Subcontractor Performance Scorecard

The problem this kills

You know which subs are good and which ones are trouble - but only in your head. When it's time to hand out the next round of jobs, you go with a gut feeling. When a sub pushes back on losing work, you've got no numbers to show them. And the data that would settle it - who showed up on time, whose jobs got called back, whose invoices never matched the quote - is scattered across dispatch notes, a callback spreadsheet, and your AP person's memory.

So the worst subs keep getting work because nobody pulled the receipts, the best subs feel taken for granted, and every "we need to talk about your performance" conversation turns into he-said-she-said. Meanwhile your customers eat the late arrivals and the callbacks.

What you'll build

A small, private web app for your service managers and procurement team that:

• Takes your completed subcontractor jobs (from a CSV export or a Google Sheet) - scheduled vs actual arrival, completion quality, callback flags, and billing-accuracy data.
• Scores each sub on four pillars - on-time arrival, completion quality, callbacks, and billing accuracy - and rolls them into one overall score using weights you set (because if billing accuracy matters more to you than callbacks, the score should say so).
• Shows each sub's scorecard per period (e.g. monthly or quarterly) and the trend over time, so you can see who's improving and who's sliding.
• Routes any rating that would change a sub's status or eligibility (e.g. "preferred," "on probation," "do not dispatch") to a manager for sign-off - the tool proposes, a human decides.
• Lets you export the scorecard as CSV to share, file, or feed into your system - and ties low scores to a corrective-action note so the data leads somewhere.

What's inside the Implementation Plan

• It starts by interviewing you about your business. Before building anything, the plan has the AI agent ask you about how you track sub jobs today, your real field names and sub-ID conventions, how you measure "on time" and "quality," what counts as a callback, how you flag a billing mismatch, the metric weights you care about, and your status tiers - then it tailors the scoring model and rules to you. No generic template that scores things you don't measure.
• A copy-paste prompt for every step, in order, so you never have to know what to type.
• A clean data model for sub jobs, period scorecards, status changes, and corrective actions.
• The full manager approval gate, audit trail, login, and per-org data isolation wired in from the start.
• Duplicate protection so the same job (sub + period) can't be scored twice.
• A guaranteed "No API yet?" fallback - run the whole thing off a sheet and export CSV, today, with zero integrations.

The governance it includes (this is the point)

This isn't a throwaway script. The plan builds in the controls that make a tool safe to actually run your subcontractor program on:

• Login so only your team can open it.
• Row-level security so each organization only ever sees its own subs and scores.
• A complete audit trail - who scored what, who approved which status change, and when. (When a sub disputes a rating, this is your record.)
• A hard human-in-the-loop gate - the tool calculates a score and proposes a status, but no sub's status or eligibility changes until a manager reviews and approves it.
• Duplicate guards keyed on (subcontractor + period) so re-importing the same job data never double-counts a callback or doubles a score.

Who it's for

Service managers, procurement leads, and project managers who hand work to outside crews and are tired of steering jobs by gut feel - and tired of performance conversations with no numbers behind them. If you can export your completed sub jobs to a spreadsheet, you can build this.

You've got this - paste the first prompt and let the agent interview you.