RGBvision RGBvision
Back
bimarket

Yandex DataLens pricing from December 2025: what BI will cost

Yandex DataLens moves to per-seat pricing. We run the numbers for teams of 5, 15 and 40 and look at when a cloud BI tool is not needed at all.

Published: 2025-11-19

On 17 November Yandex Cloud announced a new pricing model for DataLens, one of the most widely used BI tools in Russia. From 1 December the service is billed per seat rather than per active user. One seat for individual use is free with every feature, including the Neuroanalyst AI agent, while teams pay 990 rubles a month for each seat. The short answer for companies that built their reporting on the free Community plan: team use stays free until 1 March 2026, but the BI budget needs to be planned now.

What changes and when

DateWhat happensWho is affected
17 November 2025New environments get a 30-day trial with no user limitNew customers
1 December 2025Billing per seat instead of per active user, applied automaticallyBusiness plan
1 December 2025Moved to an extended trial of the unified plan, team features included, until 1 March 2026Community plan
1 March 2026Team use becomes paid; one individual seat stays freeFormer Community
1 March 2026Secure embedding of charts and dashboards is billed: 2,000 requests per seat, extra seats needed beyond that limitBusiness plan

A seat is a slot, not a person: there is a fixed number of seats, and they can be reassigned between employees. For existing customers, automatic purchase of new seats is on by default. Check this setting first, otherwise the seat count will grow with every new user.

Community users have a reason to decide early: switching to paid usage before the end of 2025 gives a 50% discount on all seats until 31 December 2026.

What it costs

We use the announced price: 990 rubles a month per seat, and for teams it applies from the first seat. The announcement does not say whether VAT is included, so check the amount against the pricing rules in the documentation.

TeamPer monthPer yearPer year with 50% discount (until end of 2026)
5 people4,950 rubles59,400 rubles29,700 rubles
15 people14,850 rubles178,200 rubles89,100 rubles
40 people39,600 rubles475,200 rubles237,600 rubles

The key difference from the old model: you used to pay for people who actually worked with data that month, now you pay for assigned seats. A manager who opens a dashboard once a quarter costs as much as an analyst who lives in DataLens. So the first question is not "how many employees do we have" but "who really needs reports every month".

If you are on Business and embed dashboards in your own portal or customer account area, count requests too. From 1 March each seat covers 2,000 secure embedding requests, and with heavy traffic you will pay for seats nobody works in.

Open-source BI: the license is free, support is not

The alternative to the cloud is hosting BI yourself. Options include Apache Superset, Metabase, and the open-source edition of DataLens itself, which Yandex released under the Apache 2.0 license back in 2023. According to the author of a Habr overview, Superset is used by large Russian companies such as Sber, MTS and Rostelecom.

The license costs nothing, but you pay in other ways:

  • a server and database for BI, plus backups;
  • updates, and fixing whatever breaks after them;
  • single sign-on and access rights, which you set up and maintain yourself;
  • an engineer's time to look after all of this, and cover while they are on holiday.

In our estimate, for a team of 5–15 people cloud DataLens is almost always cheaper than a self-hosted install: 60–180 thousand rubles a year is less than an engineer's regular time. With 40 or more users, a requirement to keep data inside your own perimeter, or heavy embedding, the calculation is worth doing seriously. For large companies Yandex also offers a commercial DataLens Enterprise on-premises edition.

Neuroanalyst is now available to everyone

There is a second piece of news inside the pricing one: features that used to be Business-only become available on the free seat as well. Among them is Neuroanalyst, an AI agent that works as a chat, gives insights on charts and dashboards, helps build calculated fields and write JavaScript for custom visualizations in Editor. It was announced in July, with external availability promised for September.

In our experience, assistants like this are only as useful as the data underneath them. If "revenue" includes VAT in one dataset and excludes it in another, if the customer directory has duplicates and fields are called col_17, the AI agent will confidently build a chart and draw the wrong conclusion from it. Before trusting it with conclusions for management, you need shared metric definitions, clean reference data, and access rights set up so that an employee sees only their own data through the chat.

When you do not need a separate BI tool

Cloud BI wins when data lives in several systems: CRM, accounting, ad accounts, spreadsheets, and has to be brought into a single picture. If almost all of your data lives in one operational system, it is often simpler to read reports right there.

We solved this in our own product, Aphorio: in this management system for service networks, reports and analytics across branches run on the same data as bookings, the till, inventory and finance, with no export to a separate BI tool. For a small chain of salons or studios that is often enough. If you need to combine many external sources and build a complex data model, that is a job for a BI platform, and then the comparison is between DataLens, open-source tools and custom development.

What this means for business

For small teams that built reports on Community, BI stops being free but stays inexpensive: 5 seats cost under 5,000 rubles a month. For Business customers the model becomes more predictable but penalizes unused seats. Anyone embedding dashboards into customer-facing services should check the load before March.

The license is not the only cost. On a corporate BI platform we took to production, customers asked not only about charts but about single sign-on, security and predictable releases, and the desktop version lifted sales by 40% (case study). The same applies to your choice: alongside the license, account for who will own access, data quality and changes.

What to do before 1 March

  1. Count your real users. Export who opened dashboards over the last 2–3 months and separate daily users from those who drop in once a quarter. The latter can get reports by email or as a PDF.
  2. Check automatic seat purchase. Decide who in the company assigns seats, and turn off automatic purchase if you do not want the bill to grow unchecked.
  3. Count embedding requests. If dashboards are embedded in a portal or customer account area, estimate monthly requests and compare them with the 2,000-per-seat limit.
  4. Decide by the end of December if you are on Community. The 50% discount until the end of 2026 applies if you switch to paid usage before the end of 2025.
  5. Sort out your metrics before adding AI. Before handing management questions to Neuroanalyst, agree on how key metrics are calculated. We help with this as part of BI implementation: the first working dashboards usually appear in 3–6 weeks.

Sources