How Algolia Crossed $20M ARR Making Engineers Answer Support Tickets

When Nathan role-played a senior engineer Algolia had just hired in San Francisco, one who didn’t want to answer support tickets, co-founder and CEO Nicolas Dessaigne didn’t blink: “Actually, yes, you should.” Algolia had decided very early not to have a support team. Customers with a problem reached the engineers directly, an expectation set with every engineer “from day one,” and each spent about a day a month on it (Dessaigne’s own phrasing: “half a day, two half days, a month”). He framed it as a product decision: the product, he said, “got way better” than it would have with a dedicated support team, and “it’s such a satisfying feeling when your customers has an issue, and you’re able to solve and help them.” By the time of this interview that was changing: “As we speak, we’re actually creating a support team now,” mostly to cover support around the clock, with engineers still taking part.

It worked on the numbers. In the interview, recorded in the closing weeks of 2018 (Nathan asks about raising money “as the year closes out”) and published in May 2019, Dessaigne said the search-API company, which powers internal search for customers such as Medium, Twitch and Under Armour, had “crossed the 20,000,000 ARR milestone last year,” meaning 2017. For 2018 he gave a forecast, not a result: “we are going to close close to double this year, so we’re going to be over 40,000,000 this year.” Asked whether Algolia was already over $40 million, he didn’t say yes: “we always need the last few days to get even more.” He also gave about 6,000 customers, a team of 300 across five offices, and $74M raised to date.

Two motions, one revenue base

Algolia’s model split cleanly in two, and Dessaigne was precise about the asymmetry: “Enterprise is 80% of our business today,” and “5% of our customers are going to bring about 80% of the revenue.”

• About 6,000 customers — his count: “6,000 today, we’re pretty much doubling year over year.” Nathan worked out that 5% of 6,000 is about 300 enterprise accounts, and Dessaigne called that a “good approximation.” The 300 is Nathan’s arithmetic, not a count Algolia gave.
• The sales-led side — enterprise contracts “close to 6 figures in the range, like 80 to 100 ks” a year, sold by a team split into SMB, mid-market and enterprise segments. He put SDRs and BDRs to account executives at “between one and one for one, one for two,” noting that he doesn’t lead sales himself.
• The long tail — developers who self-serve, from $35 a month.
• Pipeline as the metric that mattered — “at the end of the day, what matters is how much dollars we can close and get in the bank.”

That barbell, with self-serve developers at the bottom and near-six-figure enterprise contracts at the top, is a familiar playbook for API companies now. The tape catches its mechanics mid-build in 2018: headquarters in San Francisco, all of engineering in Paris, where the company started, and sales offices in London, New York and Atlanta.

Usage, not seats

Pricing scaled on two axes, with no seat model at all. One was usage: customers with more things to search for and more users querying the service paid more. The other was plan-gated features like personalization, a “new add on” the customer has to pay for. Both grow a customer’s bill as the customer grows.

“The beauty of the SaaS business,” as Dessaigne put it: “You start the year, you already know that you are going to have more revenue that new year than the year before without even acquiring any new customers.” (Where those numbers sit against the rest of the market: our CAC benchmarks breakdown.)

On fundraising timing he gave the answer worth framing. A raise was coming, “but not this year”: “probably next year, sometime next year, we’ll raise a new round,” because “you never want to raise when you have your back on the wall.” He wouldn’t size it (“I prefer not to communicate that”), and he made no claim to profitability: “at any point of time, my goal is to have an option to become profitable.” He filed the whole question under “basically a math problem.”

Algolia revenue: the forecast and the public record

Algolia doesn’t publish revenue. As TechCrunch put it in July 2021, “the company doesn’t share revenue numbers directly.” The figures that exist for these years come from two places: Dessaigne on this tape, and the revenue Algolia reported for its Inc. 5000 entries.

ARR is a run rate at a point in time; Inc. lists revenue booked over the year. A company growing this fast books less in a year than the run rate it exits that year at, so $15.2M of 2017 revenue fits a $20M ARR crossing that year. Inc.’s 2018 figure, about twice 2017’s, is in line with the near-doubling he forecast, but it can’t tell you whether year-end ARR passed $40M, and we found no Algolia statement that it did.