Atlassian set dates to stop selling and supporting its Server products, then reported the cost openly. A three-list format that brings the same discipline to marketing reviews.
Most quarterly marketing reviews are a list of things that happened. The useful ones end with three short lists: what we will stop, what we will strengthen and what we will test.
The hardest of the three is stopping. One of the clearest business examples of a deliberate stop comes from Atlassian, the Australian-founded software company behind Jira and Confluence.
Atlassian’s decision to stop
In October 2020, Atlassian announced it would end its Server products, the versions customers installed and ran on their own hardware. Business Insider reported the move at the time, describing Atlassian as a $47 billion company. The headline quoted the co-CEO’s reasoning: “The cloud is the natural future.”
The plan ran over years, not weeks. According to Atlassian’s end-of-support FAQ, new Server licence sales ended on 2 February 2021, and support for all Server products ended on 15 February 2024.
That gave customers more than three years to move to Atlassian’s Cloud products or its self-managed Data Center editions. Business Insider also reported a price rise on Server renewals and upgrades during the transition window. Atlassian told the publication this was in line with earlier, regular price increases.
What the numbers show now
Atlassian’s results for the year to 30 June 2026 show where the company placed its bet. In its earnings release, filed with the SEC on 6 August 2026, Cloud revenue was $4,410.6 million, up 28% on the year before.
Total revenue was $6,572.3 million, up 26%. Data Center revenue was $1,830.9 million, up from $1,467.2 million.
Atlassian has since made a second stop decision. Its shareholder letter for the same quarter says it announced in September 2025 that Data Center products would reach end of life in March 2029.
The letter is candid about side effects. It says the announcement pulled some Data Center revenue forward into fiscal 2026, and it expects Data Center revenue to fall by about 17% in fiscal 2027.
What marketing leaders can take from it
These are Atlassian’s figures, and they do not prove that every stop decision pays off. Many factors drive cloud growth. The lesson for marketers is about how the decision was made and communicated.
Atlassian named the stop explicitly. It did not quietly starve the Server line of attention. It published dates.
It sequenced the exit. New sales stopped first, then support, with years in between. Customers could plan.
It reported the cost. The shareholder letter explains how the decision distorts revenue in the short term. That honesty makes the long-term case more credible.
Marketing teams rarely stop anything this clearly. Old campaigns, channels and content series continue because nobody chose to end them.
Why stopping is a marketing job
A stop decision is also a communication challenge. Atlassian had to explain to existing customers why a product they relied on was going away, and what to do next.
That is marketing work: segmenting customers by how hard the move would be, sequencing messages, and giving partners something useful to say. The same skills apply inside your own team when you end a programme.
There is a budget reason too. In most marketing plans, new ideas compete for money that is already committed to old ones. Without a stop list, the test list stays small and the plan barely changes from year to year.
This is analysis rather than a finding from Atlassian’s documents, but it explains why the stop list should come first in the meeting, not last.
Sitetrail NewsPass · SEO, PR & GEO
Build your search visibility, media presence and AI discoverability
Explore NewsPass for your SEO, PR and generative engine optimization strategy. Use the calculator to compare campaign costs and find a plan that fits.
Basic
$199/moSilver
$399/moGold
$599/moSee how NewsPass works
Run the review in three lists
A quarterly review works best as a 90-minute meeting with a pre-read sent two days before. The pre-read should fit on two pages.
Each item on any list needs four fields: what it is, the evidence, the decision, and the owner.
1. Stop
Look for activities that absorb budget or time without moving a business goal. Typical candidates are a channel with falling results for two quarters, a content series nobody in sales uses, or an event kept alive by habit.
Use this test for each candidate: if we were not already doing this, would we start it today? If the answer is no, it belongs on the stop list.
Follow Atlassian’s pattern on a small scale. Set a date, tell the people affected, and write down what you expect to lose.
2. Strengthen
Strengthen means giving more money, people or senior attention to something that already works. The bar is evidence, not enthusiasm.
Good evidence includes an experiment with a control group, consistent results across several quarters, or pipeline that sales leaders confirm. A single strong month is not enough.
Limit this list to two or three items. If everything is strengthened, nothing is.
3. Test
The test list is for ideas you cannot yet judge. Each test needs a question, a budget cap, a time limit and a success threshold agreed in advance.
Ask what would make you stop the test early. Writing that down protects the team from running tests that never end.
A worked example
Hypothetical: a Dutch logistics software company reviews its quarter. Its marketing team supports three regional trade shows, a weekly webinar, a partner programme and paid social.
The review finds that one trade show produced no qualified opportunities in two years. It goes on the stop list, with notice to the organiser and an estimate of the brand presence it may lose.
Partner-sourced deals closed faster and at higher values for three straight quarters. The partner programme goes on the strengthen list, with one extra partner manager.
Nobody knows whether short video case studies would help sales in Germany. That becomes a test: six videos, a fixed budget, eight weeks, and a target agreed with the German sales lead.
The figures and choices in this example are illustrative. The structure is the point.
Keep the stop list honest
Stopping is uncomfortable because someone usually built the thing being stopped. Three habits make it easier.
Separate the work from the people. Thank the team for what they learned, and move them to the strengthen or test list.
Record what you expect to lose. If the loss turns out bigger, you can reverse the decision with evidence rather than embarrassment.
Review last quarter’s lists first. Start each meeting by checking whether the stops actually stopped and the tests reached a verdict.
What to bring to the board
The board does not need every item. It needs the three biggest decisions, the budget moved between lists, and the risks you have accepted.
Atlassian’s letter shows the tone that works. It states the decision, the expected short-term pain and the reason the company believes the long-term result will be better.
That is the standard for a quarterly marketing review. It should be possible to read your summary and know what will be different next quarter, and why.
Key takeaways
- End every quarterly review with three lists, stop, strengthen and test, and give each item evidence, a decision and an owner.
- Atlassian stopped its Server products with published dates spread over three years, and later explained the short-term revenue effects of its Data Center decision.
- Ask of every activity: if we were not already doing this, would we start it today? Record what you expect to lose when you stop.
Put it into practice
List every recurring marketing activity your team runs, and mark the three that would fail the test 'would we start this today?' Bring them to your next review as stop candidates.
A question for your next leadership meeting
What did marketing stop doing last quarter, and what did we expect to lose by stopping it?
