Digital Product Engineering: First Release to Scale
Digital product engineering from first release to scale: lifecycle stages, DORA metrics, AI evidence, 2025-2028 compliance dates and a partner checklist.
By Abhineet Jain, Product HeadDigital Product Engineering, Product Engineering Services, DORA Metrics, Cyber Resilience Act, AI in Software Development
Digital product engineering means one team designs, builds and keeps improving a software product for as long as it is in the market, from discovery and the first release through scaling and re-engineering. Project-based development ends at handover. A product team stays accountable for results after launch, ships in small increments and keeps changing the product as users and regulations change.
What is digital product engineering?
Product thinking, engineering and continuous delivery
Companies often buy these three disciplines from different suppliers. Product engineering puts them in one team. Product thinking decides which problem to solve, for which users, and how you will know a release worked. Engineering supplies architecture designed to change, automated tests, secure development and integration with existing systems. Continuous delivery (Agile planning, DevOps practices and CI/CD pipelines) lets changes ship in small batches so problems surface early.
A product is built for a market and keeps evolving after launch, so the work has no fixed end date. 3Pillar draws the same contrast with project-based development.
Product engineering vs software development
| Software development project | Digital product engineering | Product engineering services | |
|---|---|---|---|
| Goal | Deliver an agreed scope | Improve a product’s outcomes over time | Buy product engineering capacity or a full team from a partner |
| Timeline | Fixed end date | Ongoing, release after release | Contract term, often staged |
| Funding | Per project | Persistent team budget | Per team, per phase or per outcome |
| Success measure | On time, on budget, to specification | Adoption, retention, revenue and delivery health | Agreed delivery metrics, quality and handover terms |
| After launch | Handed to a maintenance team | Same team keeps evolving it | Stays with the partner, moves in-house, or transfers under build-operate-transfer |
What a digital product engineering services engagement includes
A full engagement usually spans:
- discovery and user research
- UX design
- solution architecture
- build
- QA and test automation
- DevOps and release management
- run-and-support after launch
Smaller engagements take one slice, such as an MVP or a platform upgrade.
Benefits, and the challenges that slow teams down
When the model works, small releases get value to users sooner, and each release is checked against how people actually use the product. Architecture built for change can grow without a rewrite. New revenue lines become possible, and weak ideas get dropped while dropping them is still cheap.
Experienced engineers are hard to hire, and demand is still rising: the US Bureau of Labor Statistics projects software developer employment to grow 15.8%, about 267,700 jobs, between 2024 and 2034. Legacy systems complicate integration, and technical debt piles up when teams ship without refactoring. Security work tends to arrive late. Organisations also underestimate how much a persistent product team changes funding and reporting.
The phases of digital product engineering, from first release to continuous delivery
Published guides describe the lifecycle in five, seven or nine steps (GlobalLogic, Cyntexa, Harbinger). The content barely differs: discovery, design and prototyping, development, QA, deployment and support, with re-engineering sometimes listed separately. What they rarely say is how spending, metrics and controls should change as a product matures. Highland Solutions gets nearest, with the argument that you should commit more engineering only as evidence for the product accumulates. Applied to metrics, controls and team shape, that idea gives five stages.
Discover and validate
Test the problem and the riskiest technical assumption with as little engineering as you can. If the product will handle personal data, design consent and data minimisation now. The same goes for accessibility: adding it to a finished product costs far more than building it in.
First release: MVP development on foundations you keep
The MVP is the smallest version a customer can actually buy or use. Its scope should be narrow, but its foundations are the ones you will keep: version control, automated tests, a CI/CD pipeline and a secure-development baseline. Record delivery metrics from the first deployment.
Scale-up: architecture, platform and automation
Once usage is real, harden the architecture, add observability and form a platform team so product teams stop rebuilding pipelines and environments. Regulatory duties such as vulnerability reporting need named owners by now.
Continuous delivery at scale
Releases should be frequent and dependable, running through a self-service internal platform and judged on delivery health, reliability and cost. Long-term ownership often moves in-house or to a capability centre at this point.
Modernise and re-engineer
Over time, some components start to get in the way of every change. Re-engineering replaces or refactors them, and is the moment to re-check accessibility and security support commitments. Processes that have outgrown spreadsheets need the same treatment. We built a web-based ERP for a manufacturing supplier that had been running purchase orders, challans and invoices on disconnected Excel files. It runs on Python and Next.js with a centralised database, a four-step reconciliation workflow, validation and approval controls, and built-in analytics. The process is now fully digitised, with real-time reporting and a complete audit trail.
