Skip to content

Developer and enterprise · public

How MongoDB went to market

MongoDB gave developers a free database, open source under the AGPL until 2018 and source-available under the SSPL since, then sold proprietary subscriptions through a sales force that landed one workload and expanded. Atlas, its hosted, usage-priced service, grew from 1% of revenue in fiscal 2017 to 73% in fiscal 2026.

Buyer:
Developers, Mid-market, Enterprise
Contract size:
Mixed
Checked:
October 18, 2026
The motion at a glance · From a free cluster to a company standard
  1. Developer starts free

    A developer downloads Community Server or opens a free Atlas cluster without talking to anyone.[5]

  2. Builds app, becomes advocate

    Developers build applications and champion MongoDB inside their companies.[1]

  3. Pays as usage grows

    Self-serve Atlas customers are billed monthly in arrears on what they use.[5]

  4. Usage flags sales accounts

    MongoDB reads self-serve and free-tier usage to pick the accounts its sales teams should engage.[5]

  5. Sales lands a workload

    Sales looks for the most painful application, where the budget is, to win a first workload.[11]

  6. Expands to company standard

    More workloads follow until the customer standardises on MongoDB.[1]

Back to step 1, and round again

01The phases

PrimarySupportingTried, then dropped

  1. 2009 to 2013

    A free database for developers

    MongoDB, incorporated in 2007 as 10gen, introduced Community Server in 2009 as a free download under the AGPL. Developers were the target user, and the company wrote and owned the code.

    • Open source

      Community Server was a free download under an open source licence (the AGPL at the time). MongoDB owns the code it wrote and describes the free version as analogous to freemium, not as crowdsourced R&D.[1],[11]

    • Developer-led

      Developers could evaluate the database without asking anyone, which the S-1 credits for its popularity and for driving enterprise adoption of the paid subscription. Ittycheria: the business starts and stops with developers.[1],[11]

  2. 2013 to 2016

    Enterprise Advanced and direct sales

    Enterprise Advanced arrived in 2013 and Dev Ittycheria became CEO in 2014. A field and inside sales force sold per-server subscriptions to enterprises that already ran the free version.

    • Sales-led

      Field and inside sales sold paid subscriptions, generally priced per server, to enterprises whose developers already knew Community Server. Customers sold by sales and partners were 96% of subscription revenue in fiscal 2016.[1]

    • Land and expand

      A first application lands, then more applications, departments, and geographies follow. Net ARR expansion was 128% at July 2017 and above 120% for ten straight quarters before the IPO.[1],[11]

    • Partner channel

      Global system integrators, value-added resellers, and independent software vendors resold and integrated MongoDB; the S-1 names Accenture, AWS, Microsoft, and more than 1,000 other partner organizations.[1]

  3. 2016 to 2019

    Atlas, usage pricing, and the licence change

    Atlas, the hosted database service, launched in June 2016 with self-serve signup and usage billing, and a free tier followed by 2017. In October 2018 new Community Server releases moved to the SSPL, and in November 2018 MongoDB bought mLab for about 4,200 customers.

    • Product-led

      Atlas added a self-serve path with a free tier to drive developer adoption, and it grew from 1% of revenue in fiscal 2017 to 23% in fiscal 2019. A free 512MB cluster is still offered.[1],[2],[10]

    • Usage pricing

      Atlas is charged monthly on consumption. Ittycheria says this solved the classic open source paywall problem, because a pay-as-you-go service can monetize everything.[1],[11]

    • Open source

      In October 2018 new Community Server releases moved from the AGPL to MongoDB's Server Side Public License, which makes anyone offering MongoDB as a service open source their service stack; MongoDB blamed cloud vendors capturing value. The OSI does not recognise the SSPL. Downloads rose from 30 million at the IPO to over 700 million by January 2026, and customers from over 13,400 (January 2019) to over 65,200, though the filings do not isolate the licence's effect.[2],[9],[11],[1],[3],[5]

