How to Hand Off Discovery to Delivery Without a Gate
How to hand off discovery to delivery: write down what graduates and why, name who decides, keep the boundary light and never let it harden into a stage gate.
To hand off discovery to delivery, write a short note that says what was assumed, tested and found, what is still uncertain and what outcome the work should produce, then have the person who decides confirm it in a conversation with the builders. Keep it light, or the boundary becomes a stage gate.
The two tracks meet at a boundary, and the quality of that boundary decides whether dual-track produces better products or just more meetings. This post covers what crosses it, who decides, and how to stop it from turning into bureaucracy.
How to hand off discovery to delivery: what the note contains
When something moves from discovery to delivery, it should arrive with its reasoning. The guide lists what that means, and the list is short.
| Include | Why the builders need it |
|---|---|
| What was assumed | So they know which beliefs the work rests on |
| What was tested | So they know how much weight those beliefs can bear |
| What was found | So they inherit the evidence, not just the conclusion |
| What is still uncertain | So they know where the soft spots are |
| The outcome it should produce | So there is something to check after it ships |
This is a short document, not a specification. Its purpose is that the team building the work inherits the reasoning rather than a bare conclusion.
Why the reasoning has to travel
Without the note, the most valuable part of discovery, the understanding of why, evaporates at exactly the moment it is needed. The builders then make dozens of small decisions with none of the context that would have informed them.
It helps to picture the alternative. A finding that arrives as an instruction, such as a line on a list, carries no sense of which parts mattered and which were incidental. A finding that arrives with its assumptions, its tests and its remaining doubts lets the person building decide sensibly when something unexpected comes up. They know what the work is for.
That is also why the people building should have been part of the discovery in the first place. Not for every interview, but enough that the reasoning is shared and not merely transmitted. The note then becomes a reminder of something they already understand, not an introduction to something new. This is the main reason the guide insists on one team running two tracks. For that wider argument, see what is dual track agile.
Decide who decides
A hand-off needs a decider. There must be a clear answer to the question of who says, "we have learned enough, this moves to delivery".
The guide describes what happens without one:
- Everything graduates, and the discovery track becomes a formality.
- Nothing graduates, and the team accumulates well-researched ideas that never ship.
So name the person or the small group, and give them explicit permission to say two things. The first is "not yet". The second is "we know enough, stop researching". The second is as necessary as the first and much rarer, and without it discovery quietly expands to fill the time available.
The decider does not need to be a gatekeeper in a senior sense. They need to be someone whose call the team will accept, working from a bar that was agreed in advance. The bar itself is the topic of what does validated mean in discovery.
Keep the boundary light
The final failure mode is bureaucratic. The hand-off acquires a template, then an approval, then a meeting, then a queue, and dual-track turns into waterfall with new vocabulary.
| Light boundary | Stage gate |
|---|---|
| A short document | A long template |
| A conversation between people who were already involved | A meeting with people who were not |
| One decider with a stated bar | Several approvals |
| Work moves when it is ready | Work waits in a queue for the next review |
The guide gives a practical test. If graduating an item takes longer than a small discovery test, the process has become the point. And a process that has become the point will be quietly abandoned within two quarters.
Keep delivery fed, without rushing discovery
The practical tension at the boundary is supply. Delivery needs a steady flow of validated work. Discovery cannot produce it on a fixed schedule, because questions do not resolve on demand.
The way through is a buffer. Keep enough validated work queued that delivery is never blocked waiting, but not so much that discovery is being pushed into approving things to fill the queue.
- If the buffer is too small, delivery stalls.
- If the buffer is too large, the discovery bar is probably being lowered to feed it.
- If the buffer runs dry, delivery should pick up technical work or reduction of debt, and not unvalidated features.
The last point is the one most often broken. Lowering the discovery bar to keep the queue full is the failure that turns dual-track back into a feature factory with extra ceremony.
Watch both sides of the boundary
Either track can starve, and both feel locally reasonable at every step. Discovery starves when delivery pressure absorbs everything, and the symptom is a delivery queue full of items with no evidence and nobody able to say what outcome they serve. Delivery starves when discovery becomes precious about certainty, and the symptom is shipping slowing while research documents accumulate. How to give discovery a protected share is covered in how to split capacity between discovery and delivery.
The monthly check is simple and direct: what did discovery decide this month, and what did delivery ship? Two consecutive months with a blank on either side is the signal to act.
What the builders should be able to do afterwards
A good hand-off can be judged by what the receiving team can do with it. After reading the note and having the conversation, they should be able to answer these questions without going back to anyone.
- What problem is this meant to solve, and for whom?
- Which beliefs does it rest on, and how well were they tested?
- Where might we be wrong?
- What outcome will tell us it worked?
The last question connects the boundary to what comes after it. The outcome named in the note is the thing someone will check once the work has shipped, a step that is described in how to close the loop after shipping.
The hand-off in one conversation
The guide describes the boundary as a short document and a conversation between people who were both already involved. It is worth being specific about what that conversation is for, because it is not an approval meeting.
| Purpose of the conversation | What it is not |
|---|---|
| Confirm the builders understand why the work exists | A review where discovery justifies itself |
| Surface what is still uncertain | A hunt for a signature |
| Agree what outcome will be checked after shipping | A negotiation of scope |
| Let the decider say "we know enough" | A queue for the next available slot |
If the same people were in the discovery, the conversation is short. That is the practical dividend of running one team: the hand-off becomes a brief check that everyone is holding the same picture, not a transfer between strangers.
Signs the boundary is hardening
A gate does not appear overnight. It builds up one reasonable addition at a time, and each addition feels sensible. Watch for these signs.
- The note has acquired a template that people complain about.
- Work waits for a meeting before it can move.
- More than the decider needs to approve.
- Getting an item across takes longer than running a small discovery test.
- Items are being written to satisfy the template and not to inform the builders.
The fourth is the guide's own measure. When you see it, remove something. Strip the note back to the five things listed above, and return the decision to the named person.
What should arrive, and what should not
The boundary is also a filter on the delivery side. Delivery items are things that have already earned their evidence. If something arrives without it, the right response is not to build it anyway but to ask what is missing.
That is one reason the quality of discovery matters here. When the track is full of pre-decided features, or its tests are ones nobody would have obeyed, the hand-off moves unexamined work faster. The remedy sits upstream, in how the assumptions were chosen, as explained in how to test your riskiest assumption. A light boundary only works if what crosses it has been properly looked at.
A checklist for the boundary
- Everything moving to delivery arrives with what was assumed, tested, found and still uncertain.
- It states the outcome it is expected to produce.
- The people building it were part of the reasoning, not just the conclusion.
- Graduating an item takes less effort than running a small test.
- One named person or small group decides what graduates.
- There is enough validated work queued, but not so much that discovery is rubber-stamping.
- When the buffer is empty, delivery takes technical work rather than unvalidated features.
Frequently asked questions
How long should the hand-off note be?
Short. The guide calls it a short document and not a specification. If writing it takes longer than running a small discovery test, the process has grown too heavy.
Who should decide that something is ready to graduate?
One named person or a small group, working from a bar agreed in advance. They need permission to say both "not yet" and "we know enough, stop researching".
Does every item need a hand-off note?
Everything moving from discovery to delivery should arrive with its reasoning and its expected outcome. Keep the form light so that the note does not become the point of the exercise.
What if delivery runs out of validated work?
Let delivery pick up technical work or reduction of debt. Do not lower the discovery bar to fill the queue, because that rebuilds the feature factory.
The full method, with the checklists, is in The Dual-Track Playbook, a free 34-page playbook.