| Stage | Engineering goal | What to measure | Controls to build in | AI use (evidence-led) | Typical team model |
|---|---|---|---|---|---|
| 0. Discover and validate | Test the problem and riskiest assumptions; spend little on build | Learning speed, prototype feedback | Consent notices and data minimisation if personal data is involved (DPDP Rules); accessibility in design (EAA) | Prototyping and research; low risk | Small product trio, in-house or partner |
| 1. First release (MVP) | Smallest sellable or usable product, on foundations you keep | Baseline DORA metrics | Secure development (NIST SSDF v1.2 draft); vulnerability handling ready before EU launch (CRA) | Test the speed-up: experienced developers were 19% slower in METR’s early-2025 trial | Partner pod or in-house core |
| 2. Scale-up | Harden the architecture; add platform and automation | DORA metrics plus product outcomes | CRA reporting: 24h early warning, 72h notification, final report (European Commission); AI Act Article 50 transparency from 2 Aug 2026 (CSA) | AI raises throughput but hurts stability unless testing, version control and feedback loops are strong (DORA 2025) | Partner plus a growing in-house team; platform team |
| 3. Continuous delivery at scale | Frequent, reliable releases; self-service platform | DORA metrics, reliability, cost | Full CRA obligations from 11 Dec 2027 (European Commission); high-risk AI dates 2 Dec 2027 and 2 Aug 2028 (Council of the EU) | Platform quality shapes AI value; 90% of organisations have at least one platform | In-house, or a GCC under build-operate-transfer |
| 4. Modernise and re-engineer | Replace or refactor parts that block change | Technical-debt trend, change failure rate | Re-check WCAG conformance and the CRA support period | Code comprehension and migration, with human review | Mixed |
How to measure a product engineering team
The four DORA delivery metrics
DORA’s research programme built its view of software delivery on four metrics. Deployment frequency is how often you push changes to production, and lead time for changes is the time from a commit to that code being live. Change failure rate is the proportion of releases that break something and need a rollback or fix. Time to restore service is how quickly the team recovers once something breaks.
The first pair tells you about speed and the second about safety. If speed improves while safety slips, the team is moving faster than it can check its work.
A stage-by-stage scorecard
A team can post excellent delivery numbers while shipping features nobody wants. Pair delivery metrics with product outcomes such as adoption, retention and revenue, and weight them by stage. During discovery and validation, track learning speed and prototype feedback. A first release needs a DORA baseline and its first active users. At scale-up, watch DORA trends alongside adoption and retention. Continuous delivery at scale adds reliability targets and cost to the DORA metrics. When modernising, follow the technical-debt trend and the change failure rate on the components being replaced.
AI in digital product engineering: the evidence from 2025 and 2026
Adoption is nearly universal, trust is lower
Google Cloud’s 2025 DORA report found that 90% of nearly 5,000 respondents use AI at work and more than 80% believe it raised their productivity. In the same research, 30% report little or no trust in AI-generated code. Stack Overflow’s 2025 survey found that 84% use or plan to use AI tools (76% in 2024) and 46% do not trust the accuracy of AI output (31% in 2024). The two trust figures come from different questions and different samples, so treat them as separate signals.
Felt vs measured productivity
METR’s randomised trial points the other way. Sixteen experienced open-source developers working on 246 tasks were 19% slower with AI tools, after expecting a 24% speed-up, and afterwards believed they had been 20% faster.
METR’s February 2026 update, using late-2025 tools, estimated task-time changes of −18% (confidence interval −38% to +9%) for returning developers and −4% (−15% to +9%) for new recruits. A negative figure means faster. METR still calls this weak evidence because of selection effects and is redesigning the study. Measure the effect on your own team before you plan around it.
Why platforms and testing decide whether AI helps
DORA also found that AI adoption still has a negative relationship with software delivery stability. More code arrives faster, and teams without solid tests, version control and quick feedback pay for it in failed changes.
Platforms matter for the same reason. DORA reports that 90% of organisations have adopted at least one platform. Gartner forecasts that 80% of large software engineering organisations will have platform teams by 2026, up from 45% in 2022. The two numbers count different things: DORA counts any platform in use, Gartner counts dedicated teams.
A sensible rollout trials AI tools on one team with DORA metrics recorded before and after, requires human review of AI-generated code in security-sensitive paths, and strengthens tests and pipelines before AI use widens.