    • Community

      MongoDB sponsored user groups and ran MongoDB University courses: 116 meetup groups with over 55,000 members and over 730,000 University course registrations by July 2017.[1]

    • Owned events

      MongoDB World, the annual user conference, is named in the S-1 alongside MongoDB University as a brand investment.[1]

  4. 2019 to now

    Atlas becomes the business, sold with the clouds

    Atlas passed 39% of revenue in fiscal 2020 and 73% in fiscal 2026, sold alongside AWS, Google Cloud, and Azure. Ittycheria, CEO since 2014, was succeeded by CJ Desai in November 2025; Desai left for Meta in September 2026 and Ittycheria returned as interim CEO while a search runs.

    • Usage pricing

      Atlas was 73% of revenue in fiscal 2026, and MongoDB expects more Atlas contracts billed monthly in arrears on usage, with no upfront commitment.[5]

    • Sales-assist

      Self-serve Atlas customers become Direct Sales Customers, and MongoDB analyses self-serve and free-tier usage to pick the accounts its sales teams should engage.[3],[5]

    • Sales-led

      Direct Sales Customers were 87% of subscription revenue in fiscal 2023, and 2,799 customers paid $100K or more in ARR at January 2026. Ittycheria tells salespeople to look for where the pain is.[4],[5],[11]

    • Land and expand

      Land a workload, find adjacencies, then become the company standard. Net ARR expansion was about 121% at January 31, 2026.[11],[5]

    • Cloud marketplace

      Atlas is sold on AWS Marketplace and Azure Marketplace, with unified billing on Google Cloud. The clouds co-market and co-sell, and per Ittycheria align their sellers' compensation, while also competing.[3],[4],[11]

    • Product-led

      The free, source-available Community Server (SSPL; open source under the AGPL until 2018) and the Atlas free tier still feed the funnel. Downloads passed 700 million, and the 10-K credits word of mouth.[5]

    • Community

      MongoDB University registrations passed 3.0 million at January 31, 2026, up from over one million in January 2021.[3],[5]

    • Owned events

      MongoDB World returned in person in fiscal 2023. By fiscal 2026 the 10-K names .local user conferences instead, many aimed at AI-native customers, and in-person events and digital marketing spend rose.[4],[5]

    • Developer-led

      The AI-era developer entry: an Atlas Managed MCP Server that connects coding agents to live Atlas data, and Voyage AI embedding and reranking models, acquired in 2025.[6],[5]

