The innovation pipeline: stages, gates and one metric per stage
Somewhere in your company there is a slide with seven stages and twelve gates, and nobody has ever moved an idea through it. It was drawn for a corporation with a process office, not for a two-person innovation team in a company of 1.200 people. This piece gives you five stages, one gate question per stage that a sponsor can answer with yes or no, and one metric per stage that tells you whether anything is moving.
Why five stages
Without stages, ideas live in a spreadsheet, in somebody's head and in the inbox of the person who last asked. A pipeline fixes that with little ceremony: every idea sits in exactly one stage, has a named sponsor for its next gate, and leaves a trace when it exits. Five is the number you can explain in a corridor.
The five stages
1. Capture
In: workshops, conversations, submissions, a remark in a customer meeting. What happens: the idea is written down in one or two sentences, with who said it, when and where. Nothing is judged yet. Out: an idea with a recorded source. Gate question: is this an idea or a complaint? Complaints go to the problem list, not the pipeline. Metric: ideas with a recorded source.
2. Shape
In: raw ideas, often several that say the same thing in different words. What happens: clustering, a first rough scoring, and merging a cluster into a concept: problem, proposed change, who benefits, riskiest assumption. Out: a concept one-pager. Gate question: do we know the riskiest assumption? If the answer is "it depends", the concept is not shaped yet. Metric: time from idea to concept.
3. Test
In: a concept with one named assumption. What happens: the cheapest experiment that could prove the assumption wrong: five customer conversations, a landing page, a manual version of the service for one team. Before it starts, the kill criterion is written down. Out: evidence. Gate question: did the criterion hold? Metric: tests finished on time. A test that runs for three months without a result is a project in disguise.
4. Decide
In: a concept with evidence and a recommendation from the team. What happens: the sponsor decides and gives a reason in one or two sentences, written next to the decision in the decision log. Out: go, no, or park with a date to look again. Gate question: is the decision recorded with a reason? Metric: concepts with a recorded decision.
5. Deliver
In: a concept with a go. What happens: handover to a line function or a project with budget. The innovation team supports, it does not own. Out: an owner outside the innovation team, with a name and a first milestone. Gate question: has someone outside the innovation team said yes to owning this? Metric: concepts that reached a customer. The number the board cares about and most pipelines never report.
The pipeline on one page
| Stage | In | Out | Gate question | Metric |
|---|---|---|---|---|
| Capture | Workshops, conversations, submissions | Idea with who, when, where | Is it an idea or a complaint? | Ideas with a recorded source |
| Shape | Raw ideas | Concept one-pager | Do we know the riskiest assumption? | Time from idea to concept |
| Test | Concept with assumption | Evidence | Did the kill criterion hold? | Tests finished on time |
| Decide | Concept with evidence | Go, no or park, with reason | Is the decision recorded with a reason? | Concepts with a recorded decision |
| Deliver | Concept with a go | Owner outside the innovation team | Has an outside owner said yes? | Concepts that reached a customer |
Five rules that keep it honest
- A gate is a question, not a meeting. The sponsor can answer it by email on a Tuesday afternoon.
- Every stage has a named sponsor. Not a role, a name.
- Ideas can skip a stage, never a gate. An idea that arrives with evidence can jump to decide; the sponsor still asks whether the assumption was named and the criterion held.
- Parked and dropped are exits at every stage, with a reason. Without one, the idea comes back in six months as a new idea.
- Read the pipeline from the right. How many concepts reached a customer, not how many ideas came in.
Tools that keep a state per item and record each decision with its reason (tools like IdeaHone do) turn the monthly review into a filter instead of an archaeology session.
Introducing it in one quarter
- Write the five gate questions in your own words. If "riskiest assumption" sounds foreign in your company, write "what has to be true for this to work".
- Assign a sponsor per gate. Five names, confirmed with each person, with an agreement on how fast they answer. A week is reasonable.
- Put every current idea in a stage. Most are in capture. Some are nowhere, because they died unrecorded; write them down as dropped, with what you know about why.
- Run the first monthly pipeline review. Thirty minutes, with the sponsors. Two questions only: what moved since last month, and what has been stuck longer than a stage should take.
- Report the five metrics. One number per stage, in the same order every month. The first report will be ugly. That is the point. Once a quarter the same numbers go into the one-page innovation report to the board.
The stage most companies miss is shape. Ideas jump from a sticky note straight into a business case template and die there, because nobody can write a five-year forecast for one sentence.
A pipeline with more gates than ideas in it is a bureaucracy. If the flow is thin, start with three stages (capture, test, decide) and add the other two when ideas get stuck for lack of them.
How a pipeline decays into a diagram
- Gates as committees. Once a gate needs five people in a room, it meets quarterly, and ideas wait a quarter for a one-sentence answer.
- Metrics that count inputs. Ideas submitted, workshops held, people trained: all go up while nothing reaches a customer.
- A pipeline owned by nobody. Someone has to call the review and chase the stuck items. Otherwise the pipeline is a diagram.
- Delivery without an outside owner. The innovation team runs the pilot year after year and the line never adopts it. Deliver is complete when someone else has the budget and the headache.
Never lose an idea between sessions
IdeaHone follows every idea from first mention to decision, across all your workshops and teams.
What are the stages of an innovation pipeline?
A workable pipeline for a mid-sized company has five: capture, shape, test, decide and deliver. Each has a defined input and output, one gate question and one metric.
What is a gate in innovation management?
A gate is a yes-or-no question that a named sponsor answers before an idea moves on, such as "do we know the riskiest assumption?". It is a question, not a meeting, and an idea may skip stages but never a gate.
How do you measure an innovation pipeline?
With one metric per stage, read from the right: concepts that reached a customer, concepts with a recorded decision, tests finished on time, time from idea to concept, and ideas with a recorded source. Counting submitted ideas measures input, not progress.
