Stage Gate Process: Who Decides at Each Gate

The stage gate process divides work into stages and places a gate, a decision meeting, before each stage so money is committed only on evidence. A gate ends in one of four outcomes: Go (proceed), Kill (stop), Hold (pause until stated conditions are met) or Recycle (rework, then review again).

Robert Cooper's model, as published by Stage-Gate International, runs from Discovery through five stages: Concept, Build the Business Case, Development, Testing and Validation, and Launch. Each gate looks forward. It has three parts: the deliverables from the previous stage, the criteria they are judged against, and a clear output.

For new products, Stage-Gate International lists six gate criteria: strategic fit, product and competitive advantage, market attractiveness, technical feasibility, synergies with core competencies, and financial reward and risk. The gatekeepers who apply them are cross-functional business leaders who, in the model's words, "manage by exception until they see the team at the next Gate."

No. Stage counts and outcome wording vary between guides, so write your own definitions down. For the search "stage gate process" on Google in India in November 2026, we counted seven of the nine results (the other two were an encyclopaedia entry and a video). Five list six stages and two list five, and their outcome wording falls into six different sets.

The practical lesson is that Hold, Recycle and Conditional Go are not interchangeable. Each needs a different follow-up. Our recommendation is to use the four outcomes above and add Conditional Go only when it carries an owner and a deadline.

A named senior person who does not report to the team under review, and who is not the only sponsor of the component being judged. Cooper's model gives gates to cross-functional senior leaders, and NASA's procedural requirements name a Decision Authority for each programme.

NASA's procedural requirements (NPR 7120.5F, effective 3 August 2021) define a Key Decision Point as the event where the Decision Authority determines the readiness of a program or project to progress to the next phase. For programmes, that authority is the NASA Associate Administrator, who may delegate it. The Decision Authority determines whether and how the programme proceeds, and the outcome is recorded in a Decision Memorandum.

The UK's National Infrastructure and Service Transformation Authority (NISTA) publishes six numbered gate reviews for major government projects, from Gate 0 (Strategic Assessment) to Gate 5 (Operations Review and Benefit Realisation).

Our recommendation:

• Name the decider before the evidence pack goes out.
• Let the programme or project manager assemble the pack and present it, but not decide.
• Write the decider into your decision rights and escalation rules; our programme governance framework page shows how.

In a programme, components start and finish at different times, so gates apply to each component and to the programme as a whole. In a portfolio, a gate is a funding choice among components competing for the same money. In our view, Cooper's original model is built around a single product project.

Our programme management life cycle page sets out the programme gates, each with a named decider and a pass or fail test, so we do not repeat them here. For the portfolio side, intake and periodic review are covered in our project portfolio management process page. Funding released against gates is covered in our programme financial management page.

Two further points. Agile work does not remove gates. Cooper's September 2026 article says modern versions of Stage-Gate bring in build-and-test cycles, rapid customer validation and sprints inside a stage. Teams iterate inside the stage, and the gate stays the investment decision.

In regulated sectors, we recommend putting control readiness in the evidence pack, so compliance is checked before funding rather than as a late obstacle.

A gate decision record is one page that fixes the gate's question, decider, evidence and pass criteria before the meeting, and fixes the outcome, conditions, funding and reasons after it. Ten fields do the work, and each has a test you can mark pass or fail.

Fields 4 and 5 follow a common design in which knockout questions must be passed and scored characteristics are weighed. The three-day lead time in field 3 is our recommendation, not a standard.

The outcomes map to programme governance language as follows. The mapping is our recommendation.

Tally the outcome of every gate held in the last twelve months, then check each outcome other than Go for an owner, a deadline and a re-review date. A year of only Go decisions does not prove failure, but it is the first thing to question.

1. List every gate held in the last twelve months, with date and decider.
2. Count outcomes by type: Go, Kill, Hold, Recycle, Conditional Go.
3. For each outcome other than Go, check that its conditions were closed or the item was killed. Pass: every condition has a recorded close date. Fail: any condition is still open with no owner.
4. For each Go, check that the decider was named and the pack circulated beforehand. Pass: both are recorded. Fail: either is missing.

As an illustration, with invented numbers: a board that held 12 gates and recorded 12 Go decisions has a non-Go rate of 0 of 12, or 0%. A board that recorded 9 Go, 1 Hold, 1 Recycle and 1 Kill has 3 of 12, or 25%. Neither figure is good or bad alone. The second board shows the gate can say no, and the first prompts a question about the criteria or the decider.

If a must-meet criterion was waived, record it as a Conditional Go with an owner and deadline rather than a plain Go.

We teach governance as part of PMI PgMP® (Program Management Professional) and PfMP® (Portfolio Management Professional) preparation. Our PgMP® Fast Track Program, a group mentoring program described on our PgMP training page, includes a Program Governance module covering governance practices, plans and reviews. Our PfMP® Fast Track Program and PfMP Self-Paced Program (see PfMP training) include a Portfolio Governance module.

Our programs prepare you for PMI certification; they do not award it. Our first-attempt success rate is 98.5% (how we calculate it).

To discuss your goals and suitability, book a free discovery or eligibility consultation by writing to support@augmentconsultancy.com or calling or messaging on WhatsApp at +91 9816 360 116. You can see current programs in our Course Store.