137 by December 2018, 190 by December 2019, and a peak of 238 in December 2020. Then it came down and stayed down: 214 at the end of 2022, 200 in September 2023, still 200 in October 2024. Underneath the total, the mix moved much harder than the headline.

The sales organization more than halved — 49 people in December 2020, 29 by August 2022, 23 by the end of that year and still 23 in September 2023 — while engineering barely moved. Over the same window the recorded $17.5 million becomes an estimated $48 million. If that estimate holds, the part of the machine that survived the cut is the part Levy described in 2017: customers arriving through open source and content, landing small, and expanding on their own. Current figures sit on the Logz.io GetLatka profile.

Levy was 39 at the taping, divorced with a ten-year-old daughter, sleeping six or seven hours, reading Ben Horowitz and watching Jeff Bezos. His advice on shipping early was to get over yourself — “even though your product is, sorry for the language, [expletive], and it was [expletive] in the beginning, open up. People will understand, you iterate fast.” Asked what he wished his 20-year-old self had known, he didn’t talk about open source or distribution at all: “You think that other people figured it out… but now I see more that very few people actually figured it out, and it’s okay. Take it easy, you’ll figure it out.”

SourcesNathan Latka’s June 2017 interview with Tomer Levy, Latka episode #748; GetLatka dataset rows for funding (2015–2020), revenue (November 2020 recorded, May 2024 estimated) and headcount composition through October 2024.