It wasn’t even 30 days after I left Adobe after their acquisition of my last start-up EchoSign / Adobe Sign when … they ripped out the Free edition. It had millions of users over time, was cheap to provide and simple to use, and materially helped with top-of-funnel, at least for smaller prosumer customers.
But sales always sort of hated the Free edition. Most never paid, and few every upgraded to Enterprise accounts. And when sales called into them, it always was a waste of time.
So the new GM that replaced me when I left? Weeks later, he ripped Free out. I am sure it cut down noise in support and the low end of sales. And it also led to the beginning of a decline in brand awareness and more.
Atlassian just did the same with Loom after buying it for almost $1B a few years back.
Raise prices, then kill Free, at least some of it. Atlassian is now forcing everyone to pay for viewer seats (“Creator Lite”) that used to be Free.
The version of this on LinkedIn is “Atlassian doubled the price of Loom.” Or more. Sort of. That’s the least interesting part of what happened, and for a lot of workspaces it isn’t even true.
Atlassian changed who counts as a customer. Loom had a free seat type called Creator Lite: people inside a paid workspace who could watch, comment, and record in a limited way at no cost. As part of the Atlassian integration, that role is eliminated and every one of those users is converted to a full paid Creator seat automatically.
From Atlassian’s own support documentation: the Creator Lite role is discontinued, all Creator Lite users are upgraded to full paid Creator status on your integration date, and you get a grace period until your next billing date to deactivate anyone you don’t want to pay for. Miss it and they’re on the invoice.
A workspace with 10 recorders and 90 watchers used to pay for 10. Now it pays for 100.
Worth being precise here, because Loom’s free Starter plan still exists at $0, with 25 recordings and a 5-minute cap. What got deleted is the free seat inside a paying account. That’s the population that mattered, and it’s the population that was never going to sign up for their own separate account to keep watching their coworker’s videos.
{“model_id”: “unified-v1/prod/20260724-010701”}
The free seats were the funnel
Loom’s loop was simple. One person records something, twenty people watch it, three of them decide that was easier than a meeting and start recording. The watchers were free because the watchers were the top of the funnel. Loom reached 25 million users on that loop, with business users recording close to 5 million videos a month, and Atlassian paid roughly $975 million for it in late 2023.
Charging for the watchers converts a growth loop into a collections problem. The rational admin response is not to pay for 90 seats. It’s to deactivate 85 of them. Those 85 people don’t stop needing to send video. They go find something that doesn’t bill for watching.
The LinkedIn thread that surfaced this had a customer success leader saying they cancelled Loom over it. The very next comment was someone else asking what they’d switched to.
Figma had the identical problem and made the opposite call. But at Figma, it’s closer to the core.
Figma restructured its entire seat model in March 2025, in the same window Atlassian was planning the Loom migration. Prices went up on Full seats. The people who weren’t editing got treated differently: Figma created a free View seat with view and comment access across Design, Slides, and FigJam, plus a $3/month Collab seat for stakeholders who need to participate in FigJam and Slides. Figma’s own migration table shows free viewer and viewer-restricted seats becoming the free View seat.
Figma’s documentation also states that new users automatically join with a free View seat, and that admins get upfront approval over seat upgrades that incur cost, by default.
The two defaults:
Figma: a new person lands on a free seat. If they need a paid one, an admin has to approve the charge.
Atlassian: an existing free person becomes a paid seat automatically. If you don’t want the charge, an admin has to find them and remove them before the invoice.
Same problem, same year, opposite default. Figma raised prices and still priced the viewers at zero or three dollars, because Figma understands that the stakeholder who comments on a design file is how the design file spreads through the company. Miro made the same call, keeping visitors free on paid plans and guests free on Business and above.
Atlassian looked at the same population and priced them at $15 to $24 a head. Loom isn’t Atlassian’s core product, however. That might be the explanation right there.
Mailchimp, Slack, HashiCorp: same move, same clock
We wrote last week about Mailchimp shrinking inside Intuit. The free-tier mechanic there is the same story on a longer clock.
Mailchimp’s free plan carried 2,000 contacts as recently as 2022. It dropped to 500 in 2023. Multi-step automation was stripped from the free tier in mid-2025. Then in January 2026 Mailchimp cut it again, to 250 contacts and 500 monthly sends, effective February 17. That’s roughly an 87% reduction in four years, and it came alongside a legacy-plan price increase in April for accounts created before May 2019 that never migrated.
MailerLite followed, cutting its free plan from 1,000 subscribers to 500 in September 2025.
