“I don’t know” may be the correct answer
A salesperson requesting a pitch deck or a marketer asking for campaign assets may know the business need in detail. They may have no reason to know dimensions, aspect ratios, file formats, or production tolerances.
Do not make the requester guess. Use what they do know to route the question to a trusted source.
Find out what they do know
Ask where the work will appear and how it will be used. A requester may not know that they need a 1080 × 1920 asset, but they may know it is a full-screen placement in a particular channel.
- Where will the audience see this?
- Is the placement already booked or selected?
- Does a partner, publisher, printer, or platform own the requirement?
- Has the team produced this deliverable before?
Separate standing specs from campaign direction
Standing specs are requirements that remain true across requests: dimensions, aspect ratio, file format, placement, and required versions. Campaign direction belongs to a specific brief: message, photography, offer, dates, and creative approach.
Mixing the two makes the library hard to reuse. Keep the spec record stable and let the brief carry the changing campaign context.
The minimum useful spec record
A spec entry should answer the production questions a requester or maker would otherwise ask. Keep the record compact enough to scan.
How Smart Specs handles the human decision
When Sage knows the request type, it can surface relevant approved Smart Specs for that deliverable. The requester does not have to recall the dimensions from memory.
A new deliverable can begin with custom specs. If the same completed custom specs recur across submitted requests, Sage can suggest the pattern for review. A person decides whether it belongs in the library.
The approval stays with the team. A repeated value could still be a campaign exception, a client preference, or a mistake. Only a pattern someone verifies belongs in the library.
What should stay out
Do not save one-time campaign copy, temporary exceptions, a client preference that applies once, or a value the team has not verified.
A smaller trusted library beats a large uncertain one. Every entry should answer a question the team expects to hear again.