Software sold on perpetual licenses generates revenue once. The customer owns the software indefinitely with no obligation to pay again. Annual maintenance contracts exist, typically 18-22% of the original license value, but a customer who lapses maintenance keeps the software running. The businesses that solved this stopped selling perpetual licenses, set a hard date for when the old model ended, and accepted two years of revenue recognition headwind to get there.
When a software customer buys a perpetual license, the vendor gets paid once. Three years later, the customer is still running the software -- possibly three versions behind, possibly with lapsed maintenance. The vendor's revenue from that account is zero unless it sells them something new.
This is what the valuation gap between perpetual-license software companies and their SaaS peers reflects. A business with 80% recurring revenue is valued on a multiple of ARR. A business where 60% of revenue depends on license volume in any given year is discounted against that multiple -- and the discount compounds as the installed base ages without converting. Each year the base sits on perpetual licenses is a year where the renewal lever does not exist, the pricing power does not compound, and the revenue visibility that buyers pay a premium for continues to widen against subscription peers.
The transition does not happen automatically. Guidewire's ARR in FY2020 was $514 million, the majority from on-premise perpetual installations serving property and casualty insurers -- among the most risk-averse software buyers in any industry. By FY2025, ARR reached $1.032 billion, a 101% increase at a 15% compound annual rate, with subscription and support gross margin expanding from 66% to 70.2%. The company absorbed years of revenue recognition headwind to get there. A perpetual deal recognizes revenue at close. The equivalent cloud subscription recognizes the same total value ratably over the term. A customer who paid $2 million upfront generates roughly $400,000 of recognized revenue in the first year of migration -- the reported number falls, the economic value does not. -> Guidewire
Three things separated the companies that completed this transition from the ones still running it.
The first was stopping new perpetual sales before the installed base was fully migrated. The default is to run both models in parallel: keep perpetual available for customers who prefer it while offering subscription to new logos and willing renewals. This is commercially reasonable in the short term. Over time, every perpetual deal sold delays the base migration by the length of that customer's lifecycle, and the two-speed installed base becomes a permanent condition rather than a transition state. Guidewire stopped offering perpetual as a meaningful option to new customers and rebuilt its InsuranceSuite for genuine cloud-native deployment rather than a hosted version of the on-premise product. Cloud-hosted is the faster path and the tempting middle ground -- the product ships earlier, the commercial motion looks like subscription, and the transition appears to be happening. The underlying cost structure and revenue recognition mechanics are not materially different from the original model.
The second was migrating the installed base proactively rather than waiting for contracts to expire. Varonis had $499 million in total revenue in FY2023, but only $44 million -- less than 9% -- was SaaS. Rather than waiting for on-premises term subscriptions to reach their renewal dates, Varonis offered existing customers early migration incentives, pulling forward SaaS conversion at the cost of near-term revenue recognition timing. SaaS ARR reached $340 million by Q4 FY2024, representing 53% of total ARR, more than two years ahead of schedule. -> Varonis
The third was setting a public end-of-life date for the legacy model. Varonis announced end-of-life for on-premises subscriptions effective December 31, 2026. PTC set a phased sunset on perpetual licensing across its CAD and PLM portfolio and noted in investor communications that the sunset date was the most important operational forcing function in the six-year transition timeline. Every month the legacy model remains available is a month a subset of customers uses it, a month internal sales teams sell it when it closes faster, and a month the two-speed base widens. -> PTC
Open-source businesses face a structurally different version of this problem. When Amazon Web Services built a managed service using Elastic's open-source code, the threat was a competing product available at no cost through the AWS ecosystem. Elastic's response was a license change: Apache 2.0 to SSPL in January 2021, forcing AWS to fork the project as OpenSearch and stripping the fork of access to Elastic-developed features. The mechanism that made Elastic Cloud worth migrating to was capability rather than pricing -- cloud-only features, including serverless Elasticsearch and AI-powered security analytics, created a widening gap between the managed service and the self-hosted alternative. Elastic Cloud grew from 35% of total revenue in FY2022 to approximately half of total revenue by FY2025, while total revenue grew 72% from $862 million to $1.48 billion. -> Elastic
The failure mode for both types is the same: running the transition indefinitely without completing it. The revenue dip in years one and two is real and visible. The multiple expansion at exit is real and larger. Companies that slow the transition to manage the near-term P&L impact typically extend the pain period without shortening it. The base keeps splitting. The discount to SaaS peers holds or widens. The management team that eventually exits the business faces the same conversion math as the team that started it, with less time on the clock.
The diagnostic for a software business on perpetual licenses: what percentage of total revenue recurs without a new commercial event? For a pure perpetual-license business, the honest answer is the maintenance percentage -- typically 18-25% of total revenue. That is the floor of recurring revenue. Everything above it requires a sale.
The second question is whether the company has a specific date by which it stops selling perpetual licenses to new customers. A transition roadmap without a date is a statement of intent. A date is a forcing function.
The third is the installed base profile: how many customers are still on perpetual licenses, and how far behind are they on current versions? A customer three major releases behind is already approaching a mandatory upgrade conversation. That conversation is easier to control when the vendor initiates it on a defined timeline than when the customer's operational pressure forces it.
Read these alongside the playbook — look for what each company had in common, and where their approaches diverged.
Guidewire doubled ARR to $1.03B via perpetual-to-cloud conversion of 570 P&C carriers over five years.
Guidewire Software Doubled ARR to $1.03 Billion Through Perpetual-to-Cloud Migration for P&C Insurers
Varonis grew SaaS ARR nearly 4x to $340M in one year by proactively migrating its on-premises customer base to cloud.
Varonis Accelerated SaaS ARR by 4x to $340 Million Through a Full Business Model Transition from On-Premises Subscriptions
PTC grew ARR from $1.3B to $2.2B by replacing perpetual CAD and PLM licenses with subscriptions over seven years.
PTC Grew ARR to $2.2B at 15% CAGR by Completing CAD and PLM Perpetual-to-Subscription Migration
Elastic grew total revenue 72% to $1.48B by converting open-source users to Elastic Cloud subscriptions.
Elastic Grew Total Revenue 72% from $862M to $1.48B by Converting Open-Source Elasticsearch Users to Enterprise Cloud Subscriptions