What the Baltic startup leaderboards actually measure

StartupBlink publishes a leaderboard for each national ecosystem and refreshes it monthly. The September 2026 edition is one of the few public attempts to count the companies in this part of Europe, so we read all three Baltic pages closely. Estonia's list runs to 1,183 startups, Lithuania's to 792 and Latvia's to 260. The same pages rank the ecosystems: Estonia 12th in the world and first in Eastern Europe, Lithuania 22nd and second in the region, Latvia 48th and tenth.

We used those three pages as the source for the largest expansion of the BalticNow directory so far. Thirty five new profiles went in, twelve Estonian, eleven Lithuanian and twelve Latvian, taking the directory to 90 companies. Every record cites the list it came from. Working through them country by country also made clear how much cleaning a leaderboard needs before anyone should treat it as a census.

A count of rows, not of companies

The first thing to establish is what the number is. It is a count of entries in a private database, not a company register. Nothing obliges a company to appear, and nothing removes one that has been wound up or absorbed. What ends up in the list depends on whether a company has left enough of a trace to be found, and on whether anyone filed it.

StartupBlink states its method on every page, which is more transparency than most ecosystem statistics offer.

Three inputs, then: capital raised, headcount and web traffic. Revenue is not one of them. Neither is profitability, customer retention, or whether the company is still trading.

What that rewards

• Companies that raised venture capital. A business that never took outside money scores on headcount and traffic alone, however large it has grown.
• Companies with payroll. Hiring lifts a score in a way that automating the same work, or contracting it out, does not.
• Companies with public traffic. A shop, a marketplace or a news site will out-score a B2B supplier of similar size, because a busy website can be measured and a sales team cannot.

The result is a ranking that reliably identifies the most visible and best funded companies and says rather less about everybody else. That is a defensible thing for a ranking to do. It is a poor basis for statements about how many technology companies a country contains, and it explains why the same handful of names sit at the top of all three countries.

The pages contradict themselves

Some of the contradictions are internal, which is a useful warning about precision.

• Nord Security appears in the unicorn section of the Lithuanian page while its own row gives an employee range of 11 to 50 and $8.1M raised.
• The same page says Lithuania's two unicorns have raised more than $8.4B between them, while Vinted's row credits it with $562.3M.
• Estonia's page counts Starship Technologies among the country's unicorns even though its own entry states that the company is headquartered in San Francisco.

Other problems are about hygiene, and they matter more if you are pulling the data into anything else.

• On the Latvian list, the Hodl Hodl entry describes a 3D printer for construction, Jeff's describes an Indian drone manufacturer and QuickAppNinja's describes a fitness device.
• The Estonian list files Ready Player Me under its former name Wolf3D with a description of scanning work it no longer does, and describes 5.0 Robotics as a trucking platform in Miami.
• The Lithuanian list includes Treatwell, Argyle, Revel Systems, WhiteBIT and Wargaming, each of which its own entry places outside the country. Every one of them would have flattered the national count.

None of that is fatal to the exercise. It does mean the counts should be read as approximate, and that no row should be lifted without opening it.

Where we drew the line

The directory has had strict sourcing rules from the start, and this batch was the hardest test of them yet. Only facts stated by the cited source are asserted. Taglines are written here rather than copied from a company's marketing. Founding years appear only where a listing gives one, which is why most of the new records have none.

• Companies the source itself places outside the Baltics were left out, even though including them would have made the directory look fuller.
• Rows whose description belongs to a different company were left out. An entry nobody can verify is not worth the number it adds.
• Cities, founding years and employee ranges were recorded only where the listing states them, rather than filled in from what we assume to be true.
• Where a listing is stale but the company is real, the current name was used and the outdated details left behind.

The 35 records that survived the filter are published as unclaimed, as every profile here is. A company that wants to correct or complete its entry can claim it and do so, and the correction will be credited.

What the numbers are still good for

A leaderboard is a bad census and a reasonable thermometer. Three things it does show.

• Direction of travel. The refresh is monthly, so a count moves when a cohort of companies starts raising and hiring. Watching one list over a year says more than reading it once.
• Relative weight. Estonia's 1,183 rows work out at roughly 91 per 100,000 residents, Lithuania's 792 at about 29 and Latvia's 260 at about 14. Latvia's ecosystem really is smaller, but not by a factor of six. Part of that gap is a measurement artefact: the further a company sits from Tallinn's orbit, the less likely it is to be found and filed.
• Which sectors keep recurring. Fintech, payments and identity cluster at the top of all three countries, which matches what the region is actually known for abroad.

Read per capita, the ordering of Estonia, then Lithuania, then Latvia agrees with almost every other measure of the three ecosystems, including our own directory. The absolute figures are the part to handle with care.

What to watch

• Whether the next refresh moves the Latvian count by more than a handful of entries, which would suggest coverage improving rather than a sudden wave of new companies.
• Whether any of the companies in this batch claims its profile, which would be the first sign the directory is being read by the people in it.
• Whether the internal contradictions get fixed. A unicorn with eleven employees survives only because nobody checks rows against each other.
• Whether the Baltic counts keep climbing at the rate the funding headlines imply, or flatten while the venture cycle stays tight.

Leaderboards like this are useful because somebody is willing to publish both a number and a method, which is more than most ecosystem statistics offer. The work is in reading them for what they are: a ranking of visibility built from three inputs, refreshed monthly, and worth exactly as much as the cleaning you are prepared to do yourself.