Compliance dates for product teams, 2025 to 2028
| Date | What applies | Who it affects | Source |
|---|---|---|---|
| 28 Jun 2025 | European Accessibility Act requirements | Covered products and services sold in the EU, such as e-commerce and banking | EUR-Lex, Travers Smith |
| 14 Nov 2025 | India DPDP Rules notified, with 18-month phased compliance | Organisations processing personal data in India | PIB/MeitY, PIB backgrounder |
| 17 Dec 2025 | NIST SSDF v1.2 (SP 800-218r1) initial public draft | US federal software suppliers; a wider secure-build baseline | NIST |
| 20 Apr 2026 | DOJ interim final rule moves ADA Title II web and app deadlines back one year | US state and local governments and their contractors | Reed Smith, Duane Morris |
| 11 May 2026 | HHS Section 504 web and app accessibility, first compliance date (not extended) | Organisations receiving HHS funding | Duane Morris |
| 2 Aug 2026 | EU AI Act Article 50 transparency obligations | AI-facing product features | Cloud Security Alliance |
| 11 Sep 2026 | CRA reporting obligations: 24h early warning, 72h notification | Manufacturers of products with digital elements in the EU | European Commission |
| 2 Dec 2027 | EU AI Act high-risk obligations (stand-alone, Annex III) | High-risk AI providers | Council of the EU |
| 11 Dec 2027 | CRA main obligations | Manufacturers of products with digital elements in the EU | European Commission |
| 2 Aug 2028 | EU AI Act high-risk obligations for AI embedded in products | Product manufacturers | Council of the EU |
EU Cyber Resilience Act: reporting is already live
Since 11 September 2026, manufacturers must report actively exploited vulnerabilities and severe incidents through the CRA Single Reporting Platform in three steps: an early warning within 24 hours, a notification within 72 hours and a final report (European Commission). Every product line sold in the EU therefore needs:
- a working vulnerability-handling process
- a named on-call owner
- a defined support period
Accessibility: EAA, ADA Title II and the WCAG gap
The ADA Title II extension moved the original dates of 24 April 2026 and 26 April 2027 back by one year (Reed Smith). Most sites are far from ready: the WebAIM Million found detectable WCAG failures on 95.9% of the top one million home pages in 2026 (94.8% in 2025), averaging 56.1 errors per page. Run automated accessibility checks in CI and add manual review to design sign-off.
EU AI Act transparency and high-risk dates
The Council’s 7 May 2026 release described the high-risk date changes as a provisional agreement. Later reporting says it received final approval on 29 June 2026. Check the Official Journal text before you fix roadmap dates.
India DPDP Rules
Eighteen months from notification puts the end of the phased window at roughly May 2027. That date is our own calculation, so treat it as an estimate. Products with Indian users or Indian data processing need consent notices and data minimisation designed in from the start.
NIST SSDF as a secure-build baseline
Public comments on the SSDF v1.2 draft closed on 30 January 2026 (NIST). The framework is aimed at US federal software suppliers, but it serves well as a baseline for any secure development lifecycle.
Build in-house, hire a partner, or set up a GCC?
When each model fits
Build in-house when the product is your core advantage, you can hire senior engineers as fast as you need them, and your platform and DevOps practice is already mature. Given the BLS demand projection above, hiring is usually the hardest condition to meet.
A product engineering partner fits when you need a first release quickly, need skills you lack (UX, cloud, AI, accessibility, CRA readiness), or need capacity that flexes. Build-operate-transfer or a GCC suits companies that want to own long-term capability in a lower-cost hub without carrying the setup risk themselves.
Whichever model you pick, the contract should state:
- who owns CRA reporting within the 24- and 72-hour windows
- who signs off accessibility conformance
- how AI tools are governed and measured
- which DORA metrics are reported
Product engineering services in India: the numbers
Nasscom’s 2026 strategic review puts India’s tech industry at $315B in FY26, up 6.1%, with engineering R&D at $63B and software products at $23B. Zinnov and Nasscom count 2,117 GCCs in India as of March 2026: 506 with Forbes Global 2000 parents, 583 mid-market and 504 PE-backed.
Our digital product engineering and GCC teams work from our global delivery hub in Pune. For a growing commerce and technology partner, we set up an Excellence & Operations Center on a Build-Operate model, handling entity setup, infrastructure and compliance. The team was onboarded in six weeks, supports global operations 24/7 and has kept attrition below 10%.
How to choose a digital product engineering company
Ask every shortlisted provider:
- Which DORA metrics will you report, and how often?
- Who handles CRA vulnerability reporting for our EU-facing products?
- What is your accessibility process, and who tests against WCAG?
- How do you govern AI coding tools and measure their effect on stability?
- Which secure-development baseline do you follow, such as NIST SSDF?
- Who owns the IP, and what are the transfer terms if we bring the team in-house?
- Can you give references at our product’s stage: MVP, scale-up or modernisation?
- How do you approach integration with our legacy systems?
- What is your attrition, and how do you protect knowledge continuity?
- What time-zone coverage and support model do you offer?
Why a single market-size figure misleads
Many guides open with one market-size number. For product engineering services, published estimates differ by about six times depending on what they count. For 2025, SNS Insider gives $1,310B and Market Research Future gives $218.14B. IMARC publishes “USD 1.3 Billion”, which looks like a unit error. A widely quoted MarketsandMarkets forecast of $1,510B by 2027 was last updated in May 2023.
Trusted series get revised too. Gartner moved its 2026 worldwide IT spending forecast from $6.15T to $6.31T, up 13.5%. Nasscom’s FY25 base also differs between editions: $282.6B in the 2025 review and $297B in the 2026 review. Before using any of these figures, check the date, ask what the number includes, and compare only within one edition.
A realistic starting sequence
- Name the stage your product is in and what that stage should be measured on.
- Record a DORA baseline before changing tools or teams.
- Give every relevant date in the compliance calendar a named owner.
- Run AI coding tools as a measured trial.
- Choose the operating model for the next 18 months, with transfer terms written in from the start.
Cube27 designs, builds and scales digital products from first release through continuous delivery via Digital product engineering. Talk to a Solutions Architect about your product roadmap.