CASE STUDIES / DEMO ON SYNTHETIC DATA / INTERNAL TOOL
An intake, scoring and review tool you can click through
A working intake, scoring and review flow, built for this page on synthetic data. No client usage yet Unmeasured
The problem
A common internal-tool request: a team takes requests by email or a shared spreadsheet, someone eyeballs each request, and nothing records who decided what.
This tool replaces that with a form, a set of written scoring rules and a review screen. The rules sort the queue. A person makes every final call.
Constraints
- The scoring rules are written down and shown next to every result.
- The score only sorts the queue. A reviewer decides.
- Every decision records who made it.
- This demo runs on synthetic records. No client data and no login.
Who did what
Agency or consultant
- Owns the client and runs the discovery call.
- Agrees the scoring rules with the client.
- Accepts the build against the done-checklist.
I built
- The intake form and its validation
- The scoring rules and the screen that explains them
- The review dashboard
- Tests for every rule
- Handover notes for your client's team
How the work splits on a partner build of this kind. This demo itself had no partner.
Try the demo
Demo on synthetic records. Nothing you type leaves this page.
Warehouse returns form Strong fit
- Met:Budget is approved
- Met:Needed today or this week
- Met:The request explains the job
Fix the broken sales export Possible fit
- Met:Budget is approved
- Met:Needed today or this week
- Not met:The request explains the job
Quarterly board report Weak fit
- Not met:Budget is approved
- Not met:Needed today or this week
- Met:The request explains the job
Waiting
Warehouse returns form Strong fit
Fix the broken sales export Possible fit
Quarterly board report Weak fit
Approved
Nothing here.
On hold
Nothing here.
Timeline
- This demo
- Built for this page.
- A client build like it
- Dated in the written scope, before work starts.
Scope receipt
Build record
- Milestone tier
- Internal tool
- Stack
- SvelteKit, TypeScript, scoring rules kept as plain data
- Evidence basis
- The demo runs on this page. Nothing in it is client data.
Done-checklist
- Every field is validated before a request is saved.
- Every score shows the rules that produced it.
- A reviewer can approve or hold each request.
- Every decision records who made it.
- The tool runs without any client data.
What I would do differently
On a client build I would start from the requests the team has already turned down, because the edge cases live there. A synthetic set only holds the cases I thought of.
Have a client who needs this?
I build work like this for agencies and consultants, under your brand, at a fixed partner price per milestone.
Book a partner call