Define scope for a connected audio toy across four boundaries: the offline behaviour, the online behaviour, the content the toy may reach, and the party responsible for updating it after launch. Then write down what happens when the connection is unavailable, because that case defines the product more than the connected case does.
A connected audio toy is two products
The physical toy and the connected service fail in different ways, are updated on different schedules and are owned by different teams. A project that treats them as one product will discover the split during the first integration test.
Scope work is therefore about drawing the line between them: which behaviours live entirely in the toy, which depend on the service, and what the toy does when the service is unreachable. That third question is the one teams skip, and it is the one users notice first.
Boundary one: what the toy does with no connection
The offline behaviour is the floor of the product. It is what the child experiences on a car journey, in a room with weak Wi-Fi, or after the service has an outage.
Define it explicitly rather than letting it emerge from the implementation. Decide which triggers still respond, which content is stored on the device, and what feedback the toy gives when something is unavailable. Silence is the worst option, because it reads as a broken toy.
- Which interactions still work with the radio off
- Which content is stored locally and how much space it occupies
- What the toy does when a request cannot complete
- Whether the toy indicates the difference between offline and broken
Boundary two: what the connection adds
The online behaviour should be a defined addition, not the whole experience. Write it as a list of capabilities that exist only when connected, and for each one state what the offline fallback is.
This list is also what the service contract has to support. If the connected behaviour includes free-form conversation, the scope now includes a content service with its own availability target, moderation policy and cost per interaction — none of which is a toy-manufacturing decision.
| Capability | Available offline | What scope it adds |
|---|---|---|
| Fixed phrases and songs | Yes | Storage capacity only |
| Page or card triggered audio | Yes | Content pack, no service |
| Scripted question and answer | Partly | Content set and update path |
| Free conversation | No | Service, moderation, running cost |
| Content updates after launch | No | Update process and ownership |
Boundary three: what content the toy may reach
Open conversation raises a question that a fixed-content toy never faces: what is the toy allowed to say? This is a scope decision because the answer determines how much review, filtering and escalation the project needs.
Define the boundary in three parts: the topics the toy should handle, the topics it should decline, and the mechanism that separates them. A toy that answers anything and a toy that answers nothing are both easy to scope; the middle ground is where the review effort goes.
Boundary four: who updates it after launch
A connected toy is not finished at shipment. Content ages, phrases go stale, and the service behind it changes. Someone has to own that after the project team disperses.
Name the owner before sampling. If the answer is the factory, say so in the contract. If the answer is the brand, plan for the tools and the review process the brand will need. An unnamed owner becomes a support problem within months.
- Who can change content, and through what interface
- How a content change is tested before it reaches devices
- How devices receive updates, and whether updates can be staged
- Who answers a user report that something the toy said was wrong
What happens when the connection fails
Write the failure behaviour into scope as a first-class case, not an edge case. Cover three situations: the network is absent, the service responds with an error, and the service responds too slowly.
Each needs a decided behaviour and a decided duration. A toy that waits indefinitely for a response is worse than one that gives up and answers locally, because the child cannot tell the difference between thinking and broken.
A scope document that survives the project
The scope document should be short enough to be read and specific enough to be argued with. Six lines cover most connected audio projects: the offline behaviour, the connected additions, the content boundary, the update owner, the failure behaviour and the acceptance criteria for each.
If a project later expands, the expansion shows up as a change to one of those six lines. That is the practical value of writing them down: it turns an argument about whether to add a feature into an argument about which line changes and what it costs.
OEM / ODM