02The numbers

Atlas share of total revenue, %
0%20%40%60%80%FY2017: 1%1%FY2017FY2018: 7%7%FY2018FY2019: 23%23%FY2019FY2020: 39%39%FY2020FY2021: 46%46%FY2021FY2023: 63%63%FY2023FY2026: 73%73%FY2026
Not consecutive years: fiscal 2022, 2024, and 2025 are omitted. Fiscal years end January 31.[2],[3],[4],[5]
Show the data
FY20171%From a filing
FY20187%From a filing
FY201923%From a filing
FY202039%From a filing
FY202146%From a filing
FY202363%From a filing
FY202673%From a filing
Revenue by fiscal year
$0$1B$2B$3BFY2017: $114.8M$114.8MFY2017FY2019: $267M$267MFY2019FY2020: $421.7M$421.7MFY2020FY2021: $590.4M$590.4MFY2021FY2022: $873.8M$873.8MFY2022FY2023: $1.28B$1.28BFY2023FY2025: $2.01B$2.01BFY2025FY2026: $2.46B$2.46BFY2026
Not consecutive years: fiscal 2018 and 2024 are omitted. Fiscal 2017 uses the as-adjusted figure from the FY2019 10-K, not the S-1's $101.4M.[2],[3],[4],[5]
Show the data
FY2017$114.8MFrom a filing
FY2019$267MFrom a filing
FY2020$421.7MFrom a filing
FY2021$590.4MFrom a filing
FY2022$873.8MFrom a filing
FY2023$1.28BFrom a filing
FY2025$2.01BFrom a filing
FY2026$2.46BFrom a filing
MetricValueAs ofBasis
Revenue, fiscal 2015 to 2017 (as first filed in the S-1)$40.8M, $65.3M, $101.4M[1]January 31, 2017From a filing
Sales and marketing as a share of revenue, fiscal 2016, 2017, H1 fiscal 201887%, 78%, 72%[1]July 31, 2017From a filing
Net ARR expansion rate at IPO128%, over 120% for ten straight quarters[1]July 31, 2017From a filing
Atlas customers 13 months after launchOver 1,900[1]July 31, 2017From a filing
Revenue, fiscal 2019 (as adjusted)$267.0M, up 61%[2]January 31, 2019From a filing
Direct Sales Customers share of subscription revenue, fiscal 2017 to 201996%, 92%, 85%[2]January 31, 2019From a filing
Direct Sales Customers share of subscription revenue, fiscal 202387%[4]January 31, 2023From a filing
Revenue, fiscal 2026$2,463.8M, up 23%[5]January 31, 2026From a filing
Revenue, second quarter fiscal 2027$771.8M, up 30%[6]July 31, 2026From a filing
Customers, total and on AtlasOver 65,200 total, over 63,900 on Atlas[5]January 31, 2026From a filing
Customers with $100K or more in ARR, fiscal 2024 to 20262,052, 2,396, 2,799[5]January 31, 2026From a filing
Net ARR expansion rateAbout 121%[5]January 31, 2026From a filing
Sales and marketing as a share of revenue, fiscal 2024 to 202647%, 43%, 38%[5]January 31, 2026From a filing
Downloads since February 2009 (cumulative)More than 700M[5]January 31, 2026From a filing

“Reported” figures come from trade press, not the company, and are shown with their date.

03Myth checks

  • “MongoDB makes money from open source.”

    Partly true

    The free edition is acquisition, not revenue. Revenue comes from Atlas (73% in fiscal 2026) and Enterprise Advanced, a proprietary self-managed edition, and free users are excluded from the customer count. Ittycheria calls open source a freemium strategy monetized as a service, and since October 2018 new Community Server releases ship under the SSPL, which the OSI does not consider open source. The accurate version: a free, source-available edition is the funnel for a proprietary edition and a usage-priced hosted service.[1],[5],[11],[9]

  • “Developers chose MongoDB, not companies.”

    Partly true

    Developer adoption is the entry point in every filing, and the S-1 says the go-to-market model is primarily focused on developers. But the money comes from customers sold by a sales force: Direct Sales Customers were 96% of subscription revenue in fiscal 2017 and still 87% in fiscal 2023. Ittycheria says developers can say no but not yes, so sales must turn their enthusiasm into a business case for a decision maker.[1],[2],[4],[11]

On the record

  • “We use open source as a freemium strategy to kind of let people use MongoDB quickly, play with it, tell their friends and leverage the virality of the open source community.”

    Dev Ittycheria, CEO, MongoDB[11]

    Answer to "What are the key drivers behind that?"

  • “developers don’t necessarily have the right to say yes, they have the right to say no.”

    Dev Ittycheria, CEO, MongoDB[11]

    Answer to the question on bottoms-up versus top-down go-to-market

  • “Our go-to-market model is primarily focused on driving awareness and usage of our platform among software developers with the goal of converting that usage into paid consumption”

    MongoDB, S-1 registration statement[1]

    Business, Sales and Marketing section

04What other companies can take from it

05 · GTM Now

What changes in the AI era, and what to do now

Analysis by Deepak Gupta. This is opinion built on the sourced record above, not part of it.[6],[5],[10],[11],[7],[8]

Still works

  • The free tier and usage pricing. MongoDB still offers a free Atlas cluster with pay-as-you-go growth, which fits an AI-built prototype that needs a database (analysis).
  • Landing by workload. AI projects typically start as one application, which suits a land-and-expand motion (analysis).

