Salesforce delivery has changed more in the last twelve months than in the five years before. AI has moved from pilot to production, the platform is opening up in ways we haven't seen since the introduction of Lightning, and the pace of change is no longer set by release windows but by how fast we can plan, validate, and govern what AI helps us build.
The data in this year's report points in one direction: DevOps is no longer a delivery concern, it鈥檚 the foundation for everything that comes next. The teams furthest along on Salesforce's Agentic Maturity Model are, without exception, the teams with tools and process across the full DevOps lifecycle. That鈥檚 not a coincidence. You can鈥檛 operate what you can鈥檛 observe, and you can鈥檛 scale agents on a stack you can鈥檛 see into.
At the same time, the platform itself is becoming more open and more composable. Headless capabilities, a richer component model, and partnerships with frontier AI providers like Anthropic and OpenAI bring extraordinary new possibilities, and an equally new governance challenge. Every new surface is a new release vector. Every new capability needs an architecture, a test strategy, and an owner.
The most interesting shift, in my view, is where the work moves. Build has been the longest part of our delivery cycle for years, and the report confirms it still consumes around half of team time. AI changes that. Build compresses. Plan and Validate become the bottlenecks, and the quality of our architecture decisions and our review discipline starts to matter far more than the speed of our keyboards.
For Salesforce architects and delivery leaders, the message is clear. End-to-end DevOps is no longer the destination, it鈥檚 the starting line. The teams who treat it as such will be the ones shaping what an agentic enterprise actually looks like on the Salesforce platform.
What this means in practice is that DevOps has quietly become the most strategic capability a Salesforce organization can invest in. It鈥檚 the discipline that turns AI from a productivity experiment into a reliable production capability. It鈥檚 the control plane that lets us open the platform up without losing our grip on quality, security, and trust. And it鈥檚 the muscle that decides whether the next wave of innovation lands as business value or as technical debt. The Salesforce teams who internalize this in 2026 will not just deliver faster. They will be the ones their organizations turn to when the question is no longer "can we build this?" but "can we trust it in production?" That is the conversation worth being ready for.
Salesforce DevOps adoption and performance
Share
Manual, error-prone releases used to define Salesforce delivery. Now version control and release automation are the baseline, bringing speed and stability to every release. And data from our State of Salesforce DevOps 2026 survey shows that the teams performing best aren't stopping there 鈥� they're adopting tooling and process across the full DevOps lifecycle.
Salesforce teams鈥� adoption of the DevOps lifecycle
Across all stages of the DevOps lifecycle 鈥� plan, build, validate, release, operate, observe 鈥� most teams feel they have the tools and processes in place. But teams are consistently more likely to have adopted tooling than to have defined the processes around it. The gap is widest at the 鈥淥ps鈥� end of the lifecycle 鈥� 10% of Salesforce teams have no tooling or process for the operate stage, rising to 17% for observe.
We asked teams about their level of confidence in each stage of the DevOps lifecycle. Not surprisingly, confidence levels correlate closely with adoption of tools and process. The observe stage is where teams are least confident.
Tools and processes across the DevOps lifecycle
Do you have tools and/or defined processes for each stage of the DevOps lifecycle?
Salesforce teams are most confident about the build stage 鈥� and it鈥檚 where they spend up to 50% of their time. As DevOps adoption has reduced the amount of time spent on releases, the build phase has emerged as the biggest remaining opportunity for efficiency gains.
Time spent per DevOps lifecycle stage
Which stage of the lifecycle does your team spend the most time on?
Salesforce DevOps performance (DORA metrics)
To benchmark delivery performance, we use the four key metrics defined by , Google鈥檚 long-running research program into software delivery performance. The four metrics are:
Deployment frequency. A measure of how often changes are deployed to production.
Change lead time. The time it takes for code changes to go from being committed to being deployed in production.
Change failure rate. The percentage of deployments that result in a failure.
Time to recover. The time it takes to recover from a failed deployment.
Together, the DORA metrics capture two dimensions of delivery performance: throughput and stability.
Release frequency continues to follow a bell curve centered on weekly and multiple times a week, with no significant shift toward daily deployments compared to last year. Lead times similarly cluster in the middle range for the majority of teams.
Most teams keep errors to a minimum. But a significant minority release bugs in more than 10% of releases. What鈥檚 more, 18% of teams primarily find issues in production 鈥� in the operate and observe stages of the DevOps lifecycle. Recovery times are strong: nearly two thirds of teams restore normal service within six hours of a production incident.
Deployment frequency
How frequently does your organization release to production?
Change lead time
How long does it usually take to release a new feature to production after it has been built?
Change failure rate
What percentage of releases include a bug or error?
Recovery time
On average, how long does it take your team to restore normal service after a production incident or failed deployment?
Expert opinion
At Zurich, we use DORA metrics to provide a consistent, data-driven view of DevOps performance across our Salesforce teams.
By tracking lead time, deployment frequency, and recovery, we are helping teams identify bottlenecks and prioritise improvements, regardless of their size or complexity. These metrics are increasingly central to how we drive continuous improvement across the DevOps lifecycle.
Looking ahead, we are exploring how AI can enhance this further ensuring that any innovation results in measurable gains against the core DORA metrics.
Agnieszka Oliver
Salesforce and Mulesoft Platform Lead
Zurich Insurance
Scaling Salesforce delivery
Teams face a variety of blockers to scaling delivery on the Salesforce platform, with no clear leader emerging. The average team faces the majority of these challenges all at once. But these are the very hurdles that DevOps adoption helps teams overcome 鈥� unlocking Salesforce鈥檚 potential for the business and helping teams deliver more, faster.
Biggest blockers to scaling Salesforce delivery
What are the biggest blockers to scaling Salesforce delivery in your organization? (select all that apply)
The increasing returns of DevOps adoption
This year, for the first time, we鈥檝e examined the relationship between the breadth of DevOps lifecycle adoption and performance outcomes 鈥� and the findings are unambiguous.
Teams with high lifecycle adoption outperform teams with low adoption on every metric we measured. They ship fewer bugs, fix problems faster when they occur, and restore service more quickly after incidents. Critically, the improvement isn鈥檛 just a jump from low to partial adoption: teams that go all the way across the lifecycle continue to see gains over those who stop halfway.
Returns per number of lifecycle stages adopted
Teams with an error rate of less than 5%
Teams who can restore production in less than 6 hours after an incident or failure
Average annual rework hours
Average annual production downtime hours
This finding reframes the DevOps maturity conversation. It鈥檚 not enough to have version control and a deployment pipeline. The teams pulling ahead are those treating DevOps as an end-to-end discipline 鈥� one that extends from planning through to observing what happens after release.
High adopters of the full lifecycle are four times more likely to feel extremely confident on release day, showing the positive knock-on impact of having every stage of the lifecycle in place and optimized.
The business value of Salesforce DevOps
Share
98% of survey respondents recognize ROI for Salesforce DevOps. At the same time, only 50% of teams have calculated their dollar returns.
Estimated monthly ROI from Salesforce DevOps
What's your estimated monthly return on investment from Salesforce DevOps?
Company size appears to play a role in whether teams attempt to measure ROI. Smaller businesses are more likely to have calculated their monthly ROI 鈥� often a simple tally of hours saved on release management thanks to automation. Larger businesses typically see the value of DevOps more in terms of quality than release velocity, and quality is harder to calculate. The productivity gains for DevOps at scale are clear enough, and the governance for delivering change becomes a non-negotiable requirement.
Expert opinion
At a smaller org, DevOps ROI is basically a stopwatch counting the hours your team isn't losing to deployments. At enterprise scale that math is just way harder, because the real return is in things that didn't happen. Outages you avoided. Audits that passed without a fire drill. People who didn't quit. None of that fits cleanly on a spreadsheet, which is probably why most large teams don't bother, but it's the part that actually matters.
Daniel Barckley
Senior Salesforce Solution Architect
The Wharton School of the University of Pennsylvania
Salesforce and AI
Share
The Agentforce opportunity
Salesforce remains focused on helping its customers capitalize on the potential of AI to become an agentic enterprise. Agentforce itself continues to evolve, for example, with the recent release of Agent Script to complement Agent Builder. But teams also need to be able to reliably test and release agents before they can experiment properly and begin building confidence with the business.
We asked where teams placed themselves on , and the picture is one of a community still finding its footing.
Salesforce Agentic Maturity Scale
Where is your team on the Salesforce Agentic Maturity Scale?
Agents successfully deployed
How many agents have you successfully deployed?
While they remain a minority, there are teams successfully delivering Agentforce agents. Notably, the teams at the top of the Agentic Maturity Model all have tools and/or process for every stage of the DevOps lifecycle.
AI-assisted and agentic Salesforce delivery
There鈥檚 a growing appetite for AI-assisted and agentic delivery on the Salesforce platform 鈥� and especially for development. Right now, teams are slightly leaning more towards ChatGPT as an option, but it's still a relatively level playing field across all of the AI tools used in development.
AI tools used for Salesforce development
Which AI tools are you using for Salesforce development? (select all that apply)
This reflects the pace of change in the AI market more than any settled preference. Teams are experimenting widely, and most organizations don鈥檛 yet have a standardized AI development workflow.
The increased volume of changes created with AI poses a governance challenge. How should AI-generated code be reviewed? The most common answer is that it isn鈥檛 treated differently from human-written code at all. So it鈥檚 important that any existing governance framework is robust enough to cope with the increased volume that AI-generated code delivers.
Review of AI-generated vs human-written changes
Do you review AI-generated code and configuration differently to human-written changes?
Building confidence in AI for Salesforce
The Salesforce community isn鈥檛 polarized by AI: the majority have middling levels of trust, and are building confidence in new capabilities.
Confidence in AI-generated code and configuration
How confident are you in AI-generated code or configuration in these areas?
Trust in AI across the DevOps lifecycle
How much do you trust AI involvement at each stage of the DevOps lifecycle?
The implication is significant: most teams are ready to use AI as a tool that amplifies developer capability; fewer are yet forging ahead to the world of agentic delivery.
Security and compliance remain the most commonly cited barriers to broader AI adoption, followed by data quality and budget. Notably, 鈥渓ack of a clear use case鈥� 鈥� a significant hindrance in last year鈥檚 survey 鈥� is declining as a concern. Teams increasingly know what they want from AI, and now need the guidance and tooling to make their ambitions a reality.
Biggest barriers to adopting AI
What is your team's biggest barrier to adopting AI?
Expert opinion
Confidence in AI is growing, but the jump to agentic delivery requires a level of process maturity that most organizations are still building. AI earns its keep earliest in the pipeline 鈥� in planning and requirements refinement, where feedback loops are faster and the 'blast radius' of a mistake is smaller.
Organizations that move away from 'AI for AI's sake' and instead provide clear strategic frameworks will be best positioned for success 鈥� making AI a genuine force multiplier for DevOps rather than just another layer of complexity.
Analissa Moreno
Salesforce Release Manager
GitLab
Outperform the stats with 91导航
Share
91导航 customers once again outperform the field across all key metrics.
Salesforce teams using the 91导航 end-to-end platform are:
more likely to release multiple times a week or more
more likely to restore service within half a day
more confident about releasing to production
Discover how 91导航 can help your team
Learn how 91导航 empowers your team to achieve superior speed, quality, and security throughout the DevOps lifecycle, all while ensuring robust governance for your AI initiatives, by scheduling a demo or starting a free trial today.
This year, the State of Salesforce DevOps survey involved 522 respondents who gave quality-controlled answers. 48% of respondents were 91导航 users.
Survey respondents by region
Survey respondents by region
Survey respondents by role
Survey respondents by role
Survey respondents by industry
Survey respondents by industry
Survey respondents by organization size
Survey respondents by organization size
Survey respondents by size of team
Survey respondents by size of team
Ready to get started with 91导航?
Full access for 30 days
No credit card required
Nothing to install in your orgs
What makes Salesforce teams successful with delivery, quality control, and AI?
We surveyed hundreds of Salesforce teams about their DevOps practices, delivery performance, and AI adoption. The pattern was clear: success isn't about which AI tools teams are using, it's about the DevOps practices underneath.
However, too many teams still don't catch bugs until production, and while 98% see ROI on DevOps, only half can actually prove it.
Unlock the report to see:
Where teams place confidence in AI 鈥� and where they don't
Why full lifecycle coverage translates to 4x more confidence
The gap between seeing DevOps value and proving it
Benchmark data on deployment frequency, bug rates, and recovery times