Mailchimp got to $1B+ in revenue on a free tier that was generous enough to be the entire acquisition channel for a generation of small businesses. Under Intuit that free tier has been narrowed almost annually, and Intuit has not disclosed a Mailchimp revenue figure since Q4 FY22.
Slack ran the clearest version of it. Salesforce completed the $27.7B acquisition on July 21, 2021. Twelve months later, on July 18, 2022, Slack announced that free workspaces would lose the 10,000-message searchable history and 5GB of storage, replaced by a flat 90 days, effective September 1. Pro pricing went up in the same announcement. Slack framed it as most active free teams gaining access to more history. What small teams, open source projects and community groups actually got was a rolling three-month memory wipe, and plenty of them left.
It crosses categories, too. IBM closed its $6.4B acquisition of HashiCorp in February 2025. Thirteen months later, on March 31, 2026, HCP Terraform’s legacy free plan ended, with the replacement capped at 500 managed resources.
The timing is what should get your attention:
- Salesforce / Slack: 12 months from close to the free-tier cut
- IBM / HashiCorp: 13 months
- Intuit / Mailchimp: roughly 18 months to the first cut, then again, and again
- Atlassian / Loom: 26 months, and the most aggressive version of the move
Four different acquirers running the same playbook inside a two-year window.
I watched this happen to my own company
I don’t have a neutral read on this, because I lived it.
Adobe acquired EchoSign in 2011. We had a free edition, and it did what free editions do: it put e-signature in front of people who would never have taken a sales call, and a real share of them became paying accounts later. Adobe killed the free edition within weeks of my leaving.
The logic then was identical to the logic now. Someone models the free base, sees a large population generating no revenue, and books the conversion. What the model can’t show is the deals that never start because the free edition wasn’t there to start them. That number is invisible on the day you make the decision and obvious three years later.
Why? Seat counts are under pressure everywhere in the Age of AI
This is arithmetic more than greed.
Seat counts are under pressure everywhere and every seat-priced vendor knows it. When headcount stops growing, per-seat revenue stops growing with it, and the vendor gets punished hardest when its product works best. That sends everyone looking for seats that already exist but aren’t billed: free viewers, guests, view-only users, Creator Lites. It’s the last untapped pool inside accounts you already have, and unlike new business it requires no pipeline and no product work.
Atlassian’s own reported numbers show where this lands. In Q2 FY26 they said revenue growth was primarily driven by paid seat expansion, higher ARPU, and cross-sell. Q3 FY26 came in at $1.787B, up 32%, with Cloud up 29% and non-GAAP operating margin at 34%. Converting 90 free Loom viewers into 90 paid Loom seats is paid seat expansion and higher ARPU in one move, executed by policy rather than by selling.
They report Q4 and full-year FY26 on August 6. It won’t be broken out by product, but the seat and ARPU commentary is where this shows up.
Do the new features earn it?
Atlassian’s stated justification is that Loom has shipped a lot, which is worth testing.
They have shipped. AI-generated titles, summaries and chapters. Auto meeting recaps and action items. Video-to-text that creates Jira issues and docs. Video variables for personalization at scale. A full meeting notetaker with an AI agent, the Rewatch acquisition folded in, multi-language transcription, data residency, SCIM. That is not a neglected product. Unlike Intuit with Mailchimp, Atlassian is actively investing.
Three problems with it as a defense of this particular change.
First, nearly all the AI sits behind Business + AI at $24, which is already a separate upsell. The features justify that tier existing. They don’t justify moving base-tier customers to list.
Second, Atlassian’s own document undercuts the argument. It says they made pricing adjustments in 2024 that reflected the substantial innovation, and that this is the first increase for the existing base. The pre-2026 feature work was already priced once.
Third, and most damaging: the features are irrelevant to the people actually being billed. A Creator Lite user was by definition someone who wasn’t recording much. Atlassian frames the forced conversion as a benefit, noting that upgraded users now get unlimited recording and AI access. You can’t justify charging someone $180 to $288 a year on the grounds that they now have a meeting notetaker they never asked for and weren’t using.
The bet Atlassian just made
They converted a distribution asset into a revenue asset. Those are different things, and the conversion only runs one direction.
The bull case is that Loom is embedded enough in Jira and Confluence that the loop no longer matters, the seats stick, and this shows up as clean ARPU expansion for years. The bear case is that they made async video a visible line item in a year when every CFO is auditing visible line items, while Figma and Miro keep handing out the exact seats Atlassian just started billing for.
Mailchimp’s version of this bet took most of a decade to resolve. Loom’s will resolve faster, and the first read comes at the renewal cohort twelve months out.