Breaks

  • Developer mindshare built through downloads, docs, and University courses is weaker when coding agents choose the stack for the developer (analysis).
  • The developer veto Ittycheria describes now partly sits with an agent's defaults, which no download count measures; no source here measures agent-driven database choice (analysis).

What replaces it

  • Meet the agent where it works: MongoDB launched an Atlas Managed MCP Server that connects coding agents to live Atlas data.
  • Own the AI retrieval layer: Voyage AI embeddings and marketing aimed at AI-native customers. Whether these change agent defaults is unsourced (analysis).

Apply it now

If you are

A seed to Series B infrastructure or data company (database, queue, search, observability) whose first user is a developer, with a self-managed edition today and enterprise deals at $50K to $500K ACV

  1. Decide in writing whether the free edition is distribution or a community project, and keep IP ownership (or a CLA) so the licence terms stay yours.
  2. Ship a hosted version with a permanent free tier and pure usage pricing, so the paywall question goes away.
  3. Score self-serve and free accounts on usage and route the top ones to sales, as MongoDB's 10-K describes.
  4. Train sales to win one painful workload per account, and track workload count per account as the expansion metric.
  5. List on the cloud marketplaces your customers already buy from, so their committed cloud spend can pay you.

You will know it works when: The share of subscription revenue from accounts that started self-serve or free, and net ARR expansion above 120% (MongoDB's filed range).

Based on MongoDB's Atlas, usage pricing, and the licence change, Atlas becomes the business, sold with the clouds, Atlas becomes the business, sold with the clouds

If you are

An established open-core or source-available vendor whose hosted product is growing but whose developer funnel now runs through AI coding agents

  1. Publish an MCP server or agent integration for your hosted product so agents can provision and query it directly.
  2. Make the free tier usable from an agent with no human signup friction, and count agent-originated signups separately.
  3. Add the AI capability your category needs (retrieval, embeddings, evaluation) inside the existing product, as MongoDB did with Voyage AI.
  4. Keep routing usage to sales, and add agent-originated usage as a signal.

You will know it works when: The share of new hosted accounts created through an agent or MCP integration, and their conversion to paid usage within 90 days.

Based on MongoDB's Atlas becomes the business, sold with the clouds, Atlas becomes the business, sold with the clouds

Motions in this record

Share the visuals

Every card is drawn from the sourced record above and sized 1080 by 1350 for LinkedIn and X. Free to share with a link back to this page.

Sources

  1. [1]SEC filingMongoDB, Inc. Form S-1 Registration Statement (September 21, 2017)
  2. [2]SEC filingForm 10-K, fiscal year ended January 31, 2019 (April 1, 2019)
  3. [3]SEC filingForm 10-K, fiscal year ended January 31, 2021 (March 22, 2021)
  4. [4]SEC filingForm 10-K, fiscal year ended January 31, 2023 (March 17, 2023)
  5. [5]SEC filingForm 10-K, fiscal year ended January 31, 2026 (March 11, 2026)
  6. [6]EarningsMongoDB, Inc. Announces Second Quarter Fiscal 2027 Financial Results (8-K exhibit) (September 1, 2026)
  7. [7]SEC filingMongoDB Announces Leadership Transition (8-K exhibit) (November 3, 2025)
  8. [8]SEC filingMongoDB Announces CEO Transition (8-K exhibit) (September 28, 2026)
  9. [9]CompanyServer Side Public License FAQ (undated)
  10. [10]CompanyMongoDB Pricing (undated)
  11. [11]Founder interviewIn conversation with Dev Ittycheria, CEO, MongoDB (Data Driven NYC, edited transcript) (Dev Ittycheria, May 4, 2021, at Edited transcript with no timestamps; each quote is located by the host question its answer follows)

Every phase, motion, and myth verdict cites a filing, the company's own pages, or a founder speaking on the record. See the methodology.