.webp)
Avaratak Blog
Free for Twelve Months. The Interesting Number Is Thirteen.
There is a conversation I have with founders about once a quarter, and it always lands in the same month. Not month one, when the tools go in and everybody is delighted. Month thirteen.
The free year has ended, the first real invoice has arrived, and somebody finally asks the question that quietly decides the next two years: “Why are there fifty-three people in here?”
I mention this because Atlassian published a genuinely likeable post on July 15 marking two years of Atlassian for Startups, and the numbers in it deserve a read. But the most useful thing a partner can do with a milestone post is not applaud it. It is to talk about the day after the milestone — which, in this case, arrives roughly 365 days after you sign up.
Two years in, the hypothesis held
The program launched in June 2024 on a straightforward premise: hand early-stage founders good collaborative tooling from day one and they spend less of their scarce attention on what Atlassian calls the collaboration tax — the compounding cost of not being able to find a decision, measure progress, or keep knowledge somewhere the next hire can reach it.
As of June 2026, Atlassian reports more than 5,000 startups onboarded: over 2,000 in year one, more than 3,000 in year two. That acceleration is the interesting shape in the data. Programs that grow faster in their second year are usually being recommended rather than marketed.
The distribution story supports that reading. Atlassian says it now works with more than 215 venture firms, accelerators, and incubators — Y Combinator and a16z’s Alpha and Speedrun funds among them. Investors are unusually careful about what they push into a portfolio, because a bad recommendation costs them credibility with founders they will need again later. Their willingness to forward this one is a real signal, and a harder one to manufacture than a case study.
What is actually in the box
Anniversary posts run light on specifics, so here is the inventory, because this is the part founders actually need. Eligible startups get Atlassian Cloud products at $0 for twelve months:
- Jira Premium — up to 50 seats
- Confluence Premium — up to 50 seats
- Bitbucket Premium — up to 50 seats
- Loom Business + AI — up to 50 seats
- Rovo — up to 50 seats
- Jira Service Management and Customer Service Management — up to 10 seats
- Jira Product Discovery Premium — 10 creator seats, unlimited contributors
Eligibility is narrower than the headline suggests, and it is worth checking before you get attached to the idea. You must not already be a paying Atlassian customer, you must be VC funded or attached to a partner accelerator or incubator, and you must not have raised more than US$10 million in external funding. Atlassian reviews every application and may ask for proof of funding. One twelve-month enrollment per company.
Three caveats a partner should say out loud
None of these are dealbreakers. They are simply the things that decide whether month thirteen is calm or interesting.
Marketplace apps are not included. If your intended architecture leans on a Marketplace app to function, that dependency is priced separately from day one. Better to know before you build a workflow around one.
The seat ceilings are not uniform. Fifty for most, ten for service management and for Jira Product Discovery creators. I have watched more than one team design a service desk against the free tier and then discover the ceiling was 10 rather than 50 right as support headcount started growing.
These are classed as Free or Beta Products. That is the category the program products sit in under Section 17 of the Atlassian Customer Agreement, and the product list can be modified over time. Entirely standard for a free program anywhere in this industry, and no cause for alarm — but the sort of thing you want your own counsel to read once rather than discover later.
The month-thirteen math
Here is the mechanic that catches people. Cross 50 users on a product — or 10 on service management — and you move to a paid plan for it. Grow Confluence Premium from 50 people to 75 and you are quoted at the 51–100 user tier, not for the 25 seats you added.
That is not a trap. It is how tiered pricing works across essentially every vendor in the category. But it does mean headcount growth and licensing cost do not move in a straight line, and the step tends to arrive in the same quarter as several other steps. Atlassian’s flexible commercial model for enterprises exists precisely because big organizations found rigid commitments a poor fit for unpredictable adoption. Startups have the same problem in miniature and rather less negotiating room. Which is an argument for modelling the number early, not for worrying about it.
What I would build during the free year
The scaffolding metaphor is the honest one. The program is twelve months of free scaffolding. What matters is the building underneath, because that is what is still standing when the scaffolding comes down.
Settle your project and space taxonomy in month two, not month ten. Renaming things is cheap at eight people and expensive at forty. This is the single decision with the longest half-life, and it costs nothing but an afternoon.
Use Confluence as a decision log, not a document graveyard. We made the same argument about why the Teamwork Collection rewards a clean foundation: connected tooling amplifies whatever is already there. Give it a well-kept record of why you decided things and it compounds. Give it a folder of untitled drafts and it faithfully amplifies that instead.
Treat the 10-seat products as design constraints, not disappointments. Ten Jira Product Discovery creators with unlimited contributors is a perfectly good shape for a founding team — a few people curating, everyone able to contribute. It is close to the model we described when we wrote about pulling scattered customer feedback into Jira Product Discovery. Design for it deliberately and the constraint stops mattering.
Give Rovo something worth reading. Fifty Rovo seats is a serious allocation, and its usefulness is a direct function of how much real context sits in Jira and Confluence. Atlassian also pointed at an Agentic Engineering template for teams working with coding agents, available for new space creation on Rovo-enabled instances — a sensible starting structure if you are already shipping alongside agents, and a natural companion to agents that can be assigned Jira work like teammates.
The Avaratak Take
This is a good program. I recommend it to eligible founders without hesitation, and the twelve free months of Premium tooling are worth real money to a company that has very little of it.
The thing I would push back on is the instinct it creates. Free tooling encourages you to defer architectural decisions, because nothing you do wrong costs anything yet. That is exactly backwards. The free year is the cheapest window you will ever have to make structural mistakes and correct them, and it is the same argument we made about doing the unglamorous assessment work while nothing is at stake. Stakes arrive on their own schedule. Structure does not.
So my advice is unglamorous and slightly annoying: spend part of your free year doing the work you would do if you were paying. Name things properly. Write down decisions. Watch your seat counts the way you watch runway. Do that, and month thirteen is a renewal conversation. Skip it, and month thirteen is an archaeology project with an invoice attached.
If you are inside the free year and would like a second set of eyes on the structure before it hardens — or you are staring down month thirteen and would like the seat math done properly — that is exactly the conversation we enjoy at Avaratak. Come find us at avaratak.com. Bring your org chart; we will bring the good questions.