How to unlock true ROI in software development - a deep dive into the latest DORA research

Overview

Google Cloud's DORA research team says the return on AI coding tools follows a J-curve: a short productivity dip during early adoption, then a climb to long-term gains. The report's central finding is that AI works as an amplifier, so the biggest returns come from strong engineering foundations and team workflows rather than the tools alone. Nearly all surveyed teams already use AI at work, yet financial results differ widely depending on how well an organization is set up to absorb the change.

Key Takeaways

AI value tends to dip before it rises, a pattern the report calls the J-curve, driven by learning time, extra code review, and pipeline bottlenecks.
The largest returns come from the surrounding organizational system, including platform quality and clear workflows, not from the AI tool by itself.
90% of DORA survey respondents already use AI at work, but actual financial impact varies sharply between companies.
AI raises the volume of code produced, which adds a verification cost as teams review outputs to prevent errors and meet standards.
A free interactive ROI calculator at dora.dev lets leaders model both visible costs and hidden adoption realities for their own teams.
In a sample 500-person engineering organization, the model showed a 39% first-year ROI with a payback period of roughly 8 months.

Stats & Key Facts

#90% of DORA survey respondents report using AI at work
#39% modeled first-year ROI for a sample 500-person engineering organization
#$8.4 million first-year investment in that same modeled scenario
#$11.6 million first-year value generated in that scenario
#Payback period of roughly 8 months in the modeled example
#Change failure rate rising from 5% to 6% after adoption, tied to about $344,000 in negative downtime cost

How to unlock true ROI in software development - a deep dive into the latest DORA research

Why AI coding returns follow a J-curve before they climb

The report's first insight is that financial value from AI rarely arrives in a straight line.

Most organizations hit what DORA calls a J-curve: a temporary drop in productivity and a period of instability tied to early adoption. The research frames this dip as a normal part of taking on new technology, not a sign that the strategy is failing.

Leaders who plan for this slowdown can keep their AI projects funded and moving, treating the early friction as an investment in long-term team speed. Budgeting for this learning phase is described as the key to making the transition work.

The three forces behind the early productivity dip

DORA points to three specific reasons the J-curve happens.

›Learning curve: teams need dedicated time away from feature delivery to rework daily habits and move from simple prompting toward building systems around context and intent.
›Verification tax: because AI sharply increases the volume of code, developers spend extra time reviewing outputs to confirm trustworthiness, prevent fabricated results, and meet internal architecture standards.
›Pipeline adaptation: as developers produce code faster, downstream steps like testing and change approvals turn into bottlenecks and must be scaled up to handle the higher throughput.

Near-universal adoption, but very different financial results

The second insight is that using AI and profiting from AI are not the same thing.

DORA's state of AI-assisted software development research found that 90% of survey respondents report using AI at work. Despite that near-universal adoption, the financial impact splits across the market. Some companies see clear value from their engineering investments while others run into unexpected costs.

When a project falls short, the report says the cause is often a lack of organizational support rather than the technology itself. Getting the expected returns depends on preparing workflows and teams to absorb the new tools before scaling them.

AI as an amplifier: foundations decide the payoff

The report's central argument is that AI magnifies whatever a company already has.

According to DORA team lead Nathen Harvey, the greatest returns on AI investment come not from the tools themselves but from a strategic focus on the underlying organizational system. That system includes the quality of the internal platform, the clarity of workflows, and how well teams are aligned.

In practice, AI amplifies both strengths and weaknesses. Strong version control practices, a solid internal platform, and AI-accessible internal data flow into better delivery metrics, which then improve the developer experience and finally produce cost savings and revenue growth.

Where AI adds measurable value across the lifecycle

Building a realistic financial model starts with finding where AI actually adds value.

›Reducing costs across the software development lifecycle.
›Increasing developer productivity and output.
›Improving security of the code and the delivery process.
›Delivering a better experience for both developers and end users.

The research also names an instability tax. As AI lifts individual effectiveness, the rising volume of code strains deployment pipelines and correlates with more software delivery instability until processes are scaled to match.

Modeling your own ROI with the DORA calculator

DORA pairs the report with an interactive tool so leaders can build a defensible business case.

The free ROI calculator at dora.dev/ai/roi/calculator lets a team forecast both the visible expenses and the hidden realities of AI adoption. Leaders can adjust the assumptions to match their own organization and build a custom estimate.

In one modeled example for a 500-person engineering organization, the framework showed an $8.4 million first-year investment producing $11.6 million in value, a 39% first-year ROI, and a payback period near 8 months. The same model showed change failure rate rising from 5% to 6%, tied to roughly $344,000 in negative downtime cost, a reminder that quality trade-offs belong in the math.

Frequently Asked Questions

What is the J-curve in AI software development?

It is the pattern where productivity dips temporarily during early AI adoption before rising to long-term gains. DORA attributes the dip to the learning curve, the cost of verifying AI-generated code, and the need to scale testing and approval pipelines.

Why do some companies see strong AI returns while others do not?

DORA found that 90% of respondents already use AI, but results vary because returns depend on the organization's foundations, not the tool alone. Companies with strong platforms, clear workflows, and aligned teams capture more value.

What is the verification tax?

It is the extra time developers spend reviewing AI-generated code. Because AI produces far more code, teams must check it for trustworthiness, prevent fabricated outputs, and confirm it meets internal architecture standards.

How can a business estimate its own AI ROI?

DORA offers a free interactive calculator at dora.dev/ai/roi/calculator. It lets leaders model visible costs and hidden adoption realities, adjust assumptions to their situation, and produce a custom return estimate.

What value did the report's sample model show?

For a modeled 500-person engineering organization, the framework showed a 39% first-year ROI, an $8.4 million investment generating $11.6 million in value, and a payback period of about 8 months.

The DORA research reframes AI ROI as a question of organizational readiness, urging leaders to budget for an early productivity dip and to strengthen the foundations that turn AI into lasting value. Pairing the report with a free calculator gives teams a way to model that case before scaling their AI